This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3033584, IEEE Access

Date of publication xxxx 00, 0000, date of current version xxxx 00, 0000.
Digital Object Identifier 10.1109/ACCESS.2017.Doi Number

Improving Urban Mobility by Defining a Smart

Data Integration Platform
P. Cáceres1, A. Sierra-Alonso1, C.E. Cuesta1, B. Vela1, and J.M. Cavero1
1
Universidad Rey Juan Carlos, C/ Tulipán s/n, 28933 Móstoles (Madrid) SPAIN.

Corresponding author: P. Cáceres (e-mail: [email protected]).

This work has been partially supported by the Access@City research project (TIN2016-78103-C2-1-R), funded by the Spanish Ministry of Science, Innovation
and Universities.

ABSTRACT One of the key factors employed to define the well-being of citizens in the urban environment
is mobility, since it defines a set of flows and connections that constrain those citizens’ individual and
collective behaviour. However, the complexity of this activity on the scale of a city makes this a complex
problem in computational terms. One of the main reasons for this is the asymmetry of information: different
actors have access only to partial or outdated information, and many relevant data are simply unavailable. In
this paper, we propose a data integration architecture and platform with which to combine relevant data from
many different sources and provide the results in a variety of forms. This integration uses semantic
technologies, thus ensuring that the relationships among data show their actual meaning and are appropriately
interpreted. The resulting platform amalgamates: open data, which is available from public sources; extracted
data, obtained from public sites by means of scraping techniques; pre-processed data, stored in public
databases; aggregated data, acquired from pervasive devices by means of crowdsourcing; smart data, supplied
by mobile applications and enriched with contextual information, or data concerning specific incidents, often
provided by the users themselves. The semantic integration of this information makes it possible to compute
a wide range of results, from accessible transport routes to identifiable events, in a coordinated manner. The
general public is then supplied with these results through the use of specific software, via either mobile
applications or the web. We are of the opinion that the collective use of this information may improve urban
welfare.

INDEX TERMS Crowdsourcing, Data Acquisition, Data Processing, Open Data, Pervasive Computing,
Semantics, Smart Data, Social Computing, Software Architecture, Urban Mobility

I. INTRODUCTION One of the key factors is urban mobility: the way in which
The concept of the smart city implies the notion of a better people move through the city defines their perception of the
city: one in which data is gathered in order to learn about the environment, the concentration of the population and the
problems in that city and decide upon potential solutions to interferences and interactions among individuals; in short, it
them; one in which there are fewer disruptions, the citizens’ is the framework that delimits their behaviour. Urban
lives are improved, and their experience is enhanced – in mobility is delimited by the combination of pedestrian
summary, a city whose inhabitants’ well-being increases. routes, private transport and public transport. The first is
The urban landscape is a highly complex ecosystem and can mostly relevant over short distances, while the second cannot
easily degenerate into a hostile environment; as this be controlled but only restricted. Public transport is,
complexity grows, the general public is beginning to rely on however, the backbone that defines the flows of people and
technology to make it evolve in the right direction. This is, goods within a living city. An appropriate regulation of
allegedly, the actual purpose of smart cities: using public transport has direct consequences as regards
technology to solve the challenges of the evolving urban improving citizens’ welfare – and in this case, the notion of
space, and applying software solutions in order to tackle both welfare must be understood in the most general sense, not
large-scale issues and small-scale concerns. limited to (but including) the economic perspective.

VOLUME XX, 2017 1

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3033584, IEEE Access

The use of public transport, as opposed to private transport, structured data files concerning, e.g. the pollution levels in
provides great advantages in terms of the entire urban the city. Second, many other sources, which provide
community‘s well-being. On the one hand, traffic jams, energy information in the form of an accessible API (application
consumption and carbon emissions are reduced. On the other, programme interface), are not exactly considered part of an
new parking infrastructures are not necessary, which allows open data initiative but share much of its philosophy. Third,
the construction of new pedestrian spaces, such as parks and the largest potential source of information is probably not
recreation areas, in their place. Any means employed to even these initiatives, but rather the presence of
increase the use of public transport and improve its efficiency computational devices throughout the city, that is, the
have a positive effect on the whole system. citizens’ smartphones (and other smart devices) with their
Much of the current research on smart cities has been many sensors and their huge computational capabilities. If
related to cyber-physical systems and the Internet of Things, adequate software is employed, these smartphones can be the
and much of it has, therefore, had to deal with hardware means to both capture and even pre-process information that
issues. However, the smartness of the city is, in reality, is not accessible in any other form, and provide the smart
provided by the software that processes the data captured by city’s inhabitants with the processed information in a useful
that hardware. Indeed, on the scale of a city, the actual issue form. This implication of the citizens themselves in the
is that of capturing all kinds of information in different data building of the smart city not only defines a way in which to
streams, synchronizing them, processing them, and extend the “sensor network” in that city, but also creates a
performing the corresponding analyses in order to learn crowdsourcing endeavour, which may take many different
about the current situation of the city, even in real time, and forms, and in which each new software application in fact
decide which actions would serve to improve that situation. embodies a different conception. It could even be argued that
In this context, rather than isolated efforts to process this crowdsourcing approach is required for the success of
separate fragments, an integral approach is relevant and any smart city initiative, and it should be considered as an
required. The data context of the city is defined by both essential part of any complex urban computing approach.
stable information and a dynamic status. The first of these Fourth, and finally, there are the considerable number
provides structure, which can be recovered from large databases that store all the stable information
conventional databases and includes data concerning the city mentioned above, and this information is the basic mainstay
itself, the means of transport and the combined routes they of the city’s information processing system.
define. The second comprises variable data streams obtained Coordinating all of these sources of information is highly
from different sources, and includes information regarding complex, and in many situations goes beyond what a
unexpected events, temporary barriers, accidents and computational system can deal with and enters the scope of
incidents, or simply the flow of individuals within the city at the so-called “big data” initiatives. Nevertheless, the
a particular moment: collapsed streets, traffic jams or public integration of even a small part of these data may have a
demonstrations. Many proposals concentrate their efforts on huge impact on the well-being of the individuals affected.
finding routes within the stable structure, without Even partial improvements can have very significant
considering the critical influence of temporary obstacles; and consequences.
of course, the influence of a certain event cannot be It is for this reason that our work proposes:
estimated without considering the remaining data. Both  A data integration architecture and platform, on
aspects of this information are, therefore, relevant, and both which all the data from the different sources are
must be considered in a coordinated manner, even when stored, semantically annotated and harmonised, thus
considering synchronization; e.g. an accident may cause a ensuring a feature-rich integration; and
traffic jam, but several hours later, this is no longer relevant;  A set of specific and focused software applications
a multitude might cause a standstill, but an alternative route designed for both smartphones and the web that can
could provide a detour around it. When dealing with a solve specific issues (accessibility, incidents and
dynamic status, the timing is as relevant as the place itself. infrastructure) one at a time.
Moreover, there is an enormous variety of information We intend to provide a generic architecture so as to
sources. Apart from the abovementioned emphasis on support data integration on an urban computing platform.
sensor-oriented computation and IoT devices, there are many This will lay the foundations for many different applications,
other options. First, it is necessary to consider that any smart each of which will be designed in order to improve the well-
city initiative also places emphasis on transparency and on being of urban citizens and to enrich the platform itself,
the publishing of open data concerning the city itself. These thereby allowing more sophisticated decisions to be made.
open data sources (many of which employ a form of The key aspect as regards providing this generic integration
processable Linked Open Data ) may vary in their approach, is the use of semantic technologies.
ranging from an automatic data stream, perhaps originating The main contributions of our work in addition to the
from a physical sensor, to more elaborate and even pre- Services on the move for Urban mobility (SofUR) platform
processed data sources, such as the periodic emission of are, therefore, the following:

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3033584, IEEE Access

 A Services on the Move Architecture (SoMA), which ICTs, but also aspects such as sustainability and liveability.
is structured in three following different subsystems: There is currently a tendency to consider that a
o An Open Data & Acquisition subsystem, which multidisciplinary approach is required [53] in order to make
obtains data from the open and public sources. the transition to a smart city.
o A Semantic subsystem, which processes and In this respect, several authors have drawn up lists of
integrates data and then transforms them into dimensions in order to determine which aspects actually
semantic data by using our domain-specific MAnto constitute a smart city [27][3][40][60][54]. Although each
ontology. list contains different dimensions, there are several in
o A Smart Data subsystem, which gathers smart data common, such as people (human), social aspects, the
economy, mobility, quality of life or the natural environment
concerning public transport and its accessibility by
and technology.
using various apps developed for this purpose.
Within the framework of smart cities and in line with the
 A suite of applications that offer urban services to
objective of improving the quality of citizens’ lives, it is
citizens with the aim of putting mobility in the city on worth emphasizing the need to improve urban mobility, as
the move. This suite is divided into the two following mentioned in the introduction. In their definition of a smart
groups city, Giffinger et al [40] mention mobility as a basic element
o The suite of mobile applications, which is a set of to be taken into account during its construction. Washburn et
mobility applications designed for mobile devices. al [82] also confirm the need to make services such as
o The suite of Web applications, which supplies transport more efficient, and this is directly related to
information services to data consumers, who will citizens’ mobility. Albino et al [3] state that "high-quality
use those data according to their needs or and more efficient public transport that responds to
convenience, and which also offers mobility economic needs and connects labour with employment is
services on the Internet. considered a key element for city growth". Lombardi et al.
This paper is structured as follows. Section 2 shows a [51] also opine that smart mobility is a basic component of a
review and discussion of some similar proposals, while smart city.
Section 3 provides a summary of the different technologies Indeed, urban mobility has, in recent years, become an
used throughout this work and related efforts. A detailed object of study in an attempt to improve it in order to increase
description of the structure of our proposal (the Services on citizens’ well-being [77]. One of the main aspects of this
the move for Urbanmobility –SofUR- platform and the well-being is public transport. The European Union has
underlying SoMaarchitecture) is then shown in Section 4, funded numerous projects in order to address this issue
which also includes an inclusive enumeration of the [35][1]. These projects develop infrastructures with which to
associated suite of applications. Section 5 illustrates how improve public transport and make it more comfortable and
these applications prove the validity of the proposa, in more accessible to all, and they incorporate the use of
information technologies (IT) to make public transport more
addition to providing our conclusions and an initial account
efficient and effective and to ensure that its use is a more
of some future work.
pleasant experience.
In this last respect, several software applications have
II. RELATED WORK
been developed that inform users about the public transport
In this section, we discuss some of the most representative
network and how to use it. These applications provide
software approaches employed to improve citizens’ well-
information about the lines, stops, accesses, etc. of the
being as regards urban mobility. Firstly, we examine how,
different means of public transport. Almost all of them
within the framework of smart cities, urban mobility is provide the user with the possibility of requesting a route
considered to be one of the factors affecting that well-being. (optimal in terms of time or distance) from an origin point to
The term smart city has been used for over two decades a destination point. In some cases, they include information
[3]. Numerous definitions have been proposed during this about accessibility features, or elements for people with
period; however, while they have commonalities, they also special needs or disabilities. When considering the
differ in some respects. Choubariet al [27] and Albino et al importance that people and social aspects acquire in the
[3] present a compilation of these definitions. general context and conception of smart cities, it should not
Some authors consider that the defining characteristic of be forgotten that an essential feature must be the inclusion of
a smart city is the use of ICTs when applied to citizens' daily all social groups by overcoming barriers of language,
lives [43][44][82]. This implies that it is necessary to collect culture, education and disabilities [3][53]. Taking
data by means of sensors, meters, appliances, personal accessibility into account is, therefore, essential if people
devices, or similar, in order to later integrate them and, with special needs are to be provided with a good experience
through their use, provide citizens with services and assist as regards the use of public transport. Several pieces of
them to make informed decisions. But a different trend has software use crowdsourcing to gather data about public
recently appeared, which claims that other relevant aspects transport. The information collected is then made available
are required in order to consider a city as a smart city. Of the
definitions mentioned above [27][3], some include not only

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3033584, IEEE Access

to other users in order to improve the responses that citizens when they encounter them. This is referred to as
attain from these applications. crowdsourcing. There are also applications that use
Those applications that take aspects of accessibility into crowdsourcing to improve urban public transport. For
account usually focus on a particular type of impairment or example, Tiramisu Transit [85] tells the user whether there
group of people. Wheelmate [84] or WheelMap [83] is room left for a wheelchair on a bus or how full that bus is.
consequently provide information regarding wheelchair Moovit [59], meanwhile, provides information about the
accessibility for many places in 45 countries. In addition, status of each service, calculates routes and indicates when
users can modify the degree of accessibility of registered to get off. The OneBusAway [33] project consists of a set of
places or mark new ones. Cardonha et al. [23] have tools whose intention is, among other goals, to comply with
developed an application that enables users to incorporate the bus schedules or to decrease waiting times in order to
accessibility information so as to collaborate in the creation increase well-being on urban public transport. This objective
of accessibility maps. Access Map [1] helps people with can be extended to other transit systems. In addition,
mobility needs to plan an accessible route in the city of OneBusAway permits users to make comments about these
Seattle. These applications consider only mobility tools. Swiftly [75] provides more accurate vehicle arrival
limitations. Others, such as Landmark Ontology for Hiking data for transport agencies, thus enabling them to provide
[71], are intended for elderly people. In this case, the their users with better information and allow them to better
application aids them to walk less when hiking. There are plan their journeys.
also applications that consider other types of special needs, Other initiatives that use crowdsourcing are the BUSUP
one of which is “CiudadesPatrimonio de la Humanidad” project [14] and the CIVITAS initiative [24]. The former
[15], a web application that provides accessibility allows users to book crowdsourced buses on demand, while
information concerning tourist routes in World Heritage CIVITAS is an initiative from the EU to promote a new
cities. This information refers to mobility needs, in addition urban mobility culture. One of its mobility strategies
to others related to vision and hearing. These applications are includes safe and secure transport for all users, taking into
not, however, generally customisable: their users can account a variety of needs [35]. In the CIVITAS initiative,
establish the place to which they wish to go, but cannot several pilot projects are being deployed in European cities
establish what their accessibility needs are. to test new accessibility concepts, such as smart access
As stated above, the European Union is making a great facilities for wheelchairs [35].
effort to improve public transport [35], and it is necessary to OpenTripPlanner (OTP) [63] is a project that calculates
highlight two projects that focus on the use of IT to improve routes combining different transit systems, including
mobility on urban public transport: ACCESS 2 ALL [2] and bicycles and walking routes, and takes (transport)
Mediate [55]. ACCESS 2 ALL exhaustively analyses users’ accessibility into account. OTP obtains data from
possible needs with respect to public transport. By taking OpenStreetMap [64] and GTFS (General Transit Feed
these needs into account, it establishes guidelines to ensure Specification) feeds [41].
that urban public transport is accessible to all As will be noted, there are numerous approaches with
citizens. Moreover, it proposes customised services for route which to improve users’ well-being when they use urban
guidance. The Mediate Project [52] has identified a set of public transport. However, to the best of our knowledge,
measures with which to describe the degree of accessibility none of them considers the calculation of routes by taking
of a particular means of transport and has developed an the accessibility features for any type of need into account or
application with which to quantify it. Another result of maintains real time data concerning temporary incidents (a
Mediate Project has been a Good Practice Guide for lift does not work, there is work taking place that hinders
accessibility. These two projects seek to establish a blind people, etc.) in the public transport network.
theoretical framework that will cover all aspects of citizens’ It is now necessary to discuss the need for data with which
mobility, but do not provide solutions in the form of user to provide intelligent urban public transport. These data
applications. originate from heterogeneous sources, including data
Another aspect that qualifies smart mobility is the use of scattered on the Web in several formats or data originating
data [47]. In order to achieve actual smart mobility, services from portable devices (e.g., smartphones) [50]. In order to
have to rely on a lot of information: not only a certain amount obtain greater benefit from existing data, they should be
of data, but also a significant amount of variety. For effective published in a coordinated manner, harmonised and linked,
mobility, these data must be up-to-date, and in many cases, thus uniting the efforts of public and private initiatives. The
real-time information is required. A paradigmatic example is management, maintenance and publication of data have,
the communication of incidents in the public transport therefore, become a growing challenge. Several initiatives
network. If an application calculates an accessibility route, have, however, been set up to meet this challenge, some of
but, for example, the recommended lift does not work, then which consist of governmental data portals based on CKAN
the route is not useful for the user. In order to obtain this [25]. CKAN is a data management system that facilitates the
information in real time, it is possible to take advantage of publishing and sharing of data. Other works present
the widespread use of smartphones and encourage the users platforms on which to manage, link or publish data from
themselves to provide pieces of this incident information different sources, e.g., QuerioCity [52], AECIS [36], the

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3033584, IEEE Access

proposal by Bischof et al. [12] or ATIS [7]. Some proposals

develop more comprehensive frameworks covering several BUSUP [14]
stages of the data publication process, such as Santos et al.’s CIVITAS [24]
process [70], the VIVO approach [81], the linked data platform Ferris et al. [33]
Collection
LinkedLab [30], the CITIESData framework [50] or Lau and Ismail [49]
(crowdsourcing)
CityPulse [65]. All of them handle data from different Swiftly [75]
domains, but none of them focuses on public transport data Zimmermanet al.
and less still on the accessibility of public transport. Several [85]
Publication and Bischof et al. [12]
works can also be found in the domain of urban public The need for data in harmonization CKAN ¡Error!
transport. The proposal by Schlingensiepen et al. consists of order to provide
No se encuentra
designing a framework for autonomic transport in the smart intelligent urban
mobility el origen de la
city environment, but it is for transport in general [72]. Ning
referencia.
et al. present a control system for urban rail based on
Darari et al. [30]
artificial systems, considering human factors in order to Gandon et al.[36]
improve the experience during the use of urban rail, but it is Liu et al. [50]
not intended to allow users to personalise the use of the data López et al. [52]
[61]. However, Lau and Ismail’s framework employs a Puiu et al. [65]
crowdsourcing approach to provide passengers with real Santos et al. [70]
time data in order to meet their needs [49]. But once again, VIVO [81]
nothing about accessibility is mentioned. TABLE 1. Related work classified by issues related to urban mobility in
We can conclude that there are many works in the area of smart cities
smart technologies whose objective is to improve urban
mobility so as to increase the well-being of citizens, but that III. RELATED TECHNOLOGIES
there is still much to be done, especially as regards The work described herein was carried out using various
accessibility, if urban mobility is to be accessible to all. technologies. The most important technologies used to gather
Table 1 summarises the related work. This table presents and harmonise data and to develop the server are described
the issues analysed in this work in order to improve urban below.
mobility and, therefore, the well-being of the citizens in the
context of intelligent cities. A. WEB SCRAPING
Data are scattered on the Web, distributed on many different
Albino et al. [3] sites. Furthermore, they have different formats that are
Choubari et al. [27] structured to a greater or lesser extent. Data can appear as plain
Definition and
Hall [43]
characterization of text or organised in html tables. It is possible to download files
Harrison et al. [44]
smart cities containing data in XML or JSON format or to invoke APIs
Lytras and Visvizi[53]
Washburn et al. [82] that also provide files containing data in those or other formats.
Albino et al. [3] For these data to be useful, they have to be collected, related
Definition of
Choubari et al. [27] to each other and given a common format.
dimensions in order
Giffinger et al. [40] Web scraping solves that need by obtaining data from the
to characterise a
Lytras, Visvizi and Sarirete[54][53] Internet. It consists of gathering data automatically (from the
smart city
Nam and Pardo [60]
Internet) by using programmes (denominated as bots or
Albino et al. [3]
Urban mobility as a scrapers) that query a web server, request data, parse those
Giffinger et al. [40]
factor in the well- data to extract the information required and store them for their
Lombardi et al. [51]
being of the citizens
Washburn et al. [82] later use [57].
in a smart city
Tomaszewska and Florea[77] There is a variety of programming techniques,
Accessmap. [Online]. Available: programming languages and libraries with which to attain the
The role of public
transport in urban
https://www.accessmap.io.[1] data distributed on the web. One of these languages is Python
mobility Gaggi,Fluhrer and T. Janitzek [65]. Several modules can be used to programme a scraper.
[35] Three of the most frequently used are [42]:
Accessmap. [Online]. Available: - The requests library [69], which is a wrapper over
https://www.accessmap.io.[1] the urllib Python library whose goal is to handle the
ICT applications Access2All [2]
HTTP requests. Its use is recommended by the Python
with which to Ciudades Patrimonio [15]
Cardonha et al. [23] 3 documentation, since it is friendlier than urllib
improve the
Mediate [55] [67].
experience of using
public transport OpenTripPlanner[63] - The Beautiful Soup library [9] helps to extract the
Sarjakoski et al. [71] information of interest from HTML or XML files. It
Wheelmap[83] can be used with different parsers depending on the
Wheelmate[84]

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3033584, IEEE Access

format to be analysed. This library provides a function One of these implementations is Apache FUSEKI [34],
that creates a tree from the HTML or XML file. Each which forms part of the Apache Jena project. It is a SPARQL
node in this tree contains a tag. These tags, which server and provides the means to recover and update data from
contain the information of interest, are searched by a JENA repository using SPARQL.
means of other functions, also provided by the library. Apache JENA [4] is a free and open source Java framework
Once the tag is located in the tree, its contents are that is employed to build applications and which accesses the
extracted. Semantic Web and Linked Data. JENA is used to define an
- Scrapy [73] is a more powerful framework than RDFS [13] so as to describe the underlying relationships that
Beautiful Soup and can be scaled, but is more difficult exist between data on the RDF graph. JENA provides a set of
to use. Scrapy [48] is an open source framework with functions and services in which to store, extract and publish
which to extract data from websites using XPath [26]. data as RDF triples that comply with the RDF schema.
It offers the tools for the efficient extraction of RML (the RDF Mapping Language) is a general language
unstructured data that are scattered throughout the web, that permits the definition of rules with which to map
their processing and their storage in the structure and heterogeneous data sources onto RDF graphs [31]. Mapping
format required [48]. rules are defined in order to transform data from, for example,
Scraped data are integrated using several formats, mainly XML, JSON and CSV formats into the RDF. The rules
XML, JSON or CSV. expressed by means of RML transform heterogeneous data
structures into an RDF data model. These rules transform the
B. SEMANTIC TECHNOLOGIES data format by defining new triples, semantically annotated
In this subsection, we introduce the various semantic with the vocabularies (ontologies) specific to the data domain,
technologies that are necessary to understand our work. and integrate those triples into RDF graphs already defined in
The Resource Description Framework (RDF) is a model for the same domain. The rules can be defined from any data
data interchange on the Web [68]. In the RDF, data are source, in formats as diverse as CSV, HTML, XML, or even a
described as a set of triples. A triple is formed of three format similar to that of relational databases. The result of
components: a subject, a predicate and an object. The subject applying the rules is a set of RDF triplets that uses the
is a resource that is related to an object by means of a predicate. predicates and vocabulary types of the domain.
The object can be another resource or a literal. The predicate
can describe a relationship between the subject and object or a C. DATA PROCESSING TECHNOLOGIES
property of the subject [29]. In this subsection, we introduce the various technologies
In order for the data of which the triples are formed to be related to our proposal.
referenced, the RDF identifies each component of the triple Apache Kafka [5] is a stream-processing software platform.
with a URI (Uniform Resource Identifier) [11]. The set of It implements a “publish/subscribe” message system, is
triples can be represented graphically as a graph (an RDF scalable and distributed and provides a valuable platform on
Graph) on which subjects and objects are nodes and predicates which to process streaming data between applications.
are represented with edges. For all the users of that data to be Apache NiFi [6] is an easy-to-use, powerful and reliable
able to interpret the predicate in the same way, the use of system with which to process and distribute data. It makes it
standard vocabularies is recommended, since they will allow possible to automate the movement of data between different
the definition of the predicates of the triples. Finally, the RDF systems quickly, easily and securely. It can load data from
graph is serialised to enable computers to handle data. One of different sources, and has a very powerful web interface that
the most frequently used notations is RDF-XML [36], which allows its users to visually design the data flow, act on the
is based on an XML format, while SPARQL is used to recover process and monitor that process.
data from an RDF model [62]. MongoDB [58] is a scalable, flexible and distributed
SPARQL is an RDF graph query language that has been document database. It stores data in JSON-like documents,
standardised by the World Wide Web Consortium (W3C) thus making data integration easier and faster. Searches can be
[62]. The use of SPARQL allows the definition of queries, carried out by means of fields, range queries, regular
which search for data in the RDF graph that satisfy the expressions or JavaScript functions. It supports field indexing
conditions of that query. It is a key technology in the and real time aggregation. MongoDB provides great
development of the semantic web, since the data structured in availability in the form of replica sets. It can run on multiple
RDF format could not otherwise be accessed. A SPARQL servers, and balances the load or duplicates data in order to
query recoveries RDF sentences, that is, triples. As with SQL, keep the system up and running in the case of failure.
it is necessary to distinguish between the query language and
the engine for data storage and retrieval. There are C. WEB INTEGRATION TECHNOLOGY
consequently several implementations of SPARQL that are In this subsection, we introduce the various technologies that
generally linked to different technologies. are necessary for web integration.

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3033584, IEEE Access

Spring Boot [74] is a framework that is used to develop denominated as the Open Data & Acquisition subsystem, is
applications in Java. It provides a set of tools in order to responsible for obtaining data from the open and public
configure and programme applications on any platform. The sources, while the second, the Semantic subsystem, processes
use of Spring Boot allows the programmer to focus only on and integrates data and then transforms them into semantic
the solution and forget about the complex tasks of data by using our domain-specific MAnto ontology. The third,
configuration in terms of data access, security, denominated as the Smart Data subsystem, gathers smart data
communications, etc. concerning public transport and its accessibility by using
Apache Tomcat [7] is open source software that permits various apps developed for this purpose, and those data are
web applications to be deployed in an integrated manner by then sent to the Semantic subsystem in order to transform them
sharing the same subsystem. Some of our applications are into semantic data.
directly run on top of Tomcat and do not, in these cases, Fig. 1 shows the general structure of the SofUR proposal
employ Spring. and, within it, our SoMAarchitecture, in which the above-
mentioned subsystems are represented. Each of these
IV. SERVICES ON THE MOVE FOR URBAN MOBILITY subsystems will be explained in greater detail in subsections
As stated in the related work section, several solutions B, C and D. The figure also includes the suite of apps (mobility
regarding how to deal with these issues already exist but, to and web apps), which will be explained in subsection E, and
the best our knowledge, there is no comprehensive solution which are an integral part of the SofUR platform.
that solves most of the known problems from a single
perspective.
Our Services on the move for Urban mobility (SofUR)
platform defines a set of services whose objective is to
improve the mobility of all the citizens in a city by promoting
the use of public transport. This results in an improved quality
of life and welfare in cities with regard to air quality, the flow
of traffic, etc.
We have, therefore, developed a service-based architecture
and a suite of software applications, which together comprise
the aforementioned platform. The Services on the Move
Architecture (SoMA) is structured in three different
subsystems that carry out acquisition, processing and
integrating activities with data obtained from crowdsourcing,
public and opensources and that provide enriched information
in the form of semantic data. The suite of software applications
offers different urban mobility information services obtained
from these semantic data for both citizens and data consumers
and, in some cases, also provides smart data by means of
crowdsourcing techniques. FIGURE 1. SofUR Platform.
In the following subsections, we first describe the Services
on the Move architecture, after which we specify the different B. OPEN DATA & ACQUISITION SUBSYSTEM
subsystems in which our proposal is structured. Finally, we The Open Data & Acquisition (ODA) subsystem is
present the suite of applications developed that offers a set of responsible for processing information obtained from public
services to citizens and data consumers in order to improve sources and open data related to the public transport
public transport users’ well-being. infrastructure and its corresponding accessibility features. Fig.
2 shows the details of this subsystem.
A. SERVICES ON THE MOVE ARCHITECTURE The first kinds of sources are public information sources, in
As mentioned above, this work is being developed with the which interesting public transport data are usually scattered on
aim of providing public transport and its corresponding the Web and distributed on many different sites. All these data
accessibility information. We, therefore, acquire data are collected through the use of web scraping techniques and
concerning public transport from different open and public we have, therefore, programmed several scrapers using the
sources, after which we process the data obtained. We then Python language. The Scrapy [73] and Beautiful Soup [9]
integrate and enrich them, and finally, we transform these data libraries, mentioned above, are used to access the websites,
into semantic data. and to seek and extract data of interest. We focus on the data
In order to carry out these activities, we have designed an extraction and processing of the existing information on the
extensible structure, the Services on the Move Architecture web concerning public transport, along with its accessibility,
(SoMA), which is composed of three subsystems. The first, by means of a method for the semi-automatic generation of a

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3033584, IEEE Access

data scraper for the public transport domain. This method subsystems, converting them into RDF graphs by means of
allows the extraction of public transport data and the existing programmatic (Java) processing or, more recently, RML
accessibility information from a selected website. Moreover, transformations, and then storing them in the Jena repository.
we have developed a web tool that applies the aforementioned The Jena repository is a single repository, but it stores two
method in order to generate a data scraper for the public different data collections: the public transport & accessibility
transport domain [80]. The data collected are organised and collection and the incidents collection. Details of the SEM
related according to an underlying domain model, in this case subsystem are provided in Fig. 3.
a public transport domain model, which was described in
detail in [80]. Finally, we convert them into a structured data
file with a common format (usually JSON or CSV) and these
files are transferred to the Semantic Subsystem.

FIGURE 3. Semantic Subsystem.

The SEM subsystem applies an ontological schema,

denominated as the MAnto schema, which is related to the
public transport infrastructure and the accessibility feature
FIGURE 2. Open Data & Acquisition Subsystem. domain. The definition of the MAnto schema is based on the
Transmodel (European Reference Data Model for Public
The second kinds of sources are open data sources, which Transport Information) [76] and IFOPT (Identification of
provide open access to public transport data, either through an Fixed Objects in Public Transport) [45] reference data
models. Transmodel describes a model of both public
API service, or by directly downloading data files. Each
transport concepts and data structures related to the different
source provides data in a different format (CSV, XML, KML) kinds of public transport. IFOPT extends it by including
and with diverse internal structures, but all of them must specific structures designed to specify accessibility data
satisfy the constraints of a domain model. In many cases, concerning the equipment of vehicles, stops and access areas
sources offer open transport data as CSV files by following in order to define a model for the principal so-called fixed
GTFS. Those open data files can be processed generically, objects, i.e. elements related to the access to public transport
because they have the same structure as that shown in [21]. (e.g., entrances, stop places, connection links). We have,
With regard to the sources without a GTFS structure, it is therefore, obtained a set of metadata (MAntoterminology)
necessary to process each one of them in a specific way in with which to semantically annotate the public transport
order to seek the information of interest, which is specified in infrastructure and its accessibility elements data from sources
[21].
the domain model, and taking the particular structure of each
As mentioned in the previous subsection, data from the
source into account. We process sources of both types by different sources are organised and related according to an
means of Java and Python programmes in order to extract the underlying conceptual model, which we have defined as the
specific data required for our purpose. The main reason for domain model. Our data are, therefore, always domain
using Java was initially that we had more experience with this objects and are related in the same way, that is, by following
language, and also owing to its expressive power. We have the relationships described in that model. The ontological
since used Python more frequently, essentially owing to its schema, denominated as the MAnto schema, is also based on
readability, simplicity [45] and the availability of libraries. the same domain model. It represents data in that context, i.e.
Finally, we convert them into a structured data file with a in that domain. The MAnto vocabulary (a set of terms)
common format (usually JSON and/or CSV, depending on the consequently also represents the domain objects, and the
relationships between terms are additionally contemplated in
origin) and send it to the Semantic Subsystem.
that reference model, again our domain model. Fig.4 shows
how the MAnto ontology schema is defined following the
C. SEMANTIC SUBSYSTEM structure of the domain model in the transport context.
The Semantic (SEM) subsystem is responsible for processing
structured data files from the ODA and Smart Data (SMD)

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3033584, IEEE Access

FIGURE 4. The definition of MAnto Ontology Schema

model. Moreover, we can use the MAnto ontology to define
Fig. 4 shows only a fragment of the domain model in (A) a set of consistent mappings with which to semantically
and a fragment of the MAnto ontology schema in (B). We annotate the data from the structured data file generated by
depict how an object in the domain model, denominated as the OAD subsystem. An example of a semantically annotated
oD-n, is defined as a class in the MAnto ontology schema. data fragment is shown in Fig.5.
For example, the Public Transport object in (A) is defined as mao:line 1_4 a mao:line ;
the http://com.vortic3.MANTO#PublicTransport class in (B). mao:sequence [ a <#Seq> ;
Similarly, a relationship between two objects in the domain <#_1> mao:par_4_295 ;
<#_2> mao:par_4_294 ;
model, denominated as rD-x, is defined as a data property in <#_3> mao:par_4_293 ;
the MAnto ontology schema. For example, in (A), the <#_4>mao:par_4_27 ;
.......
relationship between the Public Transport and the Line object ] ;
makes it possible to establish which lines belonging to a mao:transportFor "Metro" ;
sch:description "Pinar de Chamartín-Valdecaros" ;
means of Public Transport (Metro Madrid, the London sch:name "Line 1" .
Underground, etc.) are defined as the
http://com.vortic3.MANTO#TransportFor data property in FIGURE 5. Data annotated by means of MAnto Ontology

(B). It also specifies that this property domain is the In this figure, it will be noted that the subway has a line
http://com.vortic3.MANTO#Line class, and its range is the (sch:name “Line 1”) that is internally coded as line 1_4, and
http://com.vortic3.MANTO#PublicTransport class. which is composed of a set of stations (par_4_295, par_4_294,
As defined in [80], all data sources in the public transport par_4_29 ) ordered as a sequence.
context are specified according to the same underlying At the end of the process, these data will be stored in the
domain model. As our approach uses the same domain model semantic Jena repository, in the public transport &
to identify the objects from data sources and to define the accessibility collection, as shown in Fig.3. Data from this
MAnto ontology schema, it is easy to deduce which terms collection can be downloaded at any moment from
from the MAnto ontology are mapped onto the existing http://coruscant.my.to:8080/download/metro.xml.
objects in the data sources. The available data is now growing steadily, as this
In fact, in the SEM subsystem, we work with the data file information is increasingly more valuable to users. In this
generated from the OAD subsystem. Once again, the data respect, it might be necessary to change the ontologies in
from that file also represent the same objects from the domain order to adapt to this growth if the domain model evolves. In

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3033584, IEEE Access

fact, MAnto is a ‘living ontology’, that is, it can be expanded transformation, which is supported by a Spring Boot
through the use of an iterative and incremental process [39], substrate, in order to generate the corresponding structured
thus making it possible to semantically annotate new data that data and send them to the SEM subsystem. Pedestrian routes
could appear over time. are, meanwhile, processed by means of a NiFi configuration,
The SEM subsystem similarly includes other ontological and are stored in MongoDB to be quickly reused; this
schema related to incidents (i.e. changes in the working state) information has not yet been semantically annotated.
of the accessibility elements in the public transport network,
which could occur at any time (more information is provided E. THE SUITE OF APPLICATIONS
in [16]). Incident data, which are obtained from the Smart We have developed a suite of applications, that is to say,
Data subsystem, have to be semantically annotated and then several software applications, that offer urban services to
stored in the Jena repository, in the incidents collection, as citizens with the aim of putting mobility into the city on the
shown in Fig. 3. In this case, we have defined a set of mapping
move. The suite is divided into two different groups, the first
rules with which to automatically transform data from the
SMD subsystem into RDF graphs by means of RML. Sample of which is the suite of mobile applications and is, as its name
data from the incidents collection can be downloaded from suggests, a set of mobility applications designed for mobile
http://coruscant.my.to:8080/download/events.xml. devices. Some of these devices provide the SMD subsystem
with smart data, while others use the MongoDB support. The
D. SMART DATA SUBSYSTEM second group is a suite of Web applications, which not only
The Smart Data (SMD) subsystem employs crowdsourcing offers mobility services on the Internet, but also supplies
techniques [21] to process real-time data from different information services to data consumers, who will use those
applications that we have developed for this purpose. This data according to their needs or convenience.
real-time information is related to (a) incidents, that is to say, A detailed description of these application suites is provided
variations in the working state of the accessibility elements of below.
the public transport that may occur at any time, and (b) 1) SUITE OF MOBILE APPLICATIONS
pedestrian routes performed and requested by users The suite of mobile applications is a set of five different
themselves. Android apps. Three of them provide bus and metro mobility
One example of data in this context is a lift that does not services for the city of Madrid in Spain. These services
work at a specific transport station, which is part of a specific compute specific routes on which some of their users’ special
transport line. This event also has an associated opening date. needs will be taken into account. One of these three apps also
When the incident is solved, we can close it by indicating the
allows users to indicate incidents related to the accessibility
final date. The event data concerning accessibility features,
which are processed from many sources (i.e. a crowd), are elements in the public transport network, that it to say, to
stored in a data file (see Fig. 4). The information in this data indicate their working state. The aim of the fourth app, which
file is relevant in order to update the current accessibility is based on gamification techniques, is to capture the
features in the network. This data is our SMD output. accessibility elements in the public transport network and its
The SMD subsystem then provides the SEM subsystem working state, among other functionalities. Finally, the fifth
with the processed data. Details of the SMD subsystem are app registers the user’s pedestrian route, on which his/her
shown in Fig. 5. special needs can be indicated. We should stress that the last
three apps also collect data using crowdsourcing techniques,
with the users’ permission. A detailed description of each of
them is provided as follows.
Notify.me [38] provides users of the EMT Madrid company
with two different urban bus route services [32]: guiding the
user along the bus route and notifying the users about the
events on the route. EMT Madrid is the urban bus public
company of the city of Madrid and provides an open data
platform via web services [28]. The app accesses them in order
to obtain the specific urban bus route requested by the user. It
then shows it on a map, after which it guides the user along the
route and notifies her when a relevant stop is reached by means
FIGURE 5. Smart Data Subsystem. of acoustic, visual and sensorial messages. Fig. 6a shows an
example of a visual message. Fig. 6b displays a route on the
We use Apache Kafka technology to manage and gather map, on which each bus stop is identified by means of a blue
smart data streams, that it to say, flows of incidents regarding icon. Fig. 6c shows the corresponding information for this
the working state of public transport accessibility features
route at the top of the user interface. The first line shows the
(which usually arrive in emergent surges, and could became
massive in size) and specific pedestrian routes to be followed starting point (LEGAZPI), the destination (CLÍNICA
by users. Incidents are then processed by a specific Java MONCLOA), the time by bus (112 minutes), the time on foot

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3033584, IEEE Access

(20 minutes) and a single transfer. The second line shows the
departure and arrival time (11:38 and 14:13) how long this
route is (34 km).
We wish to highlight that this software is currently
employed by EMT Madrid buses Company [79].

FIGURE 7. MMAppcessible user interface.

MMA4A [16] is an improved version of MMAppcessible

that takes all the user’s special needs into account thanks to the
IFOPT specification regarding this means of transport. Fig. 8
(a) (b) shows two app screens. That in Fig. 8a is the screen used to
request a route and on which it is possible to indicate the user’s
need (in this case “averse to escalators”). That in Fig. 8b,
meanwhile, shows the route and, in this case, the fact that the
destination “Plaza de Castilla” is not accessible for the user,
who has an aversion to escalators. The recommended
destination route is, therefore, Ventilla.

(c)

FIGURE 6. Notify.me user interface.

Another application is MMAppcessible, which offers Metro

de Madrid (MM) users accessible routes on this means of
transport. MM is the public metro company of the city of
Madrid (Spain).
This app, therefore, works with the locally stored public
transport & accessibility collection, which was previously
downloaded from the semantic repository. MMAppcessible
permits users to indicate the origin and destination of their
route, any special needs (visual, auditory or mobility
limitations –for example, pushing a pram, or using crutches or (a) (b)
a wheelchair-) and their preferences, such as fewer transfers or
FIGURE 8. MMA4A screen.
stops. MMAppcessible then shows the route, if it exists, on a
map and permits the users to navigate along it. Fig. 7 shows
When the app starts up, it requests information from the
the user interface.
server regarding whether the public transport & accessibility
This app won an award in the open programming
and the incidents collections are up to date with respect to the
competition at the School of Computer Science and
locally stored collections. With regard to the route calculation,
Engineering at Rey Juan Carlos University, in Spain.
the MMA4A now works with the two collections, because the
incidents on the MM infrastructure could allow it to modify
the route as regards the user’s needs. As mentioned above, the
accessibility elements are part of the infrastructure of MM, and

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3033584, IEEE Access

the route (with its corresponding accessibility elements) must eventually permits the users to visualise their routes on a map
be adapted to the user’s needs. For example, a person with a (see Fig. 10c).
temporary mobility issue needs a lift to gain access to the
public transport network. It is, therefore, necessary to know
which accessibility elements are operational at each station. It
is for this reason that the notification of incidents merits
special attention: when a user detects an incident, s/he can
notify it by means of the app and MMA4A, therefore, requests
the accessibility element that does not work (lift, escalator,
steps, ramp or travellator) and the station at which it is located.
The app then sends the corresponding incident to the server to
be included in the incidents collection of the semantic
repository. Other users will be able to benefit from this
information, which may be vital if they are to follow a route
according to their special needs.
Access ’n’ Go! [17] is an Android application that has been
developed using gamification and crowdsourcing techniques, (a) (b)
and permits its users (the players) to capture public transport
accessibility elements (ACEs) such as lifts, escalators, etc. The
app then provides the SofUR proposal with smart data
concerning the working state of ACEs, that is to say, updated
information in real-time regarding those elements. Fig. 9a
shows the main user interface of Access ‘n’ Go!, on which the
red icon is the player’s current position, a green icon represents
an ACE that could be captured and a grey icon represents one
that has already been captured. The game level, points and
badges achieved by the player are shown at the top of the
(c)
screen. Fig. 9b shows the team ranking user interface.
FIGURE 10. AccessPedRo screens.

The smart data generated by the app is then sent to the SMD
subsystem. Although AccessPedRo is still in the testing phase,
it can already be used to compute accessible pedestrian routes
in its current form.
2) SUITE OF WEB APPLICATIONS
We have developed two different Web applications, which
have different purposes and audiences.
The first Web app, denominated as MMAR, provides
citizens, including those with special needs, with public
transport infrastructure and accessibility information services
concerning the Metro de Madrid (MM) company in order to
allow them to get around using this means of transport. MM is
the public metro company of the city of Madrid in Spain [56].
This Web app permits its users to know what specific
accessibility features each specific MM station has. Moreover,
(a) (b) it provides the possibility of calculating routes based on the
users’ special needs and the accessibility features of the MM
FIGURE 9. Access ‘n’ Go! screen.
network. The Web app is available at http://bit.ly/MMAR_1.
Fig. 11 shows its user interface.
The objective of AccessPedRo [78] is to obtain the
The second app, denominated as MMSemQuery, is an open
pedestrian route that the user should follow. The users must,
access Fuseki endpoint, in which the dataset can be queried by
therefore, first register and give their consent to this activity
means of SPARQL queries. MMSemQuery will normally be
(see Fig. 10a). The app then requests the users’ needs for the
used by data consumers, who can download specific MM data
route (see Fig. 10b). From that moment, AccessPedRo obtains
in different formats from the dataset. This app is available at
the users’ locations on the routes they are following. The app

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3033584, IEEE Access

http://coruscant.my.to:3030/dataset.html. The user interface With regard to the first category:

of MMSemQuery is shown in Fig. 12.  Different transport media have several types of
infrastructures (stops, stations, entrances) and
logistics (vehicles and routes). For some means of
transport, such as the Metro, this infrastructure is part
of their responsibility, and for others, such as the bus
service, the infrastructure is the responsibility of other
institutions, such as the city council. Accessibility
features in these media are, therefore, provided by the
institutions themselves, by other institutions or are, in
other cases, non-existent. The transport data for a
specific means of transport may, therefore, originate
from different sources, and the transport information
offered is consequently not uniform in terms of
format, content and completeness.
 The same means of transport are managed by
different companies in different cities, signifying that
the terminology and characteristics of their
infrastructures and logistics are different in each case.
For example, a Metro Madrid line has only one single
FIGURE 11. MMARuser interface.
route, while a London Tube line can have more than
one route. The transport information provided is,
again, not uniform in terms of format, content and
completeness.
 Some of the transport companies provide their data in
an open format, such as the London Tube. In this case,
the information is more complete and structured than
that from companies that do not provide it, such as
Metro Madrid; this signifies that it must be obtained
by means of scraping techniques. Moreover, this
information sometimes has to be completed with
additional data from other sources; it would, for
example, be necessary to add the geographical
coordinates to the entrances and stop places of Metro
Madrid.
With regard to the second category:
 Each transport company offers the information with
different levels of granularity. For example, one
FIGURE 12. MMSemQueryuser interface.
company may indicate that a station has universal
accessibility, while another may also indicate its
V. CHALLENGES ADDRESSED
specific accessibility infrastructure (lift, ramp,
The development of this data integration platform has
escalator, stairs, etc.).
involved addressing different challenges and overcoming
 The meanings of the terminology used may differ. For
several difficulties. The greatest effort is related to the
example, some transport companies state that they
integration and harmonisation of data, since the origin,
provide universal accessibility by noting that they
format and completeness of the data is highly diverse. In
serve people with visual, hearing and/or mobility
addition, in order to carry out this harmonisation, it has been
problems. However, IFOPT considers that phobias to
necessary to previously define an ontological scheme based
certain infrastructures are also a factor to take into
on transport domain standards which, as mentioned above in
account when referring to universal accessibility. For
section 4.C, are IFOPT[46] and Transmodel [76]. The
example, a person with a phobia to lifts needs to have
ontological scheme has been defined by respecting the
access to stairs, and the availability of a lift does not,
structure, syntax and semantics of the standards considered.
therefore, represent universal accessibility for that
The most relevant aspects of this effort, which have been
individual.
divided into two categories ((a) format, content and
The ontological scheme has been designed by taking into
completeness data, and (b) semantic and granularity) are
account both the reference standards (Transmodel, IFOPT)
discussed below.

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3033584, IEEE Access

and all of the aforementioned problems in order to harmonise the availability of a unified integrated data hub gathering
the great variety of data managed in terms of formats, much of this relevant information has a direct influence on
contents and meanings. This ontological scheme is improving the everyday life of those involved. A perfect
constantly evolving, since all of the above-mentioned cases example of this could be our previous efforts [19] focusing
must, according to our analysis of data from other transport on accessibility data in public transport, and the tangible
companies, be considered. usefulness of these results in the related groups of interest,
As will be observed, the applications developed have such as citizens with reduced mobility, or impaired people.
addressed these problems. For example, if we consider As mentioned in the Introduction, it is unfeasible to
different means of transport, such as Notify.me for an urban attempt to solve all the issues of public transport
bus service (Fig. 6) and MMAppcessible for a metro (Fig. 7), simultaneously; however, each separate effort in this
different meanings associated with universal accessibility direction provides an immediate benefit, not only for direct
have been included, e.g. the inclusion of passenger phobias, participants, but also for the collective common welfare. Our
as in the case of MMA4A (Fig. 8), etc. In summary, all the approach as regards both the building of the architecture and
challenges and issues mentioned in this section have been its validation is, therefore, to consider each particular
considered during the conception and development of this problem and solve it in isolation. This is, however, done by
architecture and the associated platform; in some cases, using all the existing information in the solution and
several times. integrating any acquired data into the repository, or any
single application into the platform. This means that each of
VI. VALIDATION the specific applications described in Section 4 is an integral
Our proposal is basically a data integration architecture (the part of the global system, and thus also serves to validate its
Services on the Move Architecture, SoMA); as described overall utility. By validating every single local application,
previously, this architecture gathers data from many we, therefore, also validate the full global system. Indeed,
different and diverse sources: linked open data, public APIs, each of these specific applications (the suite of mobile and
processable open data files, data scraped from dissimilar web applications) uses the public transport and accessibility
websites, data collected from smart devices, data retrieved collection and the incidents collection as its sole data source.
from public databases, spatial data available on public maps, The behaviour of these applications is based on this common
topographical data recovered from non-open public sources, source: when they capture data, they feed these collections;
smart data pre-processed by mobile phones, data streams when they process information, it is obtained from this
provided by IoT devices, open data streams, or pre-processed repository. A correct operation is that of validating the quality
data aggregations of previous crowdsourcing efforts. Joining of these data.
all these data together, reconciling their divergences and In summary, the keystone of this architecture is the
complementarities, and semantically harmonising their datastores in the Semantic Subsystem: these datasets (the
contents is both an enormous and a rather abstract effort. It aforementioned public transport and accessibility collection,
is simultaneously very specific, as all the features are derived and the incidents collection) and their contents are
from real-world constraints. The magnitude of the effort is simultaneously the main result (the core) of the architecture
related to different aspects of the elements involved: means and its extensions (the apps) and the main source of their
of transport have diverse infrastructures, which are mostly behaviour. The ultimate goal of this architecture is to capture
permanent (stops, stations, entrances), and several logistics and store this information regarding transport, accessibility
(vehicles and routes), which are usually much more variable; and incidents. Once these data have been correctly captured
and both categories have accessibility features, which are and processed, the apps in the architecture are useful and
provided on an individual basis – many instances of the same behave as expected. One of the apps also provides the
element may have different accessibility issues. information that is required by one of the others. Even when
Furthermore, accessibility has a “weakest-link” approach: an all the apps are removed, the architecture is still useful and
accessible element (e.g. a train) which passes through a non- fulfils a purpose, because it contains the relevant data in the
accessible location (e.g. a station) is, in fact, no longer semantic datastore.
accessible. It is particularly difficult to validate this work, as The validation of the architecture is, therefore, directly
this implies both evaluating its suitability and showing its related to the quality of those data. In order to validate the
benefits. architecture, we intend to verify whether it provides the
As discussed in Section 1, the benefits of the architecture infrastructure that is able to guarantee (or at least to maintain)
are directly related to its usefulness and its direct impact on the quality attributes that sustain these data.
the well-being of citizens, and obviously on the impact on Data quality is, therefore, often described in terms of the
the collective use of public transport and its consequences conjunction and composition of these attributes, namely
for the dynamics of the smart city itself. Many of the accuracy, completeness, accessibility, redundancy,
inconveniences that appear in the urban mobility system are readability, consistency, usefulness and trust [9]. The best
owing to asymmetries in the access to relevant information; way in which to validate these attributes is by means of a set

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3033584, IEEE Access

of case studies, as a single case study might not be sufficient applications, most remarkably the MMAR and
to describe all of them; and each of the apps consequently MMSemQuery applications, both of which are focused on
provides a case study. This also makes it possible to the Madrid subway network.
specifically check certain attributes, considered in isolation Apart from these semantic interfaces, the platform
from the others. The melange of these relative validations provides specific means to query specific information in its
serves to validate the full architecture. suite of mobile and web applications, particularly in certain
domains. The usefulness of these queries is sometimes
A. DIMENSIONS IN THE ARCHITECTURE defined as the integration of data originating from different
The architecture itself is an abstract construction, signifying sources; for instance, in the restricted mobility context,
that it cannot be directly validated. But as described accessibility data are often combined with Open Street Map
previously, this integration architecture has been representations, which serve to locate them and are tightly
implemented in the domain as the Services on the move for correlated with them (e.g. in the MMA4A app).
Urban mobility (SofUR) platform. Our approach in this Furthermore, some specific queries serve to separate
section is, therefore, to consider the SofUR system as a information, and consider some data in isolation; for
“SoMA in practice”. A full assessment of the platform instance, incidents and events also need to be integrated and
would, of course, require many details, and these depend on semantically correlated with their context in the transport
specific technologies (described in Section 3) and on the network, but these are distinct data, with specific features,
implementation itself. Our validation will, however, focus on such as their temporal relevance, and hence have their own
the aspects derived from the architecture itself and from its processing module and their own repository. However, they
data integration features; the implementation details are of a are still integrated into the architecture, and are essential in
less importance here. order to keep data updated, by synchronizing their
The validation of the SofUR platform will, therefore, be occurrences with the stable structure, while defining the
considered in two different dimensions, namely (1) from a current status.
global perspective of data integration, as defined by the The SofUR platform provides different kinds of data
SoMA architecture; and (2) from the local point of view processing, with different properties, and its main feature is
provided by each specific application on the platform. In not the performance level that it attains in them, but the fact
fact, the interplay between global and local viewpoints that they are semantically combined and temporally
provides the same kind of assessment already mentioned synchronised; and once they have been integrated, any
when discussing data quality and its specific attributes. particular application can be provided with them, with the
With regard to the first dimension, there are no appropriate same effort that a local store would require, but with a much
data integration metrics at the overall system level; some richer context.
internal metrics could provide certain insights into the Even when disregarding their particular origins, once
structure of data, but not into their influence on the citizens’ integrated, data have three notorious features in the SofUR
welfare. In order to validate this kind of outcome, we must platform, namely:
consider some external metrics, i.e. those properties that  They are semantic data, in the triple sense that they
have a significant influence from a global perspective. have been semantically annotated, have semantic
Two of these metrics are the degree of integration (from relationships and define a semantic context for any
the user’s point of view) and their accessibility. In order to new data;
improve the second, and to be able to evaluate the first, it is  They are smart data, as they are not only data
important to note that the integrated dataset of urban mobility gathered from specific selected sources, but they have
information (semantically annotated with the MANTO been processed, harmonised and normalised. This
ontology [20]) is accessible for download from the platform processing ensures (1) basic data quality, as any
itself [16], and has been for years, while still continuously unreliable information is filtered out; (2) analytic
growing. Raw data is accessible in the form of Linked Open services, which depend more on relationships than on
Data and can, therefore, be processed, and includes the volume of data; and (3) the capability to actively
information that has been made available only in non- respond to specific questions and problems using a
semantic formats, such as GTFS [18]. variety of interfaces.
In addition to this raw semantic (LOD) form, the SofUR  They are, finally, open data again, as the results (and
platform makes it possible to query these data in a variety of even crude data) are made openly available in order
ways. First, the platform provides a Fuseki semantic search to enable their downloading and reuse.
engine on top of the Apache Jena server, and is essentially The composition of these three features, notably the
able to answer any potential SPARQL queries. This means semantic capabilities, makes a unique combination, and this
that any users can tailor this semantic interface to their is in itself one of the values to consider when assessing this
particular needs. Moreover, in order to facilitate this kind of platform.
queries, the platform itself provides a series of suitable web

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3033584, IEEE Access

With regard to the second dimension, the impact that this resulting data stream without overflowing – which is
integration architecture has on the citizens’ welfare can be the origin of the smart data subsystem; and third, the
more easily evaluated by considering the utility of every semantic subsystem has to be able to synchronise
single application on the SofUR platform. Indeed, apart from these data and to integrate them with the rest of the
making it grow organically, driven by people’s needs, the repository, achieving a remarkable level of
specific applications define the convenience of the platform integration, as mentioned previously. The utility of
itself – in addition to the abstract benefits of the data this version is significantly expanded with respect to
integration (as described above), each specific app provides the above version, as it achieves a capability that is
a particular utility. These applications, whether mobile or rarely found in the context of mobility, i.e. the
web, determine the value of the platform implemented and possibility of avoiding local incidents that
the architecture it encloses. In fact, the applications in this compromise the accessibility of a potential valid
suite can also be described as a set of case studies, which route, which has to be temporarily circumvented.
together assess the validity of the platform.  Once the central architecture had a stable structure,
the remaining applications were simply built onto it
B. APPLICATIONS AS CASE STUDIES and, therefore, benefit from its contents. The next app,
A brief discussion of the kind of validation provided by these Access ‘n’ Go!, is described in [17], which explores
applications, whose enumeration is ordered in a strictly the context of crowdsourcing, discussing what
temporal manner, is provided below. Each discussion is motivates users to provide the information required –
related to the data quality attributes listed at the beginning of on this occasion in the form of a serious game. Our
this section at the end of this summary. initial experiments with this approach have achieved
 Notify.me, the bus route definition app [38], was the a significant amount of success and are carefully
first application in the suite, and it already defined and summarised in [17]. In the first iteration, a group of
required the initial version of our architecture. In 20 people got involved in the game in order to capture
order to obtain the bus network information, it had to (and confirm) the accessibility elements in the Madrid
access a variety of sources; in order to compute the subway network, and were able to capture more than
routes, it had to use central resources; and it already 500 of these elements in more than 1450 interactions.
provided the integrated information in the form of a The web application, MMAR, which is still under
map. It, of course, has an obvious utility; in practice, development, takes advantage of the unique
calculating transport routes is the first manifestation capabilities of a modern web interface to expand
of urban computing. Moreover, this app was used by beyond the limits of a mobile app. Its purpose is quite
the EMT bus company itself, before developing its similar to that of MMA4A, but rather than focusing on
own version. providing routes on the move, it is able to exploit the
 The first app with actual data integration, data contained in the repository in a more flexible
MMAppcessible, also had a clear utility. It already manner – as described in the first part of this section.
gathered accessibility information – mostly through  The semantic web application MMSemQuery
scraping techniques [20], and it then integrated it with similarly supplies a web-based interface in order to
pre-existing data stored in the repository and consume the semantic information contained in the
computed mobility-aware transport routes, with a triplestore. As noted above, the main purpose of this
special emphasis on impaired and blind people, and application is again to provide an adaptable interface
even had a voice interface. The social significance of so as to avoid the need to directly access the semantic
this application is, therefore, obvious. Moreover, this repository via Fuseki.
version already required a full implementation of the  Finally, the mobile app AccessPedRo, which is also
SoMA architecture, which must be considered as the under development, takes a different approach, by
first version of the SofUR platform. exploiting the same information in order to define
 The next app, MMA4A, is an evolution of the above, pedestrian routes on a street map of the city [78].
with even more emphasis being placed on These routes have been conceived in order to
accessibility [16]. It has a better usability, an evolved complement the transport network, but have a set of
interface, and a completely expanded architecture. distinct features which, despite the very similar
The idea is to add temporary information (incidences, interface, require quite different processing – and the
events) to the route computation algorithm. This has SoMA architecture has, in fact, been expanded to
three consequences: first, the application itself has to accommodate a specific datastore in order to optimise
be able to capture these incidences, and must define its reusability. The utility of this application is, when
the interface in order to allow the user to introduce considered in isolation, an additional functionality,
them; second, the architecture must be able to process directed towards a different group of interest; but when
this crowdsourcing information, accommodating the combined with the existing mechanisms, it supplies the

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3033584, IEEE Access

means to define a full connectivity mesh on the streets provide a positive validation of the complete SofUR
and in the transport network whose significance is yet platform, and the SoMA abstract architecture, which
to be explored. supports their conception and definition.
 Moreover, there are a number of additional models, It should be emphasized that this is not just an assortment
still under initial development, which will expand the of web and mobile projects in a certain domain. We have
architecture by extending the available functionality described the history of the project in order to show how
with supplementary features. For instance, one every app plays a logical role in the evolution of the platform.
module is already able to process the semantic Beyond that, this evolution has also been guided by the need
information as it arrives in the repository, and to to validate the architecture. For instance, in the initial stages
provide the required normalisation in a principled of the platform, it was difficult to update information in the
manner, while another is working to improve the repository. Updating obviously affects consistency, which is
method employed to capture open non-processable still one of the most significant data quality attributes
information by means of enhanced scraping (including the recent, and relevant, body of work on eventual
techniques. consistency). In order to validate this consistency we had to
As mentioned previously, these applications also serve as be able to update the data within the architecture in a flexible
specific case studies, which allow us to specifically study the manner. The MMA4A case study came into being as a means
data quality attributes mentioned. Each of these case studies to provide and test the update capabilities in the architecture,
targets at least one, and possibly more, of these attributes; the and it is now one of our most successful efforts.
intention is to cover them all by combining the specific In summary, all of these applications show the wide range
applications. The SoMA architecture is, of course, still of possibilities provided by this semantically integrated
responsible for the integration dimension mentioned above, architecture, the way in which it is able to organically grow
and provides global consistency. and increase its objectives, and the advantages that it already
Table 2 shows the correspondence between every single provides in order to improve the users’ well-being in the
app (case study) and the data quality attributes. Note that the urban environment. The utility of this approach has,
full set is covered by composition, and that the integration therefore, been shown, and its significance has been
architecture plays a central role. As the results of every validated by describing its existing implementation.
validation have been positive, we have to conclude that they

Notify.me MMAppcessible MMA4A Access’n’Go! MMAR MMSemQuery AccessPedrRo SoMA

Accuracy        
Completeness      
Accessibility       
Redundancy  
Readability        
Consistency     
Usefulness       
Trust    
TABLE 2. Qualitative impact of data attributes on the different applications (case studies) and global impact on the SoMA architecture

city, even in real time. Only then will we be able to address

VII. CONCLUSION AND AREAS FOR IMPROVEMENT the corresponding solutions in order to improve this
The concept of the smart city includes the use of software situation. This must be done by taking many available
solutions to meet the challenges involved in improving sources into account, including their distinct features. The
citizens’ daily lives. Urban mobility is a key factor and any first sources are those provided as open data, many using a
improvement to public transport, therefore, provides a benefit Linked Open Data structure, ranging from automatic data
for the urban community. On the one hand, traffic jams, streams to more elaborate and even pre-processed data
energy consumption and carbon emissions will be reduced, sources. There are many other sources that are not strictly
and on the other, new parking infrastructures will not be open or even processable, which supply information in the
necessary, thus liberating new spaces for pedestrians. All of form of an API (application programme interface). It is,
these changes will undoubtedly increase the welfare of however, probable that the largest potential source of
society. information is not provided by available shared data, but
In this context, data captured from hardware sensors, open rather by the presence of computational devices in the city,
and public sources, and smart devices must be synchronised specifically the citizens’ smartphones and other smart
and processed in order to perceive the current situation of the devices. When the citizens themselves are involved, a

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3033584, IEEE Access

crowdsourcing initiative is defined, and each new software [4] Apache Jena (n.d.). [Online]. Available:
https://jena.apache.org/index.html.
application embodies a different need. As noted previously, [5] Apache Kafka (n.d.). [Online]. Available:
a crowdsourcing approach of this nature is required for the https://kafka.apache.org/intro.
success of any smart city enterprise, and should be [6] Apache Nifi. [Online]. Available: https://nifi.apache.org/.
[7] Apache Tomcat. [Online] Available:
considered as a vital part of the urban computing approach. http://tomcat.apache.org/.
Processing should also include data already stored in large [8] ATIS, Alliance for Telecommunications Industry
databases, which define the basic backbone of the city’s Solutions, “ATIS Smart Cities. Data Sharing Framework
structural information. In summary, our work proposes the Report”, March 2018. [Online] Available:
https://www.atis.org/smart-cities-data-sharing/smart-
integration of even a subset of these data, and we assert that cities-data-sharing.pdf.
any combination has the potential to benefit the well-being [9] C. Batini, and M. Scannapieco, Data and Information
of the individuals affected to a tremendous extent. We have, Quality: Dimensions, Principles and Techniques,
Switzerland: Springer, 2016, pp. 23.
therefore, defined the following features for this purpose: [10] Beautiful Soup. [Online]. Available:
 A data integration architecture and platform on which https://www.crummy.com/software/BeautifulSoup/bs4/d
data from these different sources are stored, oc/.
[11] T. Berners-Lee, R. Fielding, and L. Masinter, “Uniform
semantically annotated and harmonised, thus Resource Identifier (URI): Generic Syntax” 2005.
ensuring a feature rich combination. As described [Online]. Available:
previously, the platform is organised as three different http://www.hjp.at/doc/rfc/rfc3986.html#sec_1.1.
subsystems (Open Data & Acquisition, Semantic and [12] S. Bischof, A. Polleres and S.Sperl, “City data
pipeline”,in:Proceedings of the I-SEMANTICS posters
Smart Data); and and demonstrations track, p 45, 2013.
 A suite of software applications designed for both [13] D. Brickley, R.V. Guha, “RDF Schema 1.1” W3C
smartphones and the web, and to solve specific issues Recommendation, 2014. [Online]. Available:
https://www.w3.org/TR/rdf-schema.
(accessibility, incidents and infrastructure) one at a [14] BUSUP. [Online]. Available: https://www.busup.com/.
time, proving that even small benefits have a [15] Ciudades Patrimonio de la Humanidad. [Online].
significant impact on the global ecosystem. Available:
http://www.ciudadespatrimonio.org/accesibilidad/?idio
Our goal has been to provide a generic architecture with ma=en.
which to support the integration of data into the urban [16] P. Cáceres, C. E. Cuesta, A. Sierra-Alonso, B. Vela, J. M.
mobility platform, serving as the basis for several different Cavero and M. A. Garrido, “Even Smarter Data: Using
Crowdsourcing to Improve Accessibility in Real-Time,”
applications. Each of them is designed to improve the well- in Proc. SMART ACCESSIBILITY 2019, Athens, Greece,
being of urban citizens, and to enrich the platform itself, 2019, pp. 28-32.
increasing its own aggregate intelligence in order to make [17] P. Cáceres, C. E. Cuesta, B. Vela, J. M. Cavero, and A.
more sophisticated decisions. As noted above, the key aspect Sierra-Alonso, “Smart Data at Play: Improving
Accessibility in the Urban Transport
as regards providing this generic integration is the use of System”,Behaviour& Information Technology, 39(6), pp.
semantic technologies. 681-694, 2020, DOI: 10.1080/0144929X.2019.1652852.
With regard to future work in this area, we intend to define [18] P. Cáceres, A. Sierra-Alonso, C. E. Cuesta, J. M. Cavero,
and B. Vela, “Enhancing Accessibility Information in
a systematic and generalised web scraping method that will Google Maps: Adding new pieces of information to
be based on an underlying reference data model, currently GTFS to improve accessibility” In Proceedings on the
under development, which is directly related to the data Second International Conference on Universal
found on websites. We also plan to apply generic data Accessibility in the Internet of Things and Smart
Environments (SMART ACCESSIBILITY 2017), pp. 6-9,
processing technologies (such as RML mappings) to these Nice (France), 2017.
raw collected data, thus providing a unified means to [19] P. Cáceres, A. Sierra-Alonso, C. E. Cuesta, B. Vela, and
normalise data and transform them into their native semantic J. M. Cavero, “Modelling and Linking Accessibility Data
(RDF) format. In summary, our work continues in both in the Public Bus Network”, Journal of Universal
Computer Science, vol. 21, no. 6, pp. 777-795, 2015.
directions, and any improvement made to any particular [20] P. Cáceres, A. Sierra-Alonso, B. Vela, J. M. Cavero, and
detail will now have a global impact on the whole C. E. Cuesta, “Towards Smart Public Transport Data: A
architecture. Specific Process to Generate Datasets Containing Public
Transport Accessibility Information” In Proceedings on
the Third International Conference on Universal
REFERENCES
Accessibility in the Internet of Things and Smart
[1] Accessmap. [Online]. Available:
Environments (SMART ACCESSIBILITY 2018), pp. 66-
https://www.accessmap.io.
71, Rome (Italy), 2018.
[2] Access2All, EU FP7 project, “Mobility schemes ensuring
[21] P. Cáceres, A. Sierra-Alonso, B. Vela, J. M. Cavero, M.
accessibility of public transport for all users”. [Online].
A. Garrido, and C. E. Cuesta, “Normalizing Public
Available:
Transport Accessibility Data in Real-Time”,
https://cordis.europa.eu/project/rcn/89881/factsheet/en.
International Journal on Advances in Intelligent Systems
[3] V. Albino, U. Berardi, and R.M. Dangelico, “Smart
(IJAIS), 12(3&4), pp 191-202, 2020.
Cities: Definitions, Dimensions, Performance, and
[22] P. Cáceres, A. Sierra-Alonso, B. Vela, J. M. Cavero, M.
Initiatives”, Journal of Urban Technology, 22(1), pp. 3-
A. Garrido, and C. E. Cuesta, “Adding Semantics to
21, 2015, DOI: 10.1080/10630732.2014.942092.
Enrich Public Transport and Accessibility Data from the

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3033584, IEEE Access

Web,”Open Journal of Web Technologies (OJWT), 7(1), [41] GTFS, “Google Transit Feed Specification,”. [Online].
pp 1-18, 2020. Available: https://developers.google.com/transit/gtfs/.
[23] C. Cardonha, D. Gallo, P. Avegliano, R. Herrmann, F. [42] G.L. Hajba, Website Scraping with Python. Using
Koch, and S. Borger, “A crowdsourcing platform for the BeautifulSoup and Scrapy, APress, Library of Congress
construction of accessibility maps”, In: Proceedings of Control Number: 2018957273.
the 10th International Cross-Disciplinary Conference on [43] R.E. Hall, “The Vision of a Smart City.” Proc. of the 2nd
Web Accessibility - W4A ’13, pp. 1-4, 2013, DOI: International Life Extension Technology Workshop,
10.1145/2461121.2461129. Paris, France, 2000.
[24] CIVITAS Initiative, “Cleaner and Better Transport in [44] C. Harrison, B. Eckman, R. Hamilton, P. Hartswick, J.
Cities”, 2012. [Online]. Available: http://www.civitas.eu. Kalagnanam, J. Paraszczak, and P.
[25] CKAN. [Online]. Available: https://ckan.org/. Williams,“Foundations for Smarter Cities,” IBM Journal
[26] J. Clark and S. DeRose, “XML Path Language of Research and Development, 54(4), pp. 1–16, 2010.
(XPath)”,W3C Recommendation, 2016, [Online]. [45] S.J. Humer and E. Foster, “A Comparative Analysis of
Available: https://www.w3.org/TR/1999/REC-xpath- the C++, Java, and Python Languages”, Project:
19991116/. Principles of Programming Languages, 2014.
[27] H. Chourabi, N. Taewoo, S. Walker, J.R. Gil-Garcia, S. [46] Identification of Fixed Objects In Public Transport
Mellouli, K. Nahon, T.A. Pardo, and H.J. Scholl, (IFOPT), CEN/TC 278 Intelligent transport systems -
“Understanding Smart Cities: An Integrative Public transport Standard EN 28701, 2012.
Framework,” Proc. of HICSS, 45th Hawaii Conference, [47] H. Jeekel, “Social Sustainability and Smart Mobility:
pp. 2289–2297, 2012. Exploring the relationship”, In F. Ulengin, K. Li and M.
[28] C. E. Cuesta, P. Cáceres, B. Vela, and J. M. Cavero, Boltze (Eds.), Transportation Research Procedia,
“CoMobility: A Mobile Platform for Transport Sharing”, Special issue: World Conference on Transport Research
ERCIM News, no. 93, pp. 22-23, 2013. - WCTR 2016 Shanghai, 10-15 July 2016, Vol. 25, pp.
[29] R. Cyganiak, D. Wood, M. Lanthaler, “RDF 1.1: 4296-4310, 2017.
Concepts and Abstract Syntax”. W3C Recommendation, [48] D. Kouzis-Loukas, Learning Scrapy, UK: Packt
2014. [Online]. Available: Publishing, 2016.
https://www.w3.org/TR/rdf11-concepts/. [49] S. L. Lau and S. M. S. Ismail, "Towards a real-time public
[30] F. Darari and R. Manurung, “LinkedLab: a linked data transport data framework using crowd-sourced passenger
platform for research communities”, in: Advanced contributed data", Vehicular Technology Conference
computer science and information system (ICACSIS), pp (VTC Fall) 2015 IEEE 82nd, pp. 1-6, Sept 2015.
253–258, 2011. [50] X. Liu, A. Heller and P.S. Nielsen, “CITIESData: a smart
[31] A. Dimou, M. Vander Sande, P. Colpaert, R. Verborgh, city data management framework”, Knowledge
E. Mannens, R. Van de Walle, RML: A generic language Information Systems, vol. 53, no. 3, pp 699–722, Dec.
for integrated RDF mappings of heterogeneous data, in: 2017, DOI:10.1007/s10115-017-1051-3.
Proceedings of the 7th Workshop on Linked Data on the [51] P. Lombardi, S. Giordano, H. Farouh, and W. Yousef,
Web, 2014. “Modelling the Smart City Performance”, Innovation:
[32] EMT Madrid home page. [Online] Available: The European Journal of Social Science Research, 25(2),
http://www.emtmadrid.es/ Index.aspx?lang=en-GB. pp.137–149, 2012.
[33] B. Ferris, K. Watkins, and A. Borning. “OneBusAway: [52] V. Lopez, S. Kotoulas, M.L. Sbodio, M. Stephenson, A.
Results fromProviding Real-Time Arrival Information Gkoulalas-Divanis and P.M. Aonghusa, “QuerioCity: a
for Public Transit”. In Proc. CHI 2010, Atlanta, Georgia, linked data platform for urban information management”,
April 10–15, 2010. The semantic web, pp 148–163, 2012.
[34] Fuseki. [Online]. Available: [53] M.D. Lytras, and A. Visvizi, “Who Uses Smart City
http://jena.apache.org/documentation/fuseki2/. Services and What to Make of It: Toward
[35] S. Gaggi, T. Fluhrer and T. Janitzek, “Innovation in urban Interdisciplinary Smart Cities Research”,
mobility: Policy making and planning. Directorate- Sustainability, 10(6), 2018, DOI: 10.3390/su10061998.
General for Mobility and Transport”. Luxembourg: [54] M.D. Lytras, A. Visvizi, and A. Sarirete, “Clustering
European Union. [Online]. Available: Smart City Services: Perceptions, Expectations,
https://www.kowi.de/Portaldata/2/Resources/fp/tripurba Responses”, Sustainability, 11(6), 1669, 2019, DOI:
n-mobility.pdf. 10.3390/su11061669.
[36] F. Gandon and G. Schreiber, “RDF 1.1 XML Syntax” [55] Mediate, EU FP7 project, “Methodology for describing
W3C Recommendation, 2014. [Online]. Available: the accessibility of transport in Europe. [Online].
https://www.w3.org/TR/rdf-syntax-grammar/. Available: https://trimis.ec.europa.eu/, 2011.
[37] F. Gao, M.I. Ali and A. Mileo, “Semantic discovery and [56] Metro Madrid home page. [Online]. Available:
Integration of urban data streams”, in Proceedings of the http://www.metromadrid.es/en/.
5th workshop on semantics for smarter cities, pp 15–30, [57] R. Mitchell, “Web Scraping With Python. Collecting
2014. Data from the Modern Web”, 2015, Published by
[38] M. A. Garrido, P. Cáceres, B. Vela, C. E. Cuesta, J. M. O’Reilly Media.
Cavero, and A. Sierra-Alonso, “Consuming Web Data in [58] MongoDB. [Online]. Available:
a Guiding App for Public Bus Users,” OJWT, vol. 5, no. https://www.mongodb.com/what-is-mongodb.
1, pp. 31-43, Sep. 2018. [59] Moovit. [Online]. Available: https://moovit.com/.
[39] M. A. Garrido, and P. Cáceres,"A Data-driven [60] T. Nam, and T.A. Pardo, “Conceptualizing Smart City
Methodology to Generate Living Ontologies", Applied with Dimensions of Technology, People, and
Ontology, unpublished. Institutions”, Proc. 12th Conference on Digital
[40] R. Giffinger, C. Fertner, H. Kramar, R. Kalasek, N. Government Research, College Park, MD, June 12–15,
Pichler-Milanović, and E. Meijers, “Smart Cities: 2011.
Ranking of European Medium-sized Cities”, (Vienna: [61] B. Ning, H. R. Dong, W. Ding, L. F. Li and C. J. Cheng,
Centre of Regional Science (SRF), Vienna University of "ACP-Based Control and Management of Urban Rail
Technology, 2007. Transportation Systems, " IEEE Intelligent Systems, vol.
[Online].Available:http://www.smart- 26, no. 2, pp. 84-88, 2011.
cities.eu/download/smart_cities_final_report.pdf. [62] E. Prud'hommeaux and A. Seaborne, “SPARQL 1.1
Query Language for RDF”, W3C Recommendation,

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.3033584, IEEE Access

2008. [Online]. Available: https://www.w3.org/TR/rdf- [85] J. Zimmerman et al., “Field Trial of Tiramisu: Crowd-
sparql-query/. Sourcing Bus Arrival Times to Spur Codesign”. In Proc.
[63] OpenTRipPlanner. [Online]. Available: CHI 2011, New York, 2011, pp. 1677–1686.
http://www.opentripplanner.org/.
[64] OpenStreetMap. [Online]. Available:
https://www.openstreetmap.org.
[65] D. Puiu et al., "CityPulse: Large Scale Data Analytics PALOMA CÁCERES received an MSc in
Framework for Smart Cities," in IEEE Access, vol. 4, pp. Computer Science from the Polytechnic University
1086-1108, 2016, DOI: of Madrid (Spain) in 1993 and a PhD in Computer
10.1109/ACCESS.2016.2541999. Science from Rey Juan Carlos University (Madrid,
[66] Python. [Online]. Available: https://docs.python.org/3. Spain) in 2006. She is an associate professor at Rey
[67] Python Documentation. [Online]. Available: Juan Carlos University. Her research interests focus
https://docs.python.org/3/library/urllib.request.html?hig on software engineering, web engineering, model
hlight=requests%20library. driven development, data science, open and linked data, semantic Web, and
[68] RDF Working Group, “Resource Description Framework transport and accessibility. She is the author and co-author of several
(RDF)” W3C Recommendations, 2014. [Online]. national and international papers related to these areas.
Available: http://www.w3.org/RDF/.
[69] Request Library. [Online]. Available: https://2.python-
requests.org. ALMUDENA SIERRA-ALONSO received her
[70] H. Santos, P. Pinheiro and D.L. McGuinness, “Contextual Ph.D. in Computer Science from the Universidad
data collection for smart cities”, in: Proceedings of the Politécnica de Madrid, in 2000. She is an associate
6th workshop on semantics for smarter cities, 2015. professor in the Computer Science and Statistics
[71] T. Sarjakoski et al., “Landmarks and a Hiking Ontology Department of Rey Juan Carlos University. Her
to Support Wayfinding in a National Park During research interests are focused on software
Different Seasons”. In M. Raubal, D. M. Mark, and A. U. engineering, data science, open and linked data, the
Frank (Eds.), Cognitive and Linguistic Aspects of semantic Web, and transport and accessibility. She
Geographic Space, Lecture Notes in Geoinformation and is the author and co-author of several papers related to software engineering,
Cartography. Springer, pp. 99-119, 2013. the semantic web and teaching in Engineering Education (in areas such as
DOI:10.1007/978-3-642-34359-9. operating systems and software engineering).
[72] J. Schlingensiepen, R. Mehmood and F.C. Nemtanu,
“Framework for an Autonomic Transport System in
Smart Cities”, Cybernetics And Information CARLOS E. CUESTA is an Associate Professor of
Technologies, Vol. 15, no. 5, Special Issue on Control in Software Engineering at Rey Juan Carlos University
Transportation Systems, 2015, DOI: 10.1515/cait-2015- (Madrid, Spain). He has a PhD in Information
0016. Technologies (2002) from the University of
[73] Scrapy. [Online]. Available: https://scrapy.org/. Valladolid. His main research area is Software
[74] SpringBoot. (2019). Spring Boot. [Online]. Available: Architecture, including such topics as Self-Adaptive
http://spring.io/projects/spring-boot. Systems and Systems-of-Systems, and with relation
[75] Swiftly. [Online]. Available: https://www.goswift.ly/. to Concurrency Theory, Formal Methods,
[76] Transmodel standardization, E. C. Transmodel, Road Distributed Systems or Data Engineering. He has
Transport and Traffic Telematics Public Transport. Ref. been the Program Chair of the 6th and 12th European Conferences on
Data Model, EN 12896, 2015. Software Architecture (ECSA 2012 & 2018). He has published in premier
[77] E.J. Tomaszewska and A. Florea, “Urban smart mobility conferences and journals, such as the International Journal of Information
in the scientific literature - bibliometric analysis”, Technology and Decision Making, or Future Generation Computer Systems.
Engineering Management in Production and Services,
Volume 10, Issue 2, pp. 41-56, 2018, ISSN 2543-6597,
Poland, DOI: 10.2478/emj-2018-0010. BELEN VELA has an MSc in Computer Science
[78] B. Vela, J. M. Cavero, P. Cáceres, A. Sierra-Alonso and (Carlos III University) and a PhD in Computer
C. E. Cuesta, “Defining a NoSQL document database of Science (Rey Juan Carlos University). She is now an
accessible transport routes,” in Proc. IEEE SMART Associate Professor in the Department of Computing
DATA, Exeter, England, United Kingdom, 2017, pp. Science, Computer Architecture, Programming
1125-1129. Languages and Systems and Statistics and Operative
[79] B. Vela, P. Cáceres, J. M. Cavero, A. Sierra, C. E. Cuesta Investigation at Rey Juan Carlos University (Madrid,
and M. A. Garrido, “Guíame EMT,” SpanishPatent M- Spain). She belongs to the Vortic3 Research Group.
004237/2014, Nov. 18, 2014. Her research interests focus on Software
[80] B. Vela, J. M. Cavero, P. Cáceres, and C. E. Cuesta, “A Engineering, Information System Engineering, Data Science, Open Data,
Semi-Automatic Data–Scraping Method for the Public Databases (NoSQL), Model Driven Development, Transport, Accessibility
Transport Domain,”IEEE Access, Vol.7, pp. 105627 - and Scientometrics. She has participated in and led several research projects
105637, ISSN 2169-3536,DOI: and has had numerous papers published in prestigious journals and
10.1109/ACCESS.2019.2932197. conferences.
[81] VIVO application, https://duraspace.org/vivo/
[82] D. Washburn, U. Sindhu, S. Balaouras, R.A. Dines, N.M.
Hayes, and L.E. Nelson, “Helping CIOs Understand JOSÉ MARÍA CAVERO BARCA has an MSc in
“Smart City” Initiatives: Defining the Smart City, Its Computer Science from the Polytechnic University
Drivers, and the Role of the CIO, Cambridge, MA: of Madrid and a PhD in Computer Science from Rey
Forrester Research, 2010. [Online]. Available: Juan Carlos University. He is an Associate Professor
http://public.dhe.ibm.com/partnerworld/pub/smb/smarter at the School of Computer Science Engineering at
planet/forr_help_cios_und_smart_city_initiatives.pdf. Rey Juan Carlos University. His research interests
[83] Wheelmap. [Online]. Available: https://wheelmap.org. focus on Databases, Open Data, Transport,
[84] Wheelmate. [Online]. Available: Accessibility and Scientometrics. He has been the
http://myhealthapps.net/app/details/13/wheelmate. dean of the School of Computer Science Engineering for four years.

VOLUME XX, 2017 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/