See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/344200922

Reasoning in Common Data Environments Re-thinking CDEs to enhance

collaboration in BIM processes

Conference Paper · September 2020

DOI: 10.52842/conf.ecaade.2020.2.499

CITATIONS READS

7 697

4 authors:

Davide Simeone Stefano Cursi

Salini Impregilo Sapienza University of Rome
70 PUBLICATIONS 749 CITATIONS 40 PUBLICATIONS 498 CITATIONS

SEE PROFILE SEE PROFILE

Ugo Maria Coraglia Antonio Fioravanti

University of Bologna Sapienza University of Rome
31 PUBLICATIONS 76 CITATIONS 114 PUBLICATIONS 413 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Davide Simeone on 11 September 2020.

The user has requested enhancement of the downloaded file.

Reasoning in Common Data Environments
Re-thinking CDEs to enhance collaboration in BIM processes

Davide Simeone1 , Stefano Cursi2 , Ugo Maria Coraglia3 ,

Antonio Fioravanti4
1,2,3,4
Sapienza University of Rome, Department of Civil, Construction and Envi-
ronmental Engineering
1,3,4
{davide.simeone|ugomaria.coraglia|antonio.fioravanti}@uniroma1.it
2
[email protected]

In BIM processes, the concept of Common Data Environment - CDE - has often
been depicted as a key element for successful collaboration and information
sharing among different actors but, in current practice, acts as a mere
documentation repository ineffective for true collaborative purposes. Therefore,
the idea of CDE seems to be overrated on the one hand and unexploited on the
other, while effective collaboration is still far from being decisively supported. To
overcome this lack, the present research focuses on the definition of a new
generation of CDEs, enhanced with an information level for knowledge
integration provided by different information carriers such as models and
datasets. The paper discusses its development through a graph database platform
and dedicated methodologies for data retrieval and query, to verify coherence
and consistency of information among different models.

Keywords: Collaboration in AEC processes, Common Data Environment, Graph

Databases, Building Information Modeling, Queries and data retrieval

INTRODUCTION CDE is the one proposed in the PAS 1192 (2013) and
In current BIM-enabled AEC processes, Common ISO 19650 (2018) norms, composed of different ar-
Data Environment - CDE- has been considered as a eas (WIP, Shared, Published, Archived) where project
key element for successful collaboration and infor- documentation is made selectively accessible to dif-
mation sharing among different actors and stake- ferent stakeholders. Relying on this structure, some
holders. The UK government strategy implies its software industries have proposed their platforms,
adoption within the BIM Level 2 scope and some ad- usually cloud-based, such as ProjectWise (Bentley),
ministrations are progressively mandating its imple- Aconex and, more recently, Autodesk BIM 360. As
mentation in AEC projects. At present, despite a large we look at CDEs how it is described in these norms
amount of CAAD research in the field of Collabora- and it has been implemented a big ambiguity, at least
tive AEC Design process, the most spread structure of from the CAAD research perspective, arises: are we

D2.T8.S2. THE COGNITIVE CITY (AI) - Volume 2 - eCAADe 38 | 499

 dealing with data or are current CDEs acting as doc- tion through comprehensive, coherent and interre-
umentation, although shared, repositories? And, as lated knowledge (Carrara et al., 2017). If compared
a corollary, is this CDE structure effective for collab- to this research direction, research in the CDE field
oration purposes? At present, full potentials of the has been quite static. The concept of CDE was cre-
idea of CDE seem to be unexploited, while effective ated in the IT field as a centralized, unique set of
collaboration among AEC processes actors is still far data for any kind of project that requires collabora-
from being decisively supported by CDEs. In the con- tion and mutual accessibility to data. Later, the con-
text, this research investigates new methodologies cept of CDE has shifted to the AEC field to define
for data, information and knowledge unified formal- the single source of truth for design development,
ization in CDE, based on graph databases, to improve shaped accordingly to major steps in the delivery of
coherence, reliability and accessibility of information an AEC project. On one side, its basic structure and
stored and shared, as well as improving quality of the functioning rely on the original framework provided
integration of models, documents and other infor- by BS PAS 1192 norms, while research has mainly
mation carriers that contributes to the central core of investigated procedural aspects of CDEs with only
data acting as a reference. It essentially focuses on partial attention to methods and tools to enhance
the addition of an information level in current CDEs and ensure the quality, interoperability and richness
and discusses the use of graph databases as a way to of the stored content. On the other side, AEC soft-
successfully implement and manage it. ware houses have spent much effort on CDE devel-
opment, often to improve integration with their pro-
STATE OF THE ART prietary authoring software or to provide scalable,
In the eCAADe scope and, more widely, in the CAAD cloud-based solutions for AEC projects. As a result,
and BIM world, much research has been focusing CDEs are currently used as documents storage en-
on the integration of Building Information Model- vironments, with additional metadata progressively
ing, information and knowledge ontologies (Cursi, introduced to solve documents-controlling common
2017) and linked data as a way to drive, standard- issues such as accessibility, revisioning, file codifica-
ize and extend content and semantics of AEC in- tion, issuing dates, etc. In the last years, the develop-
formative models. Potentials of this approach have ment of tools such as Autodesk Forge has made pos-
been demonstrated by research works like the one sible to extract and elaborate stored information di-
from Beetz (2005), Pauwels (2015) with a progressive rectly in the cloud environment, also to cross-link and
extension of the application to built heritage field compare different models (Yan, 2017), (Preidel, 2016).
(Simeone, 2019) (Di Mascio, 2013). In this field, the Nevertheless, this approach can be only considered a
development of IFC-OWL ontology has represented palliative that relies on the extraction of data, while a
an important shift, introducing a reference ontol- few efforts have been made regarding shaping and
ogy for AEC entities formalization for interoperabil- structuring data, information and knowledge - mod-
ity purpose and as an extendable base for domain- elled in the bim model or adjunct in a database - to
specific semantic enrichment of building informa- ensure horizontal reasoning in the CDE. This lack in
tion models. Similarly, some research has focused current CDEs is progressively emerging in the AEC
on the adoption of information ontologies as an practice, and linked data methodology represents a
approach to improve knowledge sharing and col- potential solution that needs to be further investi-
laboration between different actors and disciplines gated (Werbrouck, 2019).
(Jelokhani-Niarak, 2018) (Zhong, 2018). This is par-
ticularly relevant for this research since it targets one
of the main intended features of CDEs: collabora-

500 | eCAADe 38 - D2.T8.S2. THE COGNITIVE CITY (AI) - Volume 2

INFORMATION GRAPH DATABASES FOR Management System), three are particularly relevant
NEXT-GEN COMMON DATA ENVIRON- for our scope: 1) focus on relationships, that allows
MENTS users to depict them from each model and to create
In CDEs, two main levels can be usually depicted: new relationships between different entities of differ-
1) the information carrier level, and 2) the data/in- ent datasets; 2) flexibility, that allows the same data
formation level. The first one comprises the differ- model to be applied to the different dataset stored
ent kinds of documents, models, reports, calculation in a CDE; 3) easy evolution on time, that can adapt
notes, etc. that support data formalization and rep- to the evolution of the project models better than
resentation and that are usually produced, uploaded, any RDMS, usually static after the formalization of the
edited, and accessed by different actors. The second conceptual model.
level, instead, refers to all elementary data and in-
formation, formalized in the information carriers fol- THE ENHANCED COMMON DATA ENVI-
lowing discipline-specific methodologies and repre- RONMENT SCHEMA
sentation structures. By analyzing current CDEs ap- In this proposed framework for a new generation
plications, it is clear how CDE platforms mainly fo- of CDEs, this information level developed through
cus on the information carrier level, in some cases graph databases is mainly an environment where in-
specifying a simplified set of horizontal relationships formation, derived from different information carri-
among different documents to indicate which of ers through a process of data extraction, selection,
them are related or “linked”. No coherence and ac- and translation, is represented in a homogenous way,
tual correspondence among data stored in those in- structured and compared. To provide this required
formation carriers are declared, checked, and en- homogeneity, the proposed Information Level has
sured, and, therefore, CDEs usually result in heteroge- been conceived and implemented relying on the use
nous, incoherent, and unreliable information stor- of Information ontologies, while a general ontology
age. While information ontologies have shown great structure, based on previous research developed by
potential in abstracting entities and relationships, at this research group, serves as an ontology template
present data and information are currently managed for re-formalization of data extracted by the different
in BIM software through relational databases that Information Carriers (Simeone, 2019). As described
make the entire data structure too rigid and diffi- in figure 1, the passage from the heterogeneous for-
cult to manage for AEC design process, in particular malization of the information carriers to the homo-
when those databases are structured based on pro- geneous and common Information level requires two
prietary software requirements. This issue has been crucial passages:
discussed in the IT world, where an alternative ap-
proach, based on no-SQL databases, and in particular 1. Data selection and extraction;
graph databases, are progressively being adopted, in 2. Data re-formalization and re-structuring in
particular for knowledge formalization and creation graph databases connectable and queryable.
of data lakes that deal with complex data patterns In the proposed system, the first step depends on dif-
rather than single data. In this context, this research ferent methodologies following the variety of infor-
proposes and assesses the use of graph databases mation carriers and to the availability of specific ap-
as a way to access and make homogenous, flexi- plications and tools. In our experiments, we relied
ble and queryable the data provided by different in- on Revit DBLink application for the extraction of data
formation carriers in the CDE, in an integrated way. from models and Dynamo for extraction of specific
Among the different advantages of the adoption of datasets (Simeone, 2017), while for other information
graph databases over RDMS (Relational Databases carriers, such as reports or calculation notes, we con-

D2.T8.S2. THE COGNITIVE CITY (AI) - Volume 2 - eCAADe 38 | 501

 Figure 1
The proposed new
CDEs schema
composed of the
Information Carriers
level and the
Information Level.

sidered useful using synthesis datasheets linked to ent information carriers into a single, homogenous
actual documents. Because of the wide number of knowledge base, hence allowing reasoning opera-
information stored, this extraction activity is coupled tions as well as information comparison and general
with a filtering feature that selects only specific data coherence checking. The use of common represen-
required by the information level. Although this pro- tation standards, such as the use of IFC owl template
cess could be potentially applied to all information (Pauwels, 2015), and agreed formalization method-
formalized, limiting the amount of information for- ologies, such as schedules templates and synthesis
malized is functional to the necessity of having a lean data structures, deeply affects efficiency and speed
system to ensure qualitative coherence of informa- of the process. To operate on the Information Level, a
tion while delegating more strict and accurate data Cypher-based reasoner has been integrated into the
coherence and consistency verification (i.e. clash de- proposed system. The reasoner allows users to de-
tection) to other applications outside the Common velop and execute reasoning queries to verify and
Data Environment. The second passage is instead compare data stored in the information level. For in-
based on the re-organization of the extracted and se- stance, it allows verifying if the number of spans of a
lected data in a Common graph database that en- bridge as design in a BIM model is coherent with the
sures information to be formalized in a way to al- data stored in its calculation report, alerting the de-
low check, comparison, validation among different sign team in case of incoherencies and discrepancies.
datasets (Fig.1). Specific sets of reasoning rules can be used for spe-
Each model template requires a specific transla- cific features of the design to be checked directly in
tion method depending on the BIM authoring tool the CDE, improving actors’ awareness of the state of
and its database structure. This aspect only par- the information shared in the collaboration platform.
tially hinders the process because of the simplicity
of the translation methods and their re-usability in
similar AEC processes. This two steps process trans-
lates information heterogeneously provided in differ-

502 | eCAADe 38 - D2.T8.S2. THE COGNITIVE CITY (AI) - Volume 2

 THE IMPLEMENTATION OF THE IN- and there is a way to recognize the correspondence
FORMATION LEVEL THROUGH GRAPH (such as the same ID, or the reference to the same
DATABASES structural grid point) it is possible to create a relation-
To manage, query and capitalize the data introduced ship corresponding to and start connecting the two
in the CDE by different information carriers, this re- graphs.
search developed the information level relying on
Figure 2
the use of the graph database technology, imple-
A portion of the
mented through the Neo4J platform (Francis, 2018).
tested graph
In this way, it is possible to have a homogenous
database that
way of formalization of information extracted from
shows entities
or connected to the different models provided. Each
(nodes) and
model provides a set of information that can be or-
relationships.
ganized in a network of entities, relationships, and
Labels represent
properties that can be retrieved and used in simple or
family/type classes
complex queries. In the case of BIM models, usually
as per usual BIM
organized through the family->type->instance struc-
elements schema.
ture, the graph databases nodes are the instances of
the model (the ones usually defined by an ID in the
models, such as a specific pillar, a column, etc.) while
the labels provide the family/type classification. Re-
lationships among nodes represent the constraints
and the link between entities in the BIM Model, such
Figure 3 as a floor constrained to a specific level (fig. 2 and fig.
The same portion of 3).
the graph database In the case of other models, such as planning
in fig.2, expressed schedules, Quantity Take Off models, or 4D models,
as a table to show it is usually possible to directly access to modelling
the formalization of entities (i.e. tasks, take off items, etc.) and their con-
properties of necting relationships, recreating graph databases. A
entities as extracted bit more difficult is the extraction of data from other
by the BIM models. information carriers such as reports, etc. In that case,
it is necessary to have at least the relevant data sets to
translate them in a different data graph. At first, the
graph databases elaborated from the different mod-
els reside in the information level as independent, co-
herent graphs but, since they are formalized homo-
geneously, it is possible to construct queries and al-
gorithms to 1) run across in a transversal direction
between different graphs, 2) access and verify the
information and, if conditions are verified, generate
new relationships between nodes/entities in differ-
ent graphs. For instance, if the same building compo-
nent (i.e. a beam) is represented in different models

D2.T8.S2. THE COGNITIVE CITY (AI) - Volume 2 - eCAADe 38 | 503

 lows for a selective extraction that can be tailored fol- Figure 4
lowing the query necessities of each project. Both An example of a
these methods provide a bidirectional connection constructed query
For this research, we decided to rely on the use of between the models and the dataset elaborated in in Cypher language,
Cypher, a declarative query language that enables the CDE information level. Current Visual program- that retrieves
both queries and data formalization, specifically de- ming approaches are a relevant contribution to this structural elements
signed to operate in graph databases environment process, allowing to select, filter, and export datasets (“MATCH” part) and
and interact with nodes, relationships, properties, in formats such as .csv that allow graph databases ed- generates new
and patterns. In this proposed evolution of CDEs, the itor to access data and automatically reconstruct the relationships based
creation and control of ‘horizontal queries’ between network of entities. Another opportunity, at present on the structural
graph databases represent a new task in the cur- only available in the exporting direction, is repre- grid (“CREATE”
rent AEC information management processes, that sented by the use of the IFC schema. In this case, the part).
can impact the current role of the project informa- use of the IFC-OWL ontology (Pauwels, 2015) is a ref-
tion manager. In this task, the set of queries neces- erence for the implementation of standardized pro-
sary to connect different data models stored in the cedures for its conversion in a graph database, and
CDE is project-dependent because it has to be tai- the construction of queries accessing the IFC data
lored in accordance to the scope of each model, the (fig. 5).
formalization structure of its data, and the objectives
of the query itself. If compared to other query lan- Figure 5
guages (i.e. the SWRL - Semantic Web Rule Lan- Interoperability
guage), Cypher supports this process of queries set between BIM
customization using its syntax based on natural lan- Models and Graph
guage and its very linear logic. As in the example databases in the
shown in figure 4, the use of keywords such as MATCH CDE Information
and CREATE allows structuring complex queries to Level, based on
recognize patterns in the graph databases and gen- different data
erate new data, providing a query template that can exchange
be adapted to similar necessities in other projects.
POTENTIALS AND OPEN PROBLEMS OF
methodologies
The development and refining of query sets for AEC GRAPH DATABASES APPROACH TO COM- (RDMS, Dynamo,
projects is an open topic that can introduce a new MON DATA ENVIRONMENTS IFC schema).
standardization level in the current practice. In our experiments on the adoption of graph
For data extraction from BIM Models, and their databases as a new way of formalizing and make
connection to the graph databases in the informa- available information in AEC Common Data Environ-
tion level, different solutions can be conceived rely- ments, some potentials and limits have emerged, not
ing on approaches and tools already available in the specifically related to technical aspects but rather to
AEC sector. In the case of Autodesk Revit models, we the introduction of this approach within already con-
utilized two different methodologies that rely on the solidated workflows and protocols. The processes
use of the Revit DBLink and the use of Dynamo al- described in this paper underlay the more general in-
gorithms for data extraction in tables. We found the formation management protocols and practices that
DBLink approach useful in the case of necessity of a drive current AEC projects, essentially dealing with
bulk translation in a graph of the entire model, al- how to formalize information and knowledge. From
though some issues arise in terms of manageability an end-user perspective, it is not possible to define
and conversion from the relational database to the a unique workflow but we rather suggest that some
graph one. The use of dynamo scripts, instead, al- additional actions have to be carried out, while shar-

504 | eCAADe 38 - D2.T8.S2. THE COGNITIVE CITY (AI) - Volume 2

ing documents and models, to favour the develop- CONCLUSIONS AND FUTURE PROSPECTS
ment of a coherent and efficient Single Source of The research described in this paper aims at enhanc-
Truth as necessary in AEC projects. The data trans- ing the current Common Data Environment for BIM-
lation from models - or other information carriers - to oriented processes, by integrating the CDE reposi-
graph databases is an activity not yet considered in tory structure with a knowledge-based system, de-
current AEC collaboration processes, that requires ef- veloped through graph databases. Its scope is to ex-
forts, competencies and time, especially if performed tend CDEs usual repository function with the pos-
cyclically as in architectural design. Heterogeneity sibility to perform qualitative checking and evalua-
of models is a hindering factor that requires the de- tion of the data stored in the different models and
velopment of customized algorithms and protocols information carriers, immediately detecting poten-
for data extraction, impacting the adoption of this tial incoherence and improve design integration. The
new approach to the entire project dataset. Another conceptual framework of the presented system re-
obstacle is represented to the necessity of new fig- lies on a semi-automated process of data extraction,
ures in the process, able to manage and query this selection and translation into a homogeneous set
new CDEs: the graphs themselves can easily become of graph databases that allows reasoning operation
too complex, and database experience is, therefore, and data accessibility. As shown by first tests in sim-
necessary to control that complexity and ensure ef- ulated AEC processes, the proposed system is po-
ficacy in data retrieval. At the same time, we as- tentially able to enhance the efficacy of a crucial el-
sessed relevant improvements on the quality, con- ement of BIM processes - the Common Data Envi-
sistency and coherence of the information formal- ronment - by transforming it from a repository of
ized through graph databases, as well as in its un- not-coherent documentation to an integrated data
derstanding by the different actors involved in the environment oriented to actual collaboration. Also,
process. Simplified interfaces, for instance, allow the introduction of CDE information level developed
the actors to explore the data graphs reaching the through graph databases allows for enhanced flexi-
needed information and being supported in its in- bility in terms of queries and verifications that can be
terpretation by visualizing the connected informa- performed by connecting and moving across differ-
tion. We also noticed that the effectiveness of this ent models datasets, without constraining the users
approach is higher if not applied to the single ele- to rely only on pre-defined metadata controls as in
ments that compose the building but rather oper- current CDEs. The presented approach, although
ating at a higher level of representation where de- still explorative, opens the research to further devel-
cisions are made on systems and general aspects of opments and even integration with other research
the building, while highly detailed, discipline-specific streams that could be beneficial for the entire sector.
issues are usually more manageable through dedi- In particular, research is still needed to make the en-
cated tools. This approach also provides a relevant tire workflow more efficient by reducing the variabil-
improvement in terms of queries capabilities: hori- ity of stored data and the re-usability of queries. The
zontal queries allow to move across the different data use of ontology-based schemas such the IFC-OWL
formalized in the different models, while the Cypher could ensure a better homogeneity of the formalized
language allows to perform also complex systems of data and, at the same time, allowing for a standard-
queries and to elaborate a higher level of semantics, ized library of queries. Another aspect, not often dis-
usually not accessible by direct operations on the sin- cussed in our field, is related to the integration of
gle models. information sources different from models, that are
currently attached as external documents: further re-
search should focus on re-thinking such way of com-

D2.T8.S2. THE COGNITIVE CITY (AI) - Volume 2 - eCAADe 38 | 505

 municating a project, formalizing information with Di Mascio, D and Pauwels, P 2013 ’Improving the knowl-
different technologies and ensuring automation in edge and management of the historical built envi-
data elaboration. Another research step to be con- ronment with BIM and ontologies. The case study
of the book tower’, CONVR2013 (eProceedings of the
sidered is the use of machine learning techniques for
13th International Conference on Construction Appli-
queries and data retrieval, further enhancing the effi- cations of Virtual Reality), London, pp. 427-436
cacy of the system. To conclude, in the scope of cur- Pauwels, P and Terkaj, w 2015, ’EXPRESS to OWL for Con-
rent research AEC processes research, this work con- struction Industry: Towards a Recommendable and
tributes to the current quest for a new generation Usable ifcOWL Ontology’, Automation in Construc-
of collaborative digital environments, able not only tion, 63, pp. 100-133
Preidel, C, Borrmann, A, Oberender, CH and Trethe-
to make data and information accessible but also to
way, M 2016 ’Seamless integration of common data
integrated them, ensuring their coherence and their environment access into BIM authoring applica-
comprehension to all the actors involved in the pro- tions: The BIM integration framework’, E-work and E-
cess. business in architecture, engineering and construction
Simeone, D and Cursi, S 2017 ’A Platform for Enriching
BIM Representation Through Semantic Web Tech-
REFERENCES nologies’, LC3 2017: Volume I - Proceedings of the Joint
Beetz, J, Van Leeuwen, J and De Vries, B. 2005 ’An ontol- Conference on Computing in Construction (JC3), Her-
ogy web language notation of the industry founda- aklion, pp. 423-430
tion classes’, Conference on Information Technology in Simeone, D, Cursi, S and Acierno, M 2019, ’BIM semantic-
Construction, , CIB-W78, Dresden, p. 193–198 enrichment for built heritage representation’, Au-
BS, PAS 1192:2 2013 (eds) 2013, - Specification for infor- tomation in Construction, 97, pp. 122-137
mation management for the capital/delivery phase of Werbrouck, J, Pauwels, P, Beetz, J and van Berlo, L
construction projects using building information mod- 2019 ’Towards a Decentralised Common Data Envi-
elling, BSI, London ronment using Linked Building Data and the Solid
Carrara, G, Fioravanti, A, Loffreda, G and Trento, A Ecosystem’, Proceedings of the 36th CIB W78 Confer-
2017, Knowledge collaboration design. Theory tech- ence on Information Technology in Construction
niques and applications for collaboration in architec- Yan, W 2018 ’WP-BIM: Web-based Parametric BIM To-
ture, Gangemi Editore, Rome wards Online Collaborative Design and Optimiza-
Cursi, S, Simeone, D and Coraglia, UM 2017 ’An ontology- tion’, ShoCK! - Sharing Computational Knowledge! -
based platform for BIM semantic enrichment’, ShoCK Proceedings of the 35th eCAADe Conference, Rome,
– Sharing of Computable Knowledge - Proceedings of pp. 527-534
the 35th eCAADe Conference, Rome, pp. 649-656 Zhong, B, Ganab, C, Luo, H and Xing, x 2018, ’Ontology-
Francis, N, Green, A, Guagliardo, P, Libkin, L and Lin- based framework for building environmental moni-
daake, T 2018 ’Cypher: An Evolving Query Lan- toring and compliance checking under BIM environ-
guage for Property Graphs’, SIGMOD’18 Proceedings ment’, Building and Environment, 141, pp. 127-142
of the 2018 International Conference on Management
of Data, Jun 2018, Houston, p. 1433
ISO, 19650-1:2018 2018, Organization and digitization
of information about buildings and civil engineering
works, including building information modelling (BIM)
— Information management using building informa-
tion modelling — Part 1: Concepts and principles, ISO-
International Organization for Standardization
Jelokhani-Niaraki, M 2018, ’Knowledge sharing in Web-
based collaborative multicriteria spatial decision
analysis: An ontology-based multi-agent approach’,
Computers, Environment and Urban Systems, 72, pp.
104-123

506 | eCAADe 38 - D2.T8.S2. THE COGNITIVE CITY (AI) - Volume 2

View publication stats