Tips from the Experts

Prior Steps into Knowledge Mapping: Text Mining

Application and Comparison
Faizhal Arif Santosa
Librarian for Polytechnic Institute of Nuclear Technology
National Research and Innovation Agency
D. I. Yogyakarta, ID
[email protected]

Abstract
Bibliometrics is increasingly being used by the knowledge community and librarians to
easily analyze patterns in knowledge. In the field, the use of data from databases that
provide bibliometric information is not always completely clean, so pre-processing is
required. Several previous studies have shown that bibliometric analysis begins with a
simple pre-processing step. The goal of this research is to use text mining to perform
pre-processing to find the basic terms of the keywords that appear – to essentially
construct a controlled vocabulary for a bibliographic dataset. The method used in this
study is cleaning keywords with the stemming method using RapidMiner software.
Bibliometrix was used to compare the results. A total of 85 keywords were combined
into basic words. Using the built process, this study discovers differences in the
network built between raw data and data that has been pre-processed, resulting in
differences in the analysis that will be produced. The built process can also be reused in
a variety of real-world situations.

Keywords: Bibliometric mapping, Science mapping, Stemming, Text pre-processing

Recommended citation:

Santosa, F. A. (2022). Prior steps into knowledge mapping: Text mining application and
comparison. Issues in Science and Technology Librarianship, 102.
https://doi.org/10.29173/istl2736
Introduction
Today's publications are rapidly expanding, and volumes are larger than ever before,
making bibliometrics an excellent tool for mapping knowledge over time. Librarians are
frequently involved in this matter to assist researchers and general users in identifying
trends and gaps in a subject, and this response has resulted in bibliometrics becoming a
service offered by many libraries today. Bibliometrics has become a suitable area for
modern librarians due to the profession's extensive understanding of bibliographic
sources (Gumpenberger et al., 2012). However, bibliometrics can presents libraries with
a challenge. This mapping is very complicated and time-consuming because it
necessitates a wide range of software and many steps (Aria & Cuccurullo, 2017).

Various software packages are now available to make it easier for librarians to provide
bibliometric services, each with its own set of features and capabilities. Bibliometrix,
CiteSpace, CitNetExplorer, SciMAT, Sci2 Tool, Pajek, and VOSviewer are some
examples of bibliometric analysis software. Moral-Muñoz et al. (2020) add that
Bibliometrix provides a diverse set of techniques, whereas VOSviewer provides good
visualization to facilitate a more detailed examination. In general, tools for performing
bibliometric analysis can see the relationship between authors and the relationship
between keywords in one document with another, and several others offer analysis
using abstracts from various databases.

For librarians who have conducted bibliometric analysis, the existence of Web of
Science (WoS) and Scopus can aid in the retrieval of bibliographic data, and some
software has even provided a feature to pull data directly and process it via an API
(application programming interface). This data contains a variety of information that
was used in the analysis, such as the names of all authors, keywords used, and abstracts
if they were available. However, the resulting data is typically not clean enough for
immediate analysis, requiring a different approach during the data pre-processing
stage.

According to CRISP-DM, data quality improvement can be handled during the data
preparation phase by cleaning the data prior to the modeling phase (Schröer et al.,
2021). Data cleaning is a generic task in and of itself, but when faced with actions that
require specific situations, specialized task levels must be carried out (Chapman et al.,
2000), and data cleaning is a prerequisite for performing bibliometric analysis (Moral-
Muñoz et al., 2020). Checking the data before analyzing it becomes necessary in order to
produce precise and useful information, particularly in bibliometric analysis, which is
based on bibliographic data entered into the application.

Various researchers who employ bibliometric analysis have also investigated the
significance of this pre-processing stage. Wang et al. (2020) used VOSviewer and
CiteSpace to perform pre-processing directly on the WoS Core Collection database and
generate 831 publications, but the steps are unclear. To eliminate duplicate data, several
supporting software can be used, such as CiteSpace (Li et al., 2020; Wang et al., 2021)
and Excel (Obidat, 2022). Han et al. (2020) used Python to perform pre-processing by
removing common terms, compiling a list of synonyms, combining singular and plural
words, and academic compound nouns. Meanwhile, CheshmehSohrabi & Mashhadi
(2022) use text mining by using Excel and RapidMiner software to perform pre-
processing in the form of deeper checking of the dataset obtained from Scopus and
finding 134 book reviews listed as article reviews, as well as filling in the blank cells
with the number zero.

RapidMiner has been used in the scope of the libraries due to the ease with which
various algorithms can be applied without the need for programming skills. Lamba and
Madhusudhan (2018) performed a sentiment analysis where the data from Twitter
required for his study was obtained using RapidMiner and analyzed using AYLIEN.
Meanwhile, Moore (2017) employed RapidMiner to analyze sentiment in open-ended
qualitative LibQUAL+ comments.

Manual pre-processing of large amounts of data will be costly and time-consuming for
librarians. The author attempts to perform pre-processing using text mining using the
RapidMiner application without the need for programming skills, with the main focus
that the author proposes, namely keywords. The issue is that keywords with plural and
singular values must be combined. This process's main contribution is that it can be
reused by librarians to support bibliometric services in libraries.

Methods
As a test sample, data from Scopus was used as an application example, and 849
records about academic libraries from 2016 to 2021 were imported into RapidMiner by
selecting the UTF-8 encoding file. The information gathered was in the form of
“Citation information” and “Abstracts & keywords”. Meanwhile, the authors used
Bibliometrix to compare results.

Stemming Snowball (Porter, 2001) is used to find the root of a keyword. For example,
“librarian” and “librarians” would be changed to “librarian”, while “library” and
“libraries” would become “librari”. Both author and index keywords are processed,
with the final result being a CSV (Comma-separated values) file. The trial on a co-
occurrence network was performed in Bibliometrix without changing any of the
available parameters by selecting author keywords and the clustering algorithm,
Walktrap.

Findings
The initial file was counted as having 1987 keywords. “Academic libraries” and
“academic librarians” are the top keywords (Figure 1). According to Bibliometrix, the
use of original data from Scopus results in author keywords that are divided into many
clusters, with a total of 12 clusters (Figure 2). “Academic libraries” and “academic
librarians” are the most numerous nodes in Cluster 1. Meanwhile, “academic librarian”
is in cluster 3, and “academic library” is in cluster 8. These are actually matching words
and are separated from each other because they are treated as distinct entities, resulting
in a multitude of nets and clusters.
 Figure 1. Top ten author keywords before processing

Figure 2. Bibliometrix's keyword network using Scopus data

 Figure 3. Top ten author keywords after pre-processing

Using pre-processed data, the number of keywords is reduced to 1902, or as many as 85

keywords have been combined into certain words. The increasing number of keyword
occurrences in several keywords demonstrates this transformation (Figure 3). Different
clustering results are displayed in pre-processed files (Figure 4). The word appears to
change from “academic libraries” and “academic library” to a new word, “academic
librari”. Meanwhile, “academic librarians” was merged into the previously existing
word “academic librarian”. The formed cluster also changed significantly by having
only 5 clusters, whereas clusters 4 and 5 only had 1 keyword. The large cluster is split
into two parts, with “academic librari” and “academic librarian” as the largest nodes in
each.

Figure 4. Keyword network on Bibliometrix after pre-processing

Conclusions
Pre-processing prior to bibliometric analysis can help provide better analytical results.
“Garbage in, garbage out,” a popular computer phrase, indicates the need for anyone
performing bibliometric analysis, including librarians, to perform pre-processing such
as data cleaning and uniformity. Pre-processing keywords results in significant
changes, which have implications for changing the network analysis. Furthermore,
because of the unification of words with the same root word, this process can make
analysis easier.

Anyone can re-create this process using existing data and RapidMiner. On the other
hand, adequate hardware is required to install and run this software. The use of a web
interface for pre-processing will be very useful and convenient for librarians and the
general public.

Additional Material
RapidMiner for Academics

Data Availability: https://hdl.handle.net/20.500.12690/RIN/XMCUUX

References
Aria, M., & Cuccurullo, C. (2017). bibliometrix: An R-tool for comprehensive science
mapping analysis. Journal of Informetrics, 11(4), 959–975.
https://doi.org/10.1016/j.joi.2017.08.007

Chapman, P., Clinton, J., Kerber, R., Khabaza, T., Reinartz, T., Shearer, C., & Wirth, R.
(2000). CRISP-DM 1.0: Step-by-step data mining guide. SPSS. https://www.kde.cs.uni-
kassel.de/wp-content/uploads/lehre/ws2012-13/kdd/files/CRISPWP-0800.pdf

CheshmehSohrabi, M., & Mashhadi, A. (2022). Using data mining, text mining, and
bibliometric techniques to the research trends and gaps in the field of language and
linguistics. Journal of Psycholinguistic Research. https://doi.org/10.1007/s10936-022-
09911-6

Gumpenberger, C., Wieland, M., & Gorraiz, J. (2012). Bibliometric practices and
activities at the University of Vienna. Library Management, 33(3), 174–183.
https://doi.org/10.1108/01435121211217199

Han, J., Kang, H.-J., Kim, M., & Kwon, G. H. (2020). Mapping the intellectual structure
of research on surgery with mixed reality: Bibliometric network analysis (2000–2019).
Journal of Biomedical Informatics, 109, 103516. https://doi.org/10.1016/j.jbi.2020.103516

Lamba, M., & Madhusudhan, M. (2018). Application of sentiment analysis in libraries

to provide temporal information service: A case study on various facets of productivity.
Social Network Analysis and Mining, 8(1), 63. https://doi.org/10.1007/s13278-018-0541-y
Li, D., Dai, F.-M., Xu, J.-J., & Jiang, M.-D. (2020). Characterizing hotspots and frontier
landscapes of diabetes-specific distress from 2000 to 2018: A bibliometric study. BioMed
Research International, 2020, 1–13. https://doi.org/10.1155/2020/8691451

Moore, M. T. (2017). Constructing a sentiment analysis model for LibQUAL+

comments. Performance Measurement and Metrics, 18(1), 78–87.
https://doi.org/10.1108/PMM-07-2016-0031

Moral-Muñoz, J. A., Herrera-Viedma, E., Santisteban-Espejo, A., & Cobo, M. J. (2020).

Software tools for conducting bibliometric analysis in science: An up-to-date review. El
Profesional de La Información, 29(1). https://doi.org/10.3145/epi.2020.ene.03

Obidat, A. H. (2022). Bibliometric analysis of global scientific literature on the

accessibility of an integrated e-learning model for students with disabilities.
Contemporary Educational Technology, 14(3), ep374.
https://doi.org/10.30935/cedtech/12064

Porter, M. F. (2001). Snowball: A language for stemming algorithms.

http://snowball.tartarus.org/texts/introduction.html

Schröer, C., Kruse, F., & Gómez, J. M. (2021). A systematic literature review on
applying CRISP-DM process model. Procedia Computer Science, 181, 526–534.
https://doi.org/10.1016/j.procs.2021.01.199

Wang, X., Xu, Z., & Škare, M. (2020). A bibliometric analysis of Economic Research-
Ekonomska Istraživanja (2007–2019). Economic Research-Ekonomska Istraživanja, 33(1),
865–886. https://doi.org/10.1080/1331677X.2020.1737558

Wang, X., Xu, Z., Su, S.-F., & Zhou, W. (2021). A comprehensive bibliometric analysis
of uncertain group decision making from 1980 to 2019. Information Sciences, 547, 328–
353. https://doi.org/10.1016/j.ins.2020.08.036

Recommended Reading

Aria, M., & Cuccurullo, C. (2017). Bibliometrix : An R-tool for comprehensive science
mapping analysis. Journal of Informetrics, 11(4), 959-975.
https://doi.org/10.1016/j.joi.2017.08.007

Aria, M., & Cuccurullo, C. (n.d.). A brief introduction to bibliometrix. Bibliometrix R-

Package. https://www.bibliometrix.org/vignettes/Introduction_to_bibliometrix.html

Kalra, V., & Aggarwal, R. (2017). Importance of text data preprocessing & implementation in
RapidMiner [Paper presentation]. Proceedings of the First International Conference on
Information Technology and Knowledge Management, New Delhi, India.
https://doi.org/10.15439/2017km46

RapidMiner. (n.d.). Text and web mining with RapidMiner. RapidMiner Academy.
https://academy.rapidminer.com/courses/text-and-web-mining-with-rapidminer
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0
International License.

Issues in Science and Technology Librarianship No. 102, Winter 2023. DOI: 10.29173/istl2736