You are on page 1of 5

ACADEMIA Letters

Building AIPedia Ontology to Evaluate Research Impact


in Artificial Intelligence Area
Yasser Maatouk, Computer Science Department, King Abdulaziz University,
Jeddah, Saudi Arabia

Artificial intelligence (AI) is one of the most important research area that has been applied
extensively in several fields. Recently, it is becoming increasingly popular owing to its ability
to solve complex practical problems and integrate different systems [1]. According to the
authors of [2], AI has become a hot topic and is important in various fields. Consequently, AI
research has increased significantly over the last 20 years. The authors expect more research
to be conducted in the AI field in the future. Moreover, based on Web of Science (WoS), there
has been a clear increase in the number of AI publications in the last decade.
The WoS is one of the databases that provides a gateway to search and access the metadata
of scholarly articles in the AI field. The structure and method through which the contents are
offered on the WoS database gives useful information about the articles beside more concern
on the most important traditional measurements like bibliometrics [3]. Bibliometrics is used
to perform statistical analysis to evaluate the research impact and determine the importance
of specific scientific production. It can show the temporal evolution of research on a specific
topic by using citation number as an example. Additionally, it can be applied to evaluate the
scientific outputs of authors, such as h-index indicator or journals in certain fields such as
impact factor (IF) measurement [4][5][6].
There are several limitations in the WoS database regarding the form and the type of the
metadata provided for the scientific publications. First problem is these metadata are only
understandable by humans. Thus, humans can do the keyword-based searches through the
WoS database. Simultaneously, the big data about the scholarly papers is in silence mode. In
other words, it is neither open nor linked to other datasets. Another biggest drawback of the
WoS database is that it does not show the altmetrics indicators of these scientific documents

Academia Letters, August 2021 ©2021 by the author — Open Access — Distributed under CC BY 4.0

Corresponding Author: Yasser Maatouk, ymaatouk@kau.edu.sa


Citation: Maatouk, Y. (2021). Building AIPedia Ontology to Evaluate Research Impact in Artificial
Intelligence Area. Academia Letters, Article 2781. https://doi.org/10.20935/AL2781.

1
widely. Altmetric measurements can be used to capture attention on online platforms and dis-
cover the opinions of laypeople. Further, altmetric measurements are considered as metrics,
qualitative data and can be used as complementary measurements to traditional bibliometrics,
e.g., citation count. Altmetric data sources can include several different online platforms. For
example, exchange of views on research blogs, citations on Wikipedia, mentions on News,
bookmarks on reference managers like Mendeley, and tweets on Twitter. Overall, altmetrics
can include several online sources from the Web which clarify how specific scholarly docu-
ments are discussed and used around the world [7].
Moreover, the altmetric attention score (AAS) and the donut shape have been created to
know the amount and the type of attention has been received by a specific scholar document.
In addition to the bibliometric indicators, the AAS can work as an indicator of the amount of
attention that specific scholarly document has received. AAS represents a weighted count of
the amount of attention that is selected for a research output. The importance of this weighted
count scenario is to reflect the relative reach of all types of altmetric sources. For example,
we can imagine that the average newspaper is weighted more because it brings more attention
to the research output than a simple tweet. [7].
Semantic web (SW) technology is used to overcome the limitations of WoS database.
In this study, we developed AIPedia, which is a structured semantics-based repository that
includes a semantically enriched knowledge base comprising AI publications with more cov-
erage on bibliometrics and altmetrics indicators. Further, we evaluated AIPedia by creating a
list of competency questions. We implemented SPARQL queries for each defined question.
The semantic web (SW) technology can be used to overcome the WoS database limita-
tions. This technology can produce semantically enriched data. Its main purpose is to trans-
late the metadata of existing data in a unified form called resource description framework
(RDF) and web ontology language (OWL). Moreover, the SW technology can develop a new
schema (ontology) with well-defined relationships representing knowledge by using the rea-
soning technology, which can be useful in machine understanding [8].
Another important concept in the SW technology is linked open data (LOD). LOD is
structured data modeled as a graph and published in a manner that allows interlinking between
several datasets. It is used to access data across different and distributed data sources that are
currently scattered, accessible on a restricted scale, and exist in pieces. SPARQL can then
be used to perform sophisticated queries. SPARQL is a query language used to retrieve data
from an SW ontology. The output of such queries can be utilized for analysis studies instead
of conducting manual analysis on thousands of records [9][10][11].
Several studies have been conducted to extract structured information related to scientific
documents to build semantically enriched data from existing sources. For example, DBLP,

Academia Letters, August 2021 ©2021 by the author — Open Access — Distributed under CC BY 4.0

Corresponding Author: Yasser Maatouk, ymaatouk@kau.edu.sa


Citation: Maatouk, Y. (2021). Building AIPedia Ontology to Evaluate Research Impact in Artificial
Intelligence Area. Academia Letters, Article 2781. https://doi.org/10.20935/AL2781.

2
SPedia, CERIF, VIVO, Sapientia, PharmSci, and several other works involve ontologies that
translated the existing data source into a unified RDF format. Consequently, end-users can
query the dataset to extract useful knowledge using SPARQL. However, The major difference
between these works and ours is that our work adds bibliometrics and altmetrics data sources
into one knowledge base dataset.
In this research work, we built a structured semantics-based repository called AIPedia that
includes a semantically enriched knowledge base about publications in the AI field from the
WoS database as a major source. The AIPedia dataset knowledge base provides information
on AI publications. Moreover, AIPedia translated the data into an RDF format that mainly
offers more coverage on bibliometric and altmetric indicators. Furthermore, we utilized the
AIPedia and RDF datasets by allowing users to execute sophisticated SPARQL queries. The
query results will be used to answer questions related to AI publications and conduct various
types of analyses that can be helpful in decision making and for policy makers.
In this context, our research work makes the following contributions.

• Extracting structured information for AI publications available in WoS. This informa-


tion has altmetric data too. All relational data are triplicated mainly between authors,
publications, and journals. The end product contains valuable information about the AI
publications.

• Producing semantically enriched datasets as RDF unified triples from two different
sources (bibliometrics and altmetrics) that were extracted from the detailed information
(e.g., article DOI, author h-index, journal impact factor (JIF), altmetric score, Twitter
score) of AI scientific publications.

• Interlinking RDF datasets by combining bibliometric and altmetric sources. Moreover,


AIPedia could be interlinked with existing external open datasets such as DBPedia,
SPedia, or any other related ontologies in the web of data.

• Querying the final formed datasets using the SPARQL language. This can be achieved
by executing semantically enriched queries on AIPedia, which contains information
about AI publications in the RDF triples format.

• Evaluating the AIPedia to show the availability and reusability of the ontology by cre-
ating a list of seven competency questions that the AIPedia should be able to answer.
The competency questions were designed to consider the end-users of the final product.

• Further, we conducting an analysis based on the answered questions to obtain more


knowledge and useful informatics charts that helpful for decision makers.

Academia Letters, August 2021 ©2021 by the author — Open Access — Distributed under CC BY 4.0

Corresponding Author: Yasser Maatouk, ymaatouk@kau.edu.sa


Citation: Maatouk, Y. (2021). Building AIPedia Ontology to Evaluate Research Impact in Artificial
Intelligence Area. Academia Letters, Article 2781. https://doi.org/10.20935/AL2781.

3
• The end product of AIPedia is considered to be the first effort in calculating indexes
using both bibliometric and altmetric indicators for publications in AI by using semantic
web technology.

• Finally, we believe that our work on AIPedia is superior to that on WoS database in
many aspects. One of the most important features of AIPedia is its ability to infer new
facts from existing ones and perform reasoning based on some created rules, which is
not present in the WoS database.

References
[1] A. Mellitand S. A. Kalogirou, “Artificial intelligence techniques for photovoltaic appli-
cations: A review, ”Progress in Energy and Combustion Science, vol. 34, no. 5, pp.
574–632, 2008.

[2] Y. Lei and Z. Liu, “The development of artificial intelligence: a bibliometric analysis,
2007- 2016,” Published under license by IOP Publishing Ltd, Journal of Physics: Con-
ference Series 1168 022027, vol. 1168, no. 2, 2019.

[3] Web of Science [https://clarivate.com/webofsciencegroup/]

[4] D. H. Milanez, M. Schiavi, R. Amaral, L. de Faria, andJ. A. R. Gregolin, “Development


of carbon-based nanomaterials indicators using the analytical tools and data provided by
the Web of Science database, ”Materials Research, vol. 16, no. 6, pp. 1282–1293, 2013.

[5] E. Archambault, D. Campbell, Y. Gingras, and V. Lariviere, “Comparing of science bib-


liometric statistics obtained from the Web and Scopus, ”Journal of the American Society
for Information Science and Technology, vol. 60, no.7, pp. 1320–1326, 2009.

[6] P. H. Lv, G. F. Wang, Y. Wan, J. Liu, Q. Liu, and F. C. Ma, “Bibliometric trend analysis
on global graphene research, ”Scientometrics, vol. 88, no. 2, pp. 399–419, 2011.

[7] Altmetric [https://www.altmetric.com/]

[8] T. Berners-Lee, J. Hendler, and O. Lassila, “The semantic Web, ”Sci. Amer., vol. 284,
no. 5, pp. 34–43, 2001.

[9] C. Bizer, T. Heath, D. Ayers, and Y. Raimond, “Interlinking open data on the Web,” in
Proc. 4th Eur. Semantic Web Conf. Demonstrations Track, Innsbruck, Austria, 2007, pp.
1–2.

Academia Letters, August 2021 ©2021 by the author — Open Access — Distributed under CC BY 4.0

Corresponding Author: Yasser Maatouk, ymaatouk@kau.edu.sa


Citation: Maatouk, Y. (2021). Building AIPedia Ontology to Evaluate Research Impact in Artificial
Intelligence Area. Academia Letters, Article 2781. https://doi.org/10.20935/AL2781.

4
[10] T. Heath and C. Bizer, “Linked data: Evolving the Web into a global data space, ”in
Synth. Lect. Semantic Web, Theory Technol., vol. 1, no. 1, 2011, pp. 1–136.

[11] C. Bizer, T. Heath, and T. Berners-Lee, “Linked data: The story so far, ”Semantic Ser-
vices, Interoperability and Web Applications: Emerging Concepts. Hershey, PA, USA:
IGI Global, 2011, pp. 205–227.

Academia Letters, August 2021 ©2021 by the author — Open Access — Distributed under CC BY 4.0

Corresponding Author: Yasser Maatouk, ymaatouk@kau.edu.sa


Citation: Maatouk, Y. (2021). Building AIPedia Ontology to Evaluate Research Impact in Artificial
Intelligence Area. Academia Letters, Article 2781. https://doi.org/10.20935/AL2781.

You might also like