BIBLIOMETRIC ANALYSIS OF SCIENTIFIC PUBLICATIONS DEVOTED TO SUSTAINABLE DEVELOPMENT GOAL 13 USING NLP TECHNIQUES

Abstract

Improvements in search engines for traditional text documents have made it possible to retrieve massive text information with efficient use of resources efficiently. As the number of published scientific articles on climate change increases, analyzing trends and the current state of this field of science becomes a very time-consuming and time-consuming task. To address the ever-growing demands for systematic literature review techniques, rapid review and scoping reviews techniques have emerged. This work carried out statistical analysis of publications and popular NLP methods such as vectorization, metadata preprocessing, bag-of-words, TF-IDF, lemmatization, and others. We also applied the LDA topic modelling to identify topics from abstracts of scientific publications. According to the statistical analysis outcomes of a large dataset from the CORE API, we obtained the following results: (1) there was a significant increase in research since 2010 and peaked in 2016 with 865980 publications on the climate change field of study; (2) the two most productive institutions are the University of Pennsylvania and the Department of Energy, the United States for Climate Impact Research; (3) in almost all five centrality measures, Intergovernmental Panel on Climate Change is leading as the most productive author; (4) an obvious fact is the high level of involvement in climate change research of Western countries, while according to the results of the analysis, the second place was also taken by the representative of Asia - Japan. Additionally, we created citations and co-authorship networks using the Neo4j graph database and calculated five centrality measures, including Page Rank, Eigenvector centralities, and others.