Analysis of Word Co-occurrence Networks from Paper Abstracts in Semantic Scholar Database
Conference
65th ISI World Statistics Congress 2025
Format: IPS Abstract - WSC 2025
Keywords: statistical_network_analysis
Session: IPS 830 - Recent Advances in Large-Scale Network Data Analysis and Their Applications
Wednesday 8 October 2 p.m. - 3:40 p.m. (Europe/Amsterdam)
Abstract
The abstract is a crucial frontmatter element that provides readers with key insights into a manuscript's core ideas and subject categories. Identifying the most important words in abstracts can offer valuable clues about the central themes and evolving trends within a particular subject area. This work introduces a novel analysis method to determine the importance of words within a subject category over time, based on various centrality measures in a word co-occurrence network. The network is constructed from words extracted from the abstracts of manuscripts within a specific scientific subject. We demonstrate the effectiveness of this method using a subset of the Semantic Scholar database, focusing on the field of Statistics from 2019 to 2023.