Bibliometric Network Analysis with Natural Language Processing and Graph Clustering

Universität Ulm

MA Abschlussvortrag, Lucas Gmünder, Ort: Online, Datum: 16.02.2022, Zeit: 12:15 Uhr

Scientific publications serve as knowledge carriers and as a medium for conveying information. This wealth of knowledge is growing faster and faster and can be used to gain new insights in many ways. By analyzing the connections between publications and other bibliometric entities, bibliometric networks can identify emerging research topics and valuable knowledge bases for specific topics. Likewise, these analyses provide insight into subtopics and the evolution of topics over time. Given today's prevalent data volumes, such analyses are difficult to perform without tool support. However, tool support is scarce, does not allow for much customization, and falls short in terms of lexical and temporal analysis.

In this work, a system was developed that allows the generation and analysis of bibliometric networks of different types using a comprehensive API. Particular emphasis was placed on including textual content of the bibliometric entities via natural language processing and the dynamic generation and analysis of time periods in the data sets. Also, particular focus was put on the analyses' parameterization to cover as many use cases as possible. The system is scalable and can easily be integrated into larger bibliometric architectures, and it currently forms one of the cornerstones on which the AMBALYTICS bibliometric analysis platform works.