Anita Valmarska (2014) ANALYSIS OF CITATION NETWORKS. EngD thesis.
Abstract
In this thesis we explore the problem of detection of research subdisciplines of a chosen science, based only on the data about citing papers from a chosen batch of papers relevant to the selected science. We directed our attention to the field of Psychology. It is an interesting scientific discipline, with variety of research topics and numerous scientific publications throughout the years. Due to lack of freely accessible centralized database of psychological papers and their relevant citations, the first step of our thesis was collection of papers and their applicable citations. Data was presented in form of a citation network. Citation networks are acyclic directed information networks, where the structure of the network reflects the structure of the information stored in the network vertices. The process of differentiation of research disciplines in the network was performed by applying a state-of-the-art algorithm for community detection. Our choice was the Louvain method, known for its simplicity, effectiveness and speed. Part of the mechanism for rating the detected communities was to name them and examine their connections. Due to the vast quantity of available data and the unfamiliarity with the psychological field, we named the communities based on the measures for cosine similarity between our initially collected psychological papers and the relevant texts for each of the APA divisions of Psychology. Results obtained by the network analysis and the method for community detection are positive. However, the nature of data collection and the influence of our subjective judgment for community naming offer lots of opportunities for further improvement.
Actions (login required)
![[img]](http://eprints.fri.uni-lj.si/style/images/fileicons/application_pdf.png)
