Text Data Network Analysis Using Graph Approach

Xavier Polanco (1) , Eric San Juan (2)

1 INIST - Institut de l'information scientifique et technique
2 LITA - Laboratoire d'Informatique Théorique et Appliquée

In this paper we revisit this main idea of co-word analysis based on the computation of all geodesic paths, and it is considered that the variants of single link clustering (SLC) are better suited to extract interesting clusters formed along easily interpretable paths of associated items than algorithms based on detecting high density regions. Moreover, we propose a methodology that involves the extraction of graphs of similarities from the text-data represented on the form of a hypergraph. The mining of informative short paths in these graphs is based on a three-step graph reduction process, and the analysis of these graphs use the degree and betweenness centralities. We conclude with an application for testing this methodology.

Mots clés: Text mining; Clustering analysis; Co-word analysis; Graph theory; Hypergraph; Degree centrality; Betweenness centrality

Domaines: Web

Xavier Polanco, Eric San Juan. Text Data Network Analysis Using Graph Approach. International Conference on Multidisciplinary Information Sciences and Technology (InSciT2006), Oct 2006, Mérida, Spain. pp.586-592. ⟨hal-00165964⟩