PAR-COM-TEX: methodology for the extraction of relationships among terms of textual documents through association rules and clustering

Abstract

One of the problems related to the use of association rules refers to the number of patterns that are generated, which complicates the interpretation of the obtained rules. In order to overcome this problem, it was proposed in project FAPESP 2010/07879-0 the PAR-COM methodology. The methodology presents good results when applied to structured data. However, as a huge part of the information available today is textual, it is essential to develop techniques to manage such data.In this context, this project aims to develop the PAR-COM-TEX methodology, an adaptation of PAR-COM to the text mining context. PAR-COM-TEX will enable the extraction of relationships among terms of textual documents and their subsequent organization to support the users' decisions. PAR-COM-TEX may be used for various purposes such as to contribute to query expansion approaches in the context of information retrieval (IR). Since PAR-COM-TEX will be integrated into a major project, which targets to contribute to IR processes proposing "enriched" representations, this project also aims to integrate PAR-COM-TEX into an IR system, in order to facilitate, later, the evaluation of its possible applications in this context. (AU)