A Semantically-based Lattice Approach for Assessing Patterns in Text Mining Tasks

Abstract

In this paper, a new approach to automatically assessing patterns in text mining is proposed. It combines corpus based semantics and Formal Concept Analysis in order to deal with semantic and structural properties for concepts discovered in tasks such as generation of association rules. Experiments show the promise of our evaluation method to effectively assess discovered patterns when compared with other state-of-the-art evaluation methods.