Mining Conceptual Relations from Textual Web Content Using Leximancer

Abstract

Concept mining is a process that focuses on extracting ideas and concepts found in documents. The approach is somewhat similar to text mining, with the main difference being that mining a text focuses on the extraction of information rather than ideas. In this paper, we propose concept-based text representation, with an accent on using the proposed representation in different applications such as information retrieval, text summarization, and question answering. This work presents a new prototype for concept mining by extracting the concept-based information from a raw text using leximancer. At the text representation level, we introduce a sentence based conceptual ontological representation that builds concept-based representations for the whole document. A new concept-based similarity measure is proposed to measure the similarity of texts based on their meaning. The proposed approach is domain independent and it could be applied to general domain applications. The proposed approach is going to apply to the domain of information retrieval, and give an assertion for proceeding in the right directions of this research.