Adding Lexical Chain to Keyphrase Extraction

Abstract

Keyphrase extraction is widely used in information retrieval, automatic summarizing, text clustering, etc. KEA is a traditional and classical algorithm. But it mainly uses the statistical information and ignores the semantic information. In our paper, we propose a method which combine semantic information with KEA by constructing lexical chain that based on Reget's thesaurus. In this method, we use the semantic similarity between terms to construct lexical chain, and then the length of the chain will be used as a feature to build the extraction model. The experiment results attest that the performance of our system has an obvious improvement compare with the KEA and Nguyen and Kan's method.

Keyphrase Extraction of News Web Pages

“…In their method each step uses position-weighted TF*PDF schema. Li, Z. et al [19] propose a method based on the lexical chain to improve KEA keyphrase extraction, their experiments result shows improvements compare with KEA and Nguyen and Kan's method. Hsu, H. M. et al [20] propose subject-keyphrase concept to extract subject-keyphrases from a documents.…”