Applying OWA Operator in the Semantic Processing for Automatic Keyphrase Extraction

Abstract

The automatic keyphrases extraction from texts is a useful task for many computational systems in the natural language processing and text mining fields. Although several solutions to this problem have been developed, the semantic analysis has been one of the linguistic features less exploited in the most reported proposal, causing that the obtained results still show low accuracy and performance rates. This paper presents an unsupervised method for keyphrase extraction, which is based on the use of lexical-syntactic patterns for extracting information from texts and a fuzzy modelling of topics. An OWA operator which combines several semantics measures has been applied in the topic modelling process. This new approach was evaluated with Inspec and 500N-KPCrowd datasets and compared with other reported systems, obtaining promising results.