Acquisition and clustering for affective semantic lexicon from web

Abstract

A simple semantic lexicon extraction method is proposed based on one hypothesis and three filtering rules from Baidu Chinese Network Encyclopedia. The acquired affective lexicon includes emotional words and their lexical semantic relations including synonyms and antonyms. The acquiring method is recursive algorithm using the seed words. The extracted affective lexicon is labeled with affective tendency automatically based on the semantic relations and existing affective dictionaries. We proposed method effectively constructs and clusters the affective lexical semantics for Chinese. The method improves the word number for affective lexicon, and gets automatically and effectively tendency tagging. The experimental results show that the acquired affective semantic lexicon is extended by 166.6% emotional annotation compared to the existing dictionaries, and took 78.1% accurately tendency tagging based on the existing dictionaries.