The Construction And Application Of Chinese Emotion Word Ontology

Affective computing has received more and more interests in the field of artificial intelligence, and its goals are that the computer holds emotions. Just like human beings, the computer could communicate amiably and naturally. With the development of Internet, textual information becomes the richest interactive resources; however, few researches have focused on affective text analysis.So we constructed the Chinese affective lexicon ontology firstly. The paper analyzed the status of the emotional classification, and then classification system was determined. Finally, affective lexicon ontology which synthesizes various resources was constructed. In the process of acquiring the knowledge, the framework of ontology was filled by the combination of manual classification and acquiring the intensity automatically. The paper also describes emotional classification and lexical intensity etc, and the distribution of affective lexicons.In order to reduce the manual labor, we proposed a method of automatic emotion vocabulary acquisition based on CRF. In the experiment, we used some rules, such as the words' feature rules, the words' context rules, and, their tie-in pairs. At last, we found the best rules, analyzed the result strictly, and found the main reasons of some mistakes.In the processing of multi-affective words construction, we used the semi-automatic ways. In the automatic parts, we selected the multi-sense words by thesaurus, proposed the hypothesis that most of the multi-affective words are multi-sense words, and also used some information in the affective lexicon ontology. In the manual part, we explain the differences between the multi-affective words and those multi-affective words which needed be disambiguation in certain context. We think the differences are useful for the further word affective disambiguation.In the word affective disambiguation part, we analyzed the difference and resemblance between word affective disambiguation and word sense disambiguation. We choose Bayesian model to realize the word affective disambiguation. And, three other experiments: word affect disambiguation based on common emotion, word affect disambiguation based on part of speech, and, word affect disambiguation based on part of speech and emotional frequencies. The results show word affect disambiguation based on Bayesian model is the most effective.