Construction And Application Of A Chinese Emotion Lexicon

With the rapid development of Internet technology, more and more users have an equal opportunity to obtain information on the Internet. But the amount of online information is too huge, how to quickly identify the expressed emotional information on characters, events, products, etc., and to obtain the desired valuable information have become hot research topic in natural language processing and emotional analysis field. It also has been found wide applications in many areas, such as event analysis, product review, movie review, hotel review and public opinion analysis.Text emotional analysis is mainly to identify the emotional tendency expressed in text, and emotional words are the main basis for judging emotional tendency. Designing an efficient algorithm to build emotional lexicon is a basic and important work for text emotional analysis. Emotional lexicon is a collection of emotional words, and is constructed by two steps:collecting emotional words and labeling words’emotional information. Currently the main methods of emotional lexicon construction are semantic dictionary-based method and corpus-based method. Most of these methods are to determine the words in a given word list or words in a dictionary, thus resulting in a limited number of words in emotional lexicon, and cannot initiate to identify the emotional words in corpus and analyze words’emotional information.We propose a Chinese emotional lexicon construction method, which extract emotional words based on syntactic analysis from the corpus, and classify emotional words according to some given emotional words and classification technology. Firstly, through syntactic analysis, sentences are parsed to extract candidate emotional words. Then by making use of the words co-occurrence in the corpus, we get the similarity between words. Finally Support Vector Machine is adopted to classify the candidate emotional words, and we apply our method on unlabeled corpus to get emotional words.In addition, we apply the emotional lexicon on text emotional analysis through the combination of feature extraction methods and emotional lexicon to express text. First, CHI method is applied for feature extraction by comparative experiments. Then emotional words are added to the feature sets. The experimental results show that the proposed method can achieve a precision up to 81.63%, and the average precision is 76.95%, which are better than the results obtained by CHI method.