Chinese Evaluative Information Analysis

Proceedings of COLING 2012: Technical Papers, pages 2773–2788, COLING 2012, Mumbai, December 2012. Chinese Evaluative Information Analysis Y iou Wang Jun′ichi Kazama Takuya Kawada Kentaro Torisawa National Institute of Information and Communications Technology (NICT) 3-5 Hikaridai, Seika-cho, Soraku-gun, Kyoto 619-0289, Japan {wangyiou, kazama, tkawada, torisawa}@nict.go.jp Abstract Together with the ever-growing amount of Chinese web data, the number of opinions voiced by Chinese users is rapidly increasing, and analyzing them is an important task. This paper intro- duces a Chinese Evaluative Information Analyzer (CEIA) and proposes a method to improve its performance. We use evaluative information as a unifying term for the information about attitudes, opinions, sentiments and so on. This paper makes three contributions: (i) CEIA can identify and analyze a more diverse and richer set of evaluative information than previous studies for Chinese; (ii) to implement the system, we constructed an original annotated corpus for Chinese evaluative information and built a large sentiment dictionary; (iii) we introduce syntactic dependency, seman- tic class and distance features to improve the evaluative information extraction. The performance of the system and the effectiveness of the newly introduced features are evaluated in a series of experiments on our Chinese evaluative information corpus. Title and Abstract in Chinese � � �� � �� � �� � �� � �� � �� � �� � � ������������������������������������ ��������������������������������������� ���������CEIA��������������������������� �������������������������������������� �����i���������CEIA���������������������� ���ii�����CEIA���������������������������� �����iii��������������������������������� ��������������������������������������� ������� Keywords: Chinese Evaluative Information, Opinion Mining, Sentiment Analysis. Keywords in Chinese: ����������������. 2773 1 Introduction To automatically find or track the attitudes, feelings and evaluations in texts, opinion mining and sentiment analysis have been extensively studied from different perspectives (Pang and Lee, 2008). With the ever-growing number of Chinese users (over half a billion users only in mainland China), the amount of web opinions in Chinese is rapidly increasing, and analyzing them is an important task. However, research and resources about the Chinese opinion analysis lag behind those for extensively studied languages, such as English. Therefore, opinion analyzers, which can deal with Chinese web data of a great variety of topics and styles, are especially in great need. To meet this requirement, we introduce a Chinese Evaluative Information Analyzer (CEIA) that can mine a wide variety of evaluative information from Chinese web documents. We use evaluative information as a unifying term for the information concerning attitudes, opinions and sentiments, and so on, which is useful to provide a view of evaluation. The system automatically analyzes Chinese evaluative information through the following processes: (1) extracts evaluative expressions; (2) identifies evaluation holders; (3) extracts evaluation tar- gets; (4) determinates evaluation types; (5) determinates the sentiment polarities of the evaluative expressions. CEIA has the following two characteristics: Firstly, CEIA can analyze a more diverse and richer set of evaluative information than the previous studies for Chinese. The previous research on Chinese opinion analysis focuses on subjective expressions (opinionated sentences) (Liu, 2010), as in the Multilingual Opinion Analysis Task (MOAT) of NTCIR (Seki et al., 2010). However, some objective expressions that describe positive or negative …

Content truncated to 500 words