Title :
Research on Feature Extraction from Chinese Text for Opinion Mining
Author :
Zhu, Shanzong ; Liu, Yuanchao ; Liu, Ming ; Tian, Peiliang
Author_Institution :
Sch. of Comput. Sci. & Technol., Harbin Inst. of Technol., Harbin, China
Abstract :
More and more users and manufacturers concern about product reviews on the web, but it´s difficult to quickly find interesting content from massive information. In order to mine sentiment polarity from review sentences, two approaches for product feature extraction and sentence opinion mining are proposed in this paper. Because of the characteristics of Chinese language, lexical analyzing tools are used to process review text, and association rule model is used to mine frequent items as candidate feature. In order to get better result, several filtering algorithms are proposed. Experiment results demonstrate that relation between the precision and recall rate of feature extraction task with different minimum support thresholds in association rules mining, and the promising performance of our approach has also been shown.
Keywords :
data mining; feature extraction; information filtering; natural language processing; text analysis; Chinese text; association rules mining; feature extraction task; filtering algorithms; sentence opinion mining; Association rules; Computer aided manufacturing; Computer science; Computer vision; Data mining; Feature extraction; Filtering algorithms; Manufacturing processes; Natural languages; Safety; Association rule model; Feature Extraction; Opinion Mining; Sentiment detection;
Conference_Titel :
Asian Language Processing, 2009. IALP '09. International Conference on
Conference_Location :
Singapore
Print_ISBN :
978-0-7695-3904-1
DOI :
10.1109/IALP.2009.11
