College of Information Engineering, Capital Normal University, Beijing, China

---

Abstract—In sentiment analysis of product reviews, one important problem is to extract people's opinions based on product features. Through the summary of feature-level opinions, different consumers can choose their favorite products according to the features that they care about. At the same time, manufacturers can also improve the product features based on the opinions. Different words may be used to express the same product feature. In order to form a useful summary, the feature words need to be clustered into different groups based on the similarity. By analyzing the characteristics of Chinese product reviews on the Internet, a novel method based on feature clustering algorithm is proposed to deal with the feature-level opinion mining problems. Particularly, 1) features considered in this paper include not only the explicit features but also the implicit features. 2) opinion words are divided into two categories, vague opinions and clear opinions, to deal with the task. Feature clustering depends on three aspects: the corresponding opinion words, the similarities of the features in text and the structures of the features in comment. Moreover, the context information is used to enhance the clustering in the procedure. Experimental evaluation shows the outperformance of the proposed method.

Index Terms—feature-level, implicit features, opinion mining.

[PDF]

Cite: Lizhen Liu, Zhixin Lv, Hanshi Wang, "Extract Product Features in Chinese Web for Opinion Mining," Journal of Software vol. 8, no. 3, pp. 627-632, 2013.