REAL TIME CRAWLING AND MINING ONLINE E-COMMERCE REVIEWS WITH WAM

Abstract

The mining opinion targets and opinion words from online reviews are important tasks for fine-grained opinion mining, the key component of which detecting opinion relations among the words. In this his paper proposes a novel approach based on the partiallysupervised alignment model, which regards identifying opinion relations as an alignment process. A graph-based co-ranking algorithm is exploited to estimate the confidence of each candidate. Finally, the candidates with higher confidence are extracted as opinion words or opinion targets. Compared to the previous methods based on the nearest-neighbor rules, our model is captures opinion relations more precisely, especially for the long-span relations. Compared to syntax-based methods, our word alignment model effectively alleviates negative effects of parsing errors when dealing with informal online texts. Compared to the traditional alignment model, the proposed model obtains better precision because of the usage of partial supervision. In addition to, when estimating the candidate confidence, we penalize higher-degree vertices in our graph-based co-ranking algorithm to decreases the probability of error generation. Our experimental results on three corpora with different sizes and languages show that our approach effectively outperforms state-of-the-art methods.