Sentiment Classification Algorithm Research Articles

Fine-Grained Sentiment Classification based on HowNet

With the rapid development of Web 2.0, sentiment classification -the task of automatically classifying reviews as positive or negative has received widely attention in the past few years. However, most existing researches focus on mining the polarity of a text as a whole. In this paper, a HowNet based fine-grained sentiment classification algorithm is proposed, which address the problem at sentence level. First, HowNet tool is employed to extract the topic of each sentence by computing the similarity to a set of reference concepts. Second, supervised and unsupervised sentiment analysis methods is applied to determine its orientation. Finally, fine-grained orientation of the review set will be identified by sum up the sentiment of different topis aspect. Experiments on Chinese real hotel review dataset collected from Ctrip.com shows that the proposed algorithm obtain satisfied classification result.

Incorporating Sentiment Prior Knowledge for Weakly Supervised Sentiment Analysis

This article presents two novel approaches for incorporating sentiment prior knowledge into the topic model for weakly supervised sentiment analysis where sentiment labels are considered as topics. One is by modifying the Dirichlet prior for topic-word distribution (LDA-DP), the other is by augmenting the model objective function through adding terms that express preferences on expectations of sentiment labels of the lexicon words using generalized expectation criteria (LDA-GE). We conducted extensive experiments on English movie review data and multi-domain sentiment dataset as well as Chinese product reviews about mobile phones, digital cameras, MP3 players, and monitors. The results show that while both LDA-DP and LDA-GE perform comparably to existing weakly supervised sentiment classification algorithms, they are much simpler and computationally efficient, rendering them more suitable for online and real-time sentiment classification on the Web. We observed that LDA-GE is more effective than LDA-DP, suggesting that it should be preferred when considering employing the topic model for sentiment analysis. Moreover, both models are able to extract highly domain-salient polarity words from text.