Abstract:

With the rapid development of new network media, the quantity of online reviews has a tendency of explosive growth. Traditional manual methods for opinion mining have some problems when dealing with tremendous online texts, such as low efficiency, fuzzy classification boundary, and limited domainadaption ability. In order to solve the above problems, we improve the traditional latent Dirichlet allocation (LDA) model, and propose a LDA topic model based on domain identification for opinion mining of online reviews. Firstly, a domain layer is added to the standard LDA model to sample the domain tags of the document, and field parameters are utilized to solve the LDA model. Secondly, given the sentimental connection between sentences, we insert a sentiment layer between the topic layer and word layer. Sentimental transition variable is introduced to denote related characters, which can increase the accuracy of sentiment polarity analysis. Experimental results verify the validity of the proposed model and theory.

Key words: LDA model, domain identification, opinion mining, sentimental transition variable