PROJECT TITLE :

Complementary Aspect-Based Opinion Mining - 2018

ABSTRACT:

Side-primarily based opinion mining is finding elaborate opinions towards a topic like a product or an incident. With explosive growth of opinionated texts on the Web, mining facet-level opinions has become a promising means that for online public opinion analysis. In specific, the boom of numerous types of on-line media provides numerous nevertheless complementary data, bringing unprecedented opportunities for cross media side-opinion mining. Along this line, we tend to propose CAMEL, a completely unique topic model for complementary side-based mostly opinion mining across asymmetric collections. CAMEL gains data complementarity by modeling both common and specific aspects across collections, while keeping all the corresponding opinions for contrastive study. An auto-labeling theme known as AME is additionally proposed to assist discriminate between side and opinion words without elaborative human labeling, which is further enhanced by adding word embedding-primarily based similarity as a replacement feature. Moreover, CAMEL-DP, a nonparametric alternative to CAMEL is additionally proposed based mostly on coupled Dirichlet Processes. Intensive experiments on real-world multi-assortment reviews information demonstrate the prevalence of our strategies to competitive baselines. This is notably true when the data shared by completely different collections becomes seriously fragmented. Finally, a case study on the general public event “2014 Shanghai Stampede” demonstrates the practical worth of CAMEL for real-world applications.