Online review data contains useful information about consumers’ opinions about the reviewed products. Understanding reviews is extremely important for a producer, such as a corporation designing and producing cell phone, to predict the market and make future strategies. However, due to the large volume of the review data, it is difficult to quickly digest the reviewers’ the preferences, or to further recognize interesting patterns of reviewers’ opinions. To address this need, in this thesis, we propose a novel analysis system based on Latent Aspect Rating Analysis Model. It not only understand the feedbacks or reviews from the costumers, but also analyze the market and trend behind the data.As for the user’s basic interface, we define some operators for the analysis system and the user can conduct our predefined queries with the operators over the data to analyze the products. Futhermore, we allow the users to do more complex jobs over the system by designing a pipeline syntax of the operators. Besides, the system can visualize the results, so that the users can better understand the reviews.Overall, the proposed analysis system is general and can support analysis of any review data. It thus enables many interesting applications such as product analysis, market prediction and business intelligence.
【 预 览 】
附件列表
Files
Size
Format
View
A general tool for interactive analysis of opinions in reviews