Opinion Mining Based Product Usability Modeling And Evaluation

In the recent years, companies always want to attract consumers'eye by adding a variety of features to enhance the capability of the product. Meanwhile, consumers always would like to choose a product with as more feature as possible. However, consumers will often find that some features are difficult to be used and even there are no demands to use some features. These situations will cause a negative evaluation that affect future purchase behavior. The performance of these products in the market can't meet prospect is mainly because excessive features effect the product usability. Therefore, we are committed to research the product usability which is the key factor affect the success in the market. We want to provide decision support for product improvement and redesign by modeling and evaluating product usability.On the basis of the previous results and preliminary studies, some further studies such as raw data about product usability obtaining method, feature usability modeling approach and evaluating standard are focused in the thesis. A novel product usability modeling and evaluation approach based on opinion mining is raised to deal with the product usability issue. The data of our approach is the comments from the network. Firstly, opinion mining was used to process the unstructured product comment to structured comment database. Then we'll choose the feature related to product usability and use factor analysis method to extract common factors which effected product usability. At last, a case study on handset was conducted to demonstrate the usefulness of the proposed methodology.In addition, on the basis of the theoretical studies we introduce in the second paragraph, we design and develop the product usability system which use the web comment as data source, establish usability evaluation model from user's perspective and provide reference to usability researchers. We use the hierarchy table to store data and solve the problem that network reviews are lack of standardization. The system not only can quickly collect the raw data with high versatility and also can provide decision support for users and researchers.