A Novel Feature Cloud Visualization for Depiction of Product Features Extracted from Customer Reviews

There has been an exponential growth of web content on the World Wide Web and online users contributing to majority of the unstructured data which also contain a good amount ofinformation on many different subjects that may range from products, news, programmes and services. Many a times otherusers read these reviews and try to find the meaning of the sentences expressed by the reviewers. Since the number and the length of the reviews are so large that most the times the user will read a few reviews and would like to take an informed decision on the subject that is being talked about. Many different methodshave been adopted by websites like numerical rating, star rating, percentage rating etc. However, these methods fail to giveinformation on the explicit features of the product and their overall weight when taking the product in totality. In this paper, aframework has been presented which first calculates the weightof the features depending on the user satisfaction or dissatisfaction expressed on individual features and further afeature cloud visualization has been proposed which uses two level of specificity where the first level lists the extracted features and the second level shows the opinions on those features. A font generation function has been applied which calculates the font size depending on the importance of the features vis-a-vis with the opinion expressed on them.

Keywords:

Opinion Mining , Natural Language Processing , Feature Cloud , Visualization