Mining consumer reviews to generate ratings of different product attributes while producing feature-based review-summary. Issue 13 (2nd October 2016)

• Record Type: Journal Article
• Authors: Kangale, Akshay; Kumar, S. Krishna; Naeem, Mohd Arshad; Williams, Mark; Tiwari, M. K.
• Is Part Of: International journal of systems science. Volume 47:Issue 13(2016)
• Page Start: 3272
• Page End: 3286
• Publication Date: 2016-10-02
• Subjects: Natural language processing -- crowd sourcing -- feature-based summarisation -- opinion spam -- part-of-speech tagging -- Naïve Bayes -- logistic regression -- classification; System analysis -- Periodicals; 003.3
• DOI: 10.1080/00207721.2015.1116640
• Languages: English
• ISSNs: 0020-7721

Abstract

ABSTRACT: With the massive growth of the internet, product reviews increasingly serve as an important source of information for customers to make choices online. Customers depend on these reviews to understand users' experience, and manufacturers rely on this user-generated content to capture user sentiments about their product. Therefore, it is in the best interest of both customers and manufacturers to have a portal where they can read a complete comprehensive summary of these reviews in minimum time. With this in mind, we arrived at our first objective which is to generate a feature-based review-summary. Our second objective is to develop a predictive model to know the next week's product sales based on numerical review ratings and textual features embedded in the reviews. When it comes to product features, every user has different priorities for different features. To capture this aspect of decision-making, we have designed a new mechanism to generate a numerical rating for every feature of the product individually. The data have been collected from a well-known commercial website for two different products. The validation of the model is carried out using a crowd-sourcing technique.