Sentiment Analysis for intelligent ratings management
Khayyam Hashmi, Erfan Najmi, Nariman Ammar, Zaki Malik, Brahim Medjahed · 2014
This paper investigates the problem of rating propagation in composite systems. We propose a method for reputation distribution among component services in Web service composition environments. The main idea lies in providing component services with the appropriate amount of share received for the overall rating. The amount should be proportional to the contribution and performance of the component service. The method ensures that any component service is neither over rated at the expense of a higher performing component nor penalized due to a low performing component. We make use of the textual information present in the service reviews to extract different aspects and their individual sentiments to provide a better rating distribution mechanism. The proposed method attempts to accurately distribute the rating so that it closely reflects the performance of each component in the system. The experimental results show the applicability of our approach and the improved ranking distribution.