Applying MCDM Techniques for Ranking Products Based on Online Customer Feedback

Abstract

Text analytics is to distill out structured information from unstructured or semi-structured text. User feedback analysis or sentiment analysis on products enables to highlight the best and worst of features and recommend the product to new buyers. The model extracts the positive and negative comments and identifies the emotions in the piece of text or n-way analysis and classification like very-positive, positive, neutral, negative or very-negative. Natural Language Processing (NLP) tools play vital role in classifying the sentiment polarity of sentences while data analytics has the role in recommendation of the product. In this paper, we propose a recommender system model to rank the products based on the feedback given by the users. Features, the topics of interest, are identified from the set of review text. Sentiments are detected from each review and thus senti-score is calculated for each feature of the product. We use the Analytic Hierarchy Process and the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS), which are Multi-Criteria Decision Making techniques to rank a set of products. This method provides a logical framework to determine the benefits of each product based on the features and thus the products are ranked.

Keywords: Sentiment Analysis, User Feedback Analysis, Multi-Criteria Decision Making, Technique for Order Preference by Similarity to Ideal Solution, Analytic Hierarchy Process

View PDF