Aspect Based Mobile Recommendation System

Abstract

Now-a-days, there is a trend of changing mobile phone models in very short duration. To achieve benefit of choosing a mobile phone satisfying our requirements, mobile phone recommendation system is of great importance. As we know, there are many web sites that provide number of reviews and ratings for each and every mobile phone model available in market. From this, we can understand the consumer opinions and reviews about any mobile product. Existing systems were based on complete review of product as good or bad. But, there is need to have a way with which we can review the product with view of each aspect such as camera, battery, look etc. For this, we have developed a system which provides a recommendation of mobile phone model by considering aspects with user’s choice. For this, we have used reviews, ratings provided by group of users on social website, Amazon. The data is collected with the use of “Octoparse”, a web scraping software and the text data collected is analyzed using Stanford's CoreNLP for sentiment analysis. Our approach provides recommendation, considering user provided aspects (i.e. camera, battery, look etc.) with the use of apache mahout and hybrid recommendation. Our approach showed outstanding performance for mobile phone recommendation.