Personality Based Recommendation System Using Social Media

Abstract

Recommendation system is the reason of success for most of the social media companies as well as e-commerce sites. Giving recommendation to the uses is one of the interesting and challenging tasks nowadays, it helps to generate revenue, to increase number of users, to reduce the searching time for particular item. Recommendation system helps for making interest in user and eventually it increases the popularity of any site. Huge number of items (product, users, movies, songs, hotels etc.) and its feature sets makes it hard to predict the accurate items to the user. It is important to keep all historic data of user as well as all information about the items to generate recommendation. In this paper, the personality of the user is used with the combination on the most popular recommendation techniques like collaborative filtering (CF) and content based filtering (CB) proposed on the amazon review data set. In the first model the personality of the user is calculated by using the big five model on the twitter account. In the second module Collaborative filtering is used to generate the recommendation based on the historic information of the user wherries in third module, Content based filtering is uses to generate recommendation based on the feature set of the item. Pearson-correlation algorithm is applied on both modules and ranking are generated. Finally union of the both vector space are taken as the final recommendation.