A keyword-aware recommender system using implicit feedback on Hadoop

Recommendation mechanisms offer an effective way to utilize user observations and ratings. The cold-start problem is most prevalent to affect the accuracy and reasoning of recommendation tasks, since new users or items have no side information in recommendation domains. A common architecture of keyword-aware recommender systems is proposed to improve the problem of cold-start users using extra information from external user or item domains. Keywords in a textual description of the extra information are significant factors in estimating user rating of items. Gathering the extra information through WEB scraping methods advances the dimensionality of keyword spaces. To implement the proposed architecture, textual descriptions for the users, who have installed APPs, and the movie-related items are expressed as extra information of the user and item domains. The keyword datasets of users and items drive the estimation of user initial ratings, and make further movie's recommendations to cold-start users. A Big Data environment is essential to perform rating estimation with extensive datasets to establish a high-quality recommendation model. Experimental results obtained from the analysis of keyword similarity and rating accuracy represent that the proposed architecture of keyword-aware recommenders is effective and promising, as well offer profitable recommendation services on improving the recommendation accuracy.