The advancement of the Internet has brought us into a world that represents a huge amount of information items such as movies, web pages, etc. with fluctuating quality. As a result of this massive world of items, people get confused and the question “Which one should I select?” arises in their minds. Recommendation Systems address the problem of getting confused about items to choose, and filter a specific type of information with a specific information filtering technique that attempts to present information items that are likely of interest to the user. A variety of information filtering techniques have been proposed for performing recommendations, including contentbased and collaborative techniques which are the most commonly used approaches in recommendation systems. This dissertation introduces a new recommendation model, a feature weighting technique to cluster the user for recommendation top-n movies to avoid new user cold start and scalability problem. The distinctive point of this study lies in the methodology used to cluster the user and the methodology which is utilized to recommend movies to new users. The model makes it possible for the new users to define a weight for every feature of movie based on its importance to the new user in scale of one (with an increment of 0.1). By using these weights, it finds nearest cluster of users to the new user and suggests him the top-n movies (with the highest rate and most frequency) which are reviewed by users that are in the targeted cluster. Rating and Movie dataset were are used during this study. Firstly, purity and entropy are applied to evaluate the clusters and then precision, recall and F1 metrics are used to assess the recommendation system. Eventually, the results of accuracy testing of proposed model are compared with two traditional models (OPENMORE and Movie Magician Hybrid) and based on the evaluation the level of preciseness of the proposed model is more better than Movie Magician Hybrid but worse than OPENMORE.
展开▼
