The focus of our work is to apply association rule mining to collaborative recommender systems, which recommend articles to a user on the basis of other users' ratings for these articles as well as the similarities between this user's and other users' tastes. In this work, we propose a new algorithm for association rule mining specially tailored for use in collaborative recommendation. We make recommendations by exploring associations between users, associations between articles, and a combination of the two. We experimentally evaluated our approach on real data for many different parameter settings and compared its performance with that of other approaches under similar experimental conditions. Through our analysis and experiments, we have found that association rules are quite appropriate for collaborative recommendation domains and that they can achieve a performance that is comparable to current state of the art in recommender systems research.
Our research focuses on extending a system which mines association rules for collaborative filtering ( Lin, Alvarez, and Ruiz) by enhancing the framework to utilize available content information in formulating predictions. Association rule mining is well-suited for recommendation, because association rules can naturally find and describe multiple dependencies between those aspects which determine user preferences, such as article properties or other users' ratings of article properties. Additionally, mined rules come quantified by two measures, the confidence and the support, which measure the strength and the significance of the dependencies among factors, respectively.
To enhance recommendation, we propose mining two new types of associations: content associations, which relate a target user to the properties of articles they find of interest, and content-based collaborative associations, which attempt to find relationships between users through users' ratings of article properties. We test these approaches using the domain of movies. Our results show that we achieve high quality recommendations and in most cases with real time performance. Specifically, our results show that overall, content-based collaborative filtering yields the highest precision, but may fail to produce recommendations for some users. We therefore combine our content-based collaborative filtering with other system modes, including regular collaborative associations, and content associations, which have slightly lower precision but yield recommendations for all users.
We designed and built a web-based movie recommender system. We used association rule mining to implement two data filtering methods. Content-based filtering identifies sets of common attributes of the movies that the user has liked in the past, while collaborative filtering associates users with each other based on similarities in taste. By combining content- and collaborative-based filtering, we obtained recommendations with a higher precision than either method individually.
In this project we apply artificial neural networks trained using error back-propagation to construct three different systems for automated information filtering. These systems utilize content-based filtering, collaborative filtering, and a combination of both methods. Extensive experimental evaluation of the systems has been carried out with movie recommendation as a test domain, using data collected from our own on-line survey.
In this project we investigate the use of artificial neural networks (ANNs) as the core prediction function of a recommender system. In the past, research concerned with recommender systems that use ANNs have mainly concentrated on using collaborative-based information. We look at the effects of adding content-based information and how altering the topology of the network itself affects the accuracy of the recommendations generated. In particular, we investigate a mixture of experts topology. We create two expert clusters in the hidden layer of the ANN, one for content-based data and another for collaborative-based data. This greatly reduces the number of connections between the input and hidden layers. Our experimental evaluation shows that this new architecture produces the same accuracy of recommendation as the fully connected configuration with a large decrease in the amount of time it takes to train the network. This decrease in time is a great advantage because of the need for recommender systems to provide real time results to the user.
This project investigated a mixture of experts neural network architecture for a combined collaborative and content-based recommender system. The effect of first reducing the dimensionality of the input data using the singular value decomposition was also studied. We showed that the mixture of experts architecture achieves the same recommendation quality as a fully-connected architecture while requiring less computation time, or, if desired, higher quality can be achieved with only a slight increase in running time.