Li-Tung Weng

Improving Recommendation Novelty Based on Topic Taxonomy

Clustering has been a widely applied approach to improve the computation efficiency of collaborative filtering based recommendation systems. Many techniques have been suggested to discover the item-to-item, user-to- user, and item-to-user associations within user clusters. However, there are few systems utilize the cluster based topic-to-topic associations to make recommendations. This paper suggests a taxonomy-based recommender system that utilizes cluster based topic-to-topic associations to improve its recommendation quality and novelty.

Towards Information Enrichment through Recommendation Sharing

Nowadays most existing recommender systems operate in a single organisational basis, i.e. a recommender system recommends items to customers of one organisation based on the organisation’s datasets only. Very often the datasets of a single organisation do not have sufficient resources to be used to generate quality recommendations. Therefore, it would be beneficial if recommender systems of different organisations with similar nature can cooperate together to share their resources and recommendations. In this chapter, we present an Ecommerce-oriented Distributed Recommender System (EDRS) that consists of multiple recommender systems from different organisations. By sharing resources and recommendations with each other, these recommenders in the distributed recommendation system can provide better recommendation service to their users. As for most of the distributed systems, peer selection is often an important aspect. This chapter also presents a recommender selection technique for the proposed EDRS, and it selects and profiles recommenders based on their stability, average performance and selection frequency. Based on our experiments, it is shown that recommenders’ recommendation quality can be effectively improved by adopting the proposed EDRS and the associated peer selection technique.

Improve Recommendation Quality with Item Taxonomic Information

Recommender systems’ performance can be easily affected when there are no sufficient item preferences data provided by previous users. This problem is commonly referred to as cold-start problem. This paper suggests another information source, item taxonomies, in addition to item preferences for assisting recommendation making. Item taxonomic information has been popularly applied in diverse ecommerce domains for product or content classification, and therefore can be easily obtained and adapted by recommender systems. In this paper, we investigate the implicit relations between users’ item preferences and taxonomic preferences, suggest and verify using information gain that users who share similar item preferences may also share similar taxonomic preferences. Under this assumption, a novel recommendation technique is proposed that combines the users’ item preferences and the additional taxonomic preferences together to make better quality recommendations as well as alleviate the cold-start problem.

An Efficient Neighbourhood Estimation Technique for Making Recommendations

Recommender systems produce personalized product recommendations during a live customer interaction, and they have achieved widespread success in e-commerce nowadays. For many recommender systems, especially the collaborative filtering based ones, neighbourhood formation is an essential algorithm component. Because in order for collaborative-filtering based recommender to make a recommendation, it is required to form a set of users sharing similar interests to the target user. Forming neighbourhood by going through all neighbours in the dataset is not desirable for large datasets containing million items and users. In this paper, we presented a novel neighbourhood estimation method which is both memory and computation efficient. Moreover, the proposed technique also leverages the common “fixed-n-neighbours” problem for standard “best-k-neighbours” techniques, therefore allows better recommendation quality for recommenders. We combined the proposed technique with a taxonomy-driven product recommender, and in our experiment, both time efficiency and recommendation quality of the recommender are improved.

Improvement of web data clustering using web page contents

This paper presents an approach that discovers clusters of Web pages based on Web log data and Web page contents as well. Most existing Web log mining techniques are access-based approaches that statistically analyze the log data without paying much attention on the contents of the pages. The log data contains various kinds of noise which can significantly influence the performance of pure access-based web log mining. The method proposed in this paper not only considers the frequence of page co-occurrence in user access logs, but also takes into account the web page contents to cluster Web pages. We also present a method of using information entropy to prune away irrelevant papges which improves the performance of the web page clustering.