Abstract
In today’s electronic commerce and knowledge economy environments, informationusers may experience information overload and seek for e-services to help them infiltering and selecting from an overwhelming array of product information. Onlinemerchandisers may also seek to better manage customer relationships that lead tohigher customer satisfaction and loyalty. In response, recommendation systems haveemerged as a class of e-service that are not only address the challenge of informationoverload by suggesting products of greatest interest to users, but also facilitatesorganizations better managing their customer relationships. Among variousrecommendation techniques, the collaborative filtering approach is the mostsuccessful and widely adopted one. However, the basic design of traditionalcollaborative filtering approach ignores item heterogeneities during therecommendation process. That is, all the user preferences on items are deemedidentically important and given an equal weight in measuring user similarities andpredicting user preferences. This may take unreliable users into consideration andtherefore, lead to poor recommendations. In this research, we design an item-basedcontent-weighted collaborative filtering approach to improve the recommendationeffectiveness by considering the content similarities of items. We conduct a serious ofexperiments to evaluate our proposed approach. The empirical evaluationdemonstrates that our proposed approach can achieve better prediction accuracy thanthose of the benchmarks. To further promote the prediction coverage of our proposedapproach, we also create a hybrid approach by combing our proposed approach withthe traditional one. The corresponding experimental results also show that the hybridapproach can achieve the same coverage as the traditional one and its predictionaccuracy is still better than that of the traditional one.