Personalization Services based on Collaborative Filtering Algorithms
HUANG Tianshou -, Yongliang Shi, HUANG Yiqun -, RONG Wei - · INTERNATIONAL JOURNAL ON Advances in Information Sciences and Service Sciences · 2011
A good personalization strategy can increase sales by improving customer conversion ratio, enhance customer loyalty by improving relationship with customers, or in other words, increase revenue and profit. The heart of personalization is to serve individual customers unique needs. However, customers needs are hard to pin-down; every customer is unique and so are his or her needs. With the constraints of traditional collaborative filtering (CF) methods, the heat conduction process is introduced into the CF method to enhance the personalization service performance. By introducing a new user similarity formulation base on the heat conduction process, we introduce a modified collaborative filtering algorithm (MCF), which has remarkably higher accuracy than the standard collaborative filtering. Furthermore, by introducing a tunable parameter l , the degree effects of the termination nodes on the performance are investigated. The numerical simulation results on a benchmark data set shows that the algorithmic accuracy of the MCF, measured by the ranking score, is further improved by 6.3% in the optimal case. In addition, two significant criteria of algorithmic performance, diversity and popularity, are also taken into account. Numerical results show that when the recommendation list L = 50 the diversity of the presented algorithm is improved 14.3%. There has long been a trade-off between satisfying more customers and better meeting the needs of every individual customer. This work may shed some light on improving the personalization service performance.