Products recommend algorithm based on customer preference model and affective computing

Products recommending on personalized preference of customer is one kind of effective product recommend algorithm base on contents, in which the difficulties are uncertainty and descriptive fuzziness of customer preference modeling. The model of customer preference base on feature space of products has been build by introducing the concept of affective computing. The degree of customer preference to each value of each product feature is well described by a value which is similar to membership in fuzzy space, and method of match degree computing between feature of product and preference of customer is proposed. On these bases, the complete products recommend algorithm base on customer preference model and affective computing is proposed. This products recommending algorithm recommends some products dynamically to customer, re-computing customer preference matrix with customer affective evaluation to products recommended, then go to the next iteration. It's proved that the result of products recommending of this algorithm converges to the result of products sorting with “real” customer preference matrix. The simulations also verify the convergence and effective of this algorithm.