Hybrid Personalized Recommended Model Based on Genetic Algorithm

Amount of customer and product is huge and variant. It usually decreases the accuracy of personalized recommender system. To solve this problem, a hybrid model is proposed to integrate outputs produced by every recommender at the basis of genetic algorithm. According to features of recommend task, weight vectors are constructed to represent different forecasting performances of each recommender. Then, the hybrid problem is translated into optimizing problem about weight vectors. Fitness function is designed based on forecasting accuracy, and a selecting strategy is constructed to avoid pre-mature of GA. Experiments express the proposed model has higher capability than traditional method.