Mixed Recommendation Models Based on Rating Matrix and Review Text
With the rapid application of recommendation system in E-business services, the data sparsity brought by it has increased to the factor of inaccurate score prediction.A mixed recommendation model ( MRB) was proposed that combines scoring and comment text.The weights of "user common friends"was added to User-based-prediction(UBP) and the "project-time" was added to Item-based-prediction ( IBP) .The experimental verification of the mean absolute error( MAE) showed that the model was obvi-ously superior to the traditional model when the number of nearest neighbors was near 50 .
万方数据
维普
