Study on Recommendation Method Based on Product Evaluation Concept Tree and Collaborative Filtering Algorithm

In view of the data sparseness of traditional collaborative filtering algorithms, this paper introduces product evaluation concept tree to optimize the calculation of similarity, and uses the concepts similarity replace the items similarity. The hypothesis of this new algorithm is that the customers tend to purchase products according with themselves. So if a customer has selected a product, then he or she is more likely to choose a similar product. Finally, we implement this algorithms improvement by c#. Experimental raw data is got from tablet PC reviews of JingDong Mall. We process the product feature scores and get the recommendation results based on the reviews mining. The experiment data proves that the recommendation result is reasonable.