Personalized Online Product Review Selection Study

Traditional review selection algorithm based on the usefulness of reviews and the amount of information contained in the review,which leads to the higher redundancy and repetition rate for selected review sets.Therefore,based on the user's preference,a review that is most consistent with the user's preference is selected from the mass merchandise review sets to truly satisfy the user's needs.In order to solve this problem,this paper completes the following work: 1.We combine the classical TF-IDF feature extraction algorithm with Skip-gram model and Canopy clustering algorithm to build a hybrid multi model.The model solves user preferences based on text content mining and user evaluation of product attributes.2.Improve the traditional view diversification review selection algorithm,and the improved review selection algorithm,it can select the best set of review texts that meet the user's preference requirements,so as to achieve personalized online product review selection.3.Design model and algorithm evaluation functions,performance evaluation of related models and algorithms.For related models,we use RMS values for model performance evaluation. For the review selection algorithm,we perform performance evaluation and analysis of the algorithm based on two measures of accuracy and efficiency.4.The related models constructed and the improved review selection algorithm are applied in practice.Based on the results,the performance of related models and algorithms is evaluated and summarized.