个性化商品推荐系统的混合推荐算法研究  

Research on the Hybrid Recommendation Algorithm for the Personalized Product Recommendation System

作　　者：王晖[1] 张慧[1] WANG Hui;ZHANG Hui(Taizhou Mechanical and Electrical Higher Vocational School,Taizhou,Jiangsu Province,225300 China)

机构地区：[1]泰州机电高等职业技术学校,江苏泰州225300

出　　处：《科技资讯》2023年第22期248-252,共5页Science & Technology Information

摘　　要：推荐算法是推荐系统的核心内容,推荐算法的评价标准包含预分类准确性和测准确性。传统的推荐算法有两个明显缺陷,使用词频作为搜索文本的特征向量与无法克服高频词汇干扰。通过TF/IDF特征词加权改进算法提升分类准确性。提出混合模型LDTF,从信息增益的角度计算每个词性对词义的贡献增益,来判断一个特定词在此词性下能够代表的词义权重,用动态的计算不同词性的词性比,解决传统TF/IDF算法在文本识别的缺陷,使用CW-TF/IDF优化算法提升特征词的分类效果综合提升推荐准确度。为了解决内容推荐稀疏矩阵问题引入WSBCF协作推荐算法,提升推荐系统的用户体验,实验结果表明能在不同评分矩阵稀疏度下,统计能显著且明显提高。A recommendation algorithm is the core content of a recommendation system,and the evaluation criteria for the recommendation algorithm include pre-classification accuracy and measurement accuracy.The traditional recommendation algorithm has two obvious drawbacks:using word frequency as the feature vector for search texts and being unable to overcome the interference of high-frequency words.This article improves classification accuracy through the improved TF/IDF feature word weighting algorithm.This article proposes a hybrid model LDTF,which calculates the contribution gain of each part of speech to the meaning of a word from the perspective of information gain to determine the semantic weight that a specific word can represent under this part of speech,solves the shortcomings of the traditional TF/IDF algorithm in text recognition by dynamically calculating the part of speech ratio of different parts of speech,and improves the classification effect of feature words by using the CW-TF/IDF optimization algorithm to comprehensively improve recommendation accuracy.In order to solve the sparse matrix problem of content recommendation,this paper introduces the WSBCF collaborative recommendation algorithm to improve the user experience of the recommendation system,and the experimental results show that the statistics can be significantly and obviously improved under different scoring matrix sparsity.

关 键 词：商品推荐 特征词加权 推荐算法 稀疏矩阵 词义权重 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术]