基于用户行为的产品关键词优化研究与实现

发布时间：2018-06-16 13:57

本文选题：用户行为 + 加权轨迹数据集　；参考：《东南大学》2016年硕士论文

【摘要】：近年来互联网经济快速发展,B2B电商平台连接生产企业与消费企业,在互联网经济中发挥着日益重要的作用。生产企业在平台发布的产品描述关键词直接影响平台用户对该产品的检索效果,对生产企业的经营至关重要。如何提升生产企业所发布产品描述关键词的受关注度直接体现电商平台的服务效果。论文工作结合电商平台实际需求,对基于用户行为的产品关键词优化进行研究,并开发原型系统。实现对目标产品关键词优化推荐,以期促进电子商务平台服务的个性化和定制化,提升服务质量。主要工作如下：(1)电商平台积累了大量用户访问信息,结合挖掘用户搜索行为模式主题,确定目标数据源,并进行预处理,提出一种加权轨迹数据集构建方法,兼顾数据规模的同时体现用户行为模式信息,生成挖掘数据集；(2)结合用户特征及其搜索产品类别对用户进行分类,设计频繁项集挖掘方法获取各类用户搜索主题词与其所关注产品关键词间关联模式。并从语义相关性和产生式前后件关联角度,设计聚类算法生成关键词关联词库和热词库；(3)结合所构建的关联词库和热词库,设计基于加权k近邻思想的关键词优化方法,实现对目标产品关键词优化推荐。并设计评估机制对基于词库的产品关键词优化效果进行评估,验证所提方法的有效性。在此基础上,实现基于用户行为的产品关键词优化原型系统。帮助用户合理制定产品描述关键词,提高产品的受关注度。目前,系统已通过委托研发验收,处于部署应用中。
[Abstract]:In recent years, the rapid development of Internet economy, B2B ecommerce platform connecting production enterprises and consumer enterprises, plays an increasingly important role in the Internet economy. The key words of product description issued by the production enterprise directly affect the retrieval effect of the product by the platform users, and are very important to the management of the production enterprise. How to improve the attention of product description keywords published by manufacturing enterprises directly reflects the service effect of e-commerce platform. According to the actual requirements of e-commerce platform, this paper studies the optimization of product keywords based on user behavior, and develops a prototype system. In order to promote the individualization and customization of E-commerce platform service and improve the service quality, we can optimize and recommend the target product keyword. The main work is as follows: (1) the ecommerce platform accumulates a large amount of user access information, combines with mining user search behavior pattern topic, determines the target data source, carries on the preprocessing, proposes a weighted track data set construction method. Taking into account the data scale and reflecting the user behavior pattern information, the mining data set is generated to classify the user in combination with the user characteristics and their search product categories. The frequent itemsets mining method is designed to obtain the associated patterns between the various user search subject words and their concerned product keywords. And from the point of view of semantic correlation and production relation, we design the clustering algorithm to generate keyword association lexicon and hot lexicon, and design a keyword optimization method based on weighted k-nearest neighbor thought combined with the related lexicon and hot lexicon. To achieve the target product keyword optimization recommendation. An evaluation mechanism is designed to evaluate the effect of product keyword optimization based on thesaurus to verify the effectiveness of the proposed method. On this basis, the prototype system of product keyword optimization based on user behavior is implemented. Help users to develop product description keywords, improve product attention. At present, the system has been commissioned by R & D acceptance, in the deployment of applications.
【学位授予单位】：东南大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP311.13

【参考文献】