基于Phrase-LDA主题模型的茶产品群组推荐研究

发布时间：2018-01-26 05:00

本文关键词： 茶产品群组推荐 Phrase-LDA 融合策略　出处：《安徽农业大学》2016年硕士论文　论文类型：学位论文

【摘要】：随着网络和信息技术的进步,农产品电子商务也飞速发展。根据国家商务部的数据显示,2014年全国涉农电子商务平台正在蓬勃发展,目前已超三万家。以茶叶为代表的特色农产品,也正在抓住机遇,从线下迈向线上,通过网络走向千家万户。由于涉农电子商务的爆发式增长,网购人群的迅速增加,如何在海量信息中找到符合用户兴趣的商品和如何向一群有相同兴趣爱好的用户推荐商品,成为亟待解决的问题。推荐系统,利用消费者在电商平台上的隐性或者显性的行为,分析其偏好,在过载的信息中,推荐满足其偏好的商品。目前的推荐系统主要都是针对个人的,考虑用户和商品数量的急剧增加,一对一的推荐的成本过高,如何向群组进行推荐是非常的热门研究方向。本文通过介绍现有群组推荐的相关理论,分析了目前电子商务环境下茶叶推荐存在的问题和基于评分推荐的不足,提出了基于Phrase-LDA模型的茶产品的群组推荐模型。具体研究内容如下:(1)基于茶产品的评论信息,提取用户和茶产品评论中的主题,去表示用户偏好和茶产品的主题特征。分析了目前传统的用户偏好的表示方法的优点和不足,综合其他方法的优点,提出主题信息表示法。(2)针对传统的LDA是基于单词的,单词相对短语没有明确的语义。本文提出Phrase-LDA,并用其去融合群组用户偏好。Phrase-LDA主题模型一方面融合了群组用户的偏好,另一方面在主题表示时,短语的明确语义更细致的描述了用户偏好。(3)利用京东商城部分茶叶的数据进行实验仿真,验证所提Phrase-LDA主题模型的有效性,并实现了一个基于Phrase-LDA茶产品的群组推荐原型系统。本文的研究成果,进一步细化了用户偏好的研究。另外在群偏好的聚合策略方面,本文的方法有一定的新颖性,具有一定的借鉴意义。
[Abstract]:With the development of network and information technology, the rapid development of e-commerce of agricultural products. According to the national Ministry of commerce data show that in 2014 the national agricultural e-commerce platform is booming, currently has exceeded thirty thousand. Tea is the representative of the characteristics of agricultural products, is to seize the opportunity, from the line to line, through the network to thousands of households due to the explosive growth of agricultural e-commerce, online shopping population increased rapidly, how to find the mass of information goods and how to match the user's interest to a group of users have the same hobby to recommend commodities, has become an urgent problem. Recommendation system, the use of consumers in the electronic business platform of the implicit or explicit behavior and its analysis in preference, information overload, recommended to satisfy their preferences of goods. The current recommendation system is mainly for individual, considering the number of users and commodities sharply Increase, one of the recommended cost is too high, how to group recommendation is a popular research direction is. This paper introduces the related theory of existing group recommendation, analysis of the current e-commerce environment tearecommended problems and score recommended based on recommendation Phrase-LDA model based on the group of tea products specific contents are as follows: (1) the tea product reviews based on information extraction of user and tea product reviews the theme, theme features to represent user preferences and tea products. Analyzes the traditional user preference representation methods and the advantages and disadvantages, the comprehensive advantages of other methods, put forward the topic of information representation. (2) according to the traditional LDA is based on the word, the word phrase relative no clear semantics. In this paper Phrase-LDA, and used to merge user preferences in.Phrase-LDA model group On the one hand, the fusion group of user preferences, on the other hand, the theme said, clear semantic phrases more detailed description of the user preference. (3) simulation using part of the tea mall Jingdong data, verify the validity of the proposed Phrase-LDA model, and implements a prototype system of Phrase-LDA tea products recommended the group based on the results of this study, further refinement of the user preferences. In addition the group preference aggregation strategy, this method has a certain novelty, has a certain significance.

【学位授予单位】：安徽农业大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP391.3

【参考文献】