基于Web日志挖掘的农业信息个性化推荐系统研究

发布时间：2018-05-07 10:01

本文选题：农业信息 + Web日志挖掘　；参考：《湖南农业大学》2016年硕士论文

【摘要】：随着Internet的快速发展以及国家对“三农”问题的愈发重视,各类农业信息网站如雨后春笋应运而生。面对各种类型的农业信息网站,在海量的数据信息中寻找有价值、感兴趣的内容,对于技术水平有限的农业用户来说,是一件有困难的事情。如何打造一个有效的推荐系统,为农业用户提供个性化信息推荐服务,帮助用户从“人找信息”变成“信息找人”,成为农业信息领域亟需解决的问题。本文在综合研究课题研究背景、研究意义、国内外研究现状的基础上,深入研究了数据挖掘、Web挖掘、个性化推荐的相关理论知识。针对农业信息网站存在的问题,提出了基于Web日志挖掘的农业信息个性化推荐系统解决方案。本文的具体研究内容如下：1、提出了通过双重阈值控制的方法,提高基于文本语义特征的Item-Based协同过滤推荐结果准确度的改进算法。2、研究了基于内容推荐的核心流程,包括特征提取、文本向量表示、用户兴趣模型建立。考虑新注册用户的情况,总结了基于用户注册信息的推荐算法。3、提出了基于内容推荐、基于用户注册信息推荐和Item-Based协同过滤推荐的混合推荐算法,解决了农业信息个性化推荐领域存在的问题。4、设计并实现基于Web日志挖掘的农业信息个性化推荐系统。实验验证了算法和系统的有效性。
[Abstract]:With the rapid development of Internet and the increasing attention of the country to the issue of agriculture, rural areas and farmers, all kinds of agricultural information websites emerge as the times require. In the face of all kinds of agricultural information websites, it is difficult for the agricultural users with limited technology to find the valuable and interesting content in the huge amount of data information. How to build an effective recommendation system, to provide personalized information recommendation service for agricultural users, and to help users change from "looking for information" to "finding information" has become a problem that needs to be solved in the field of agricultural information. On the basis of comprehensive research background, research significance and current research situation at home and abroad, this paper deeply studies the relevant theoretical knowledge of Web mining and personalized recommendation of data mining. Aiming at the problems of agricultural information website, a solution of personalized recommendation system of agricultural information based on Web log mining is put forward. The specific research contents of this paper are as follows: 1. An improved algorithm of improving the accuracy of Item-Based collaborative filtering recommendation based on semantic features of text is proposed by double threshold control. The core flow of content-based recommendation is studied. Including feature extraction, text vector representation, user interest model establishment. Considering the situation of new registered users, this paper summarizes the recommendation algorithm. 3 based on user registration information, proposes a hybrid recommendation algorithm based on content recommendation, user registration information recommendation and Item-Based collaborative filtering recommendation. This paper solves the problems in the field of personalized recommendation of agricultural information, and designs and implements a personalized recommendation system of agricultural information based on Web log mining. Experiments show that the algorithm and the system are effective.
【学位授予单位】：湖南农业大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP391.3

【参考文献】