基于兴趣学习的Web内容推荐及其优化研究

发布时间：2018-03-24 17:02

本文选题：用户兴趣学习　切入点：Web内容推荐　出处：《华中科技大学》2012年硕士论文

【摘要】：随着Internet的飞速发展，互联网已成为全球最大的分布式信息数据库。一方面，信息化给人们带来了极大的便利；另一方面，由于过量冗余的信息充斥网络，想要在网络上快速有效的提取有效信息也变得越来越困难。传统搜索是基于关键词检索的，但这种方法无法有效提取和检索到语义间的关联内容和隐含信息，在知识发现和查准查全率方面都有所欠缺。而个性化Web搜索技术的出现，可以有效缓解上述问题的出现，为用户提供更精细、准确和自动化的搜索。本文研究基于兴趣学习的Web内容推荐系统并对其进行优化，根据用户搜索所涉及的领域本体添加用户兴趣领域至用户本体，，通过概念和语义间的关系计算用户兴趣权重，并根据用户浏览行为实时更新本体，得到更准确的用户兴趣模型。由于用户兴趣作为搜索限制条件加入搜索语句，无疑增加了系统响应时间，本文通过研究图论算法，对搜索条件进行重新排序，通过选择估值减少中间结果集，选择高效的执行计划，提高连接查询效率，从而减少搜索响应时间，给用户创造更准确快捷的结果返回。本文首先介绍基于兴趣学习的Web内容推荐涉及的核心技术，在此基础上，研究用户兴趣学习算法，以达到提高用户查询搜索准确度的目的。由于用户兴趣增加了查询条件的复杂性，又通过查询优化策略优化查询时间，以达到提高用户查询搜索效率的目的。并对查询优化策略进行实验和其他方法的搜索引擎进行对比，验证了该方法可有效提高查询效率。通过研究及优化，改进后的基于兴趣学习的Web内容推荐系统在为用户推荐信息上将更符合用户的兴趣，同时查询效率也将有所提升。通过实验，将搜索结果按照用户兴趣模型重新排序后返回给用户，用户的满意度有所提高，可以看出改进后的用户兴趣模型更接近用户真实兴趣，可以减少翻页和搜索时间，给用户更愉悦的用户体验。将用户兴趣作为限制条件加入查询语句后的搜索系统，查询时间将会有所增加，经过本文方法的查询优化，在查询效率上也比优化前有所提高，尤其针对查询条件和语句关系较为复杂的情况，优化效果更为显著。
[Abstract]:With the rapid development of Internet, the Internet has become the largest distributed information database in the world. It is also becoming more and more difficult to extract effective information quickly and effectively on the network. Traditional search is based on keyword retrieval, but this method can not effectively extract and retrieve the associated content and hidden information between semantics. The emergence of personalized Web search technology can effectively alleviate the above problems and provide users with more precise accurate and automated search. This paper studies and optimizes the Web content recommendation system based on interest learning, adds the domain of interest to user ontology according to the domain ontology involved in user search, and calculates the weight of user interest through the relationship between concepts and semantics. According to the user browsing behavior, the ontology is updated in real time, and a more accurate user interest model is obtained. Because user interest is added to the search sentence as a search constraint, the response time of the system is undoubtedly increased, and the graph theory algorithm is studied in this paper. The search conditions are reordered to reduce the intermediate result set by selecting the estimation and the efficient execution plan to improve the efficiency of the join query so as to reduce the search response time and create a more accurate and fast result return for the user. This paper first introduces the core technology of Web content recommendation based on interest learning, and then studies the algorithm of user interest learning. In order to improve the search accuracy of users, because of the complexity of the query conditions, the query time is optimized by the query optimization strategy, because the interest of the user increases the complexity of the query conditions. In order to improve the search efficiency of users, the experiment of query optimization strategy is compared with the search engine of other methods, and it is proved that this method can effectively improve the efficiency of query. The improved Web content recommendation system based on interest learning will be more in line with the user's interest in recommending information, and the query efficiency will also be improved. Through experiments, the search results are reordered according to the user interest model and returned to the user. The user satisfaction is improved. It can be seen that the improved user interest model is closer to the real user interest and can reduce page turning and searching time. To give users a more pleasant user experience. After adding user interest as a restriction condition to the query system, the query time will be increased, and the query efficiency will also be improved after the optimization of the method in this paper. Especially in the case of complex query condition and statement relationship, the optimization effect is more remarkable.
【学位授予单位】：华中科技大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TP391.3

【参考文献】