基于Play的用户匹配与内容推荐系统设计与实现

发布时间：2018-07-10 19:54

本文选题：用户匹配 + 支持向量机　；参考：《电子科技大学》2013年硕士论文

【摘要】：随着各种互联网社区中的用户量和相关数据量的快速增长，用户越来越难以获取到对自己有价值的信息，为此，需要通过计算机帮助用户在海量信息中筛选出对每个用户有价值的内容。在较早时期发展起来的分类列表网站、搜索引擎等都是为了解决这个问题，然而这两种方式存在着各自的局限性，，分类列表网站的内容不会针对不同的用户分别提供内容，而是为了满足大多数用户的普遍需求而确定的，所以不能够实现个性化，也不利于对用户兴趣的挖掘；而搜索引擎则依赖于用户每次主动输入的关键词，因此容易将用户局限在已知的一个范围之内，不能够达到内容新颖的效果。为了克服上述局限，各种互联网社区开始结合机器智能算法来为用户提供更优质的服务，通过机器学习的方法来归纳和总结用户的行为习惯，达到理解用户偏好的目的，并根据这种学习到的偏好为用户提供个性化服务。但是，对于特定的应用领域应该使用何种方法，以及如何将已有的各种智能算法结合到实际系统中仍然是需要进一步研究的问题。本论文以为用户提供个性化的用户匹配服务和内容推荐服务为目标，总结了相关领域的研究现状，在对线性分类器、支持向量机和协同过滤等技术进行研究的基础上，以教师和学生之间的匹配需求以及教学资源的共享需求为背景，设计并实现了一种基于Play框架的用户匹配与内容推荐系统，其中用户匹配功能使用LIBSVM实现、内容推荐功能使用LensKit实现，两者都良好地整合到了基于Play框架实现的应用系统之中。本论文相关工作的先进性主要体现在以下两个方面：（1）使用支持向量机理论来解决教师和学生用户之间的匹配问题。支持向量机理论一般用于解决分类问题和回归分析问题，论文通过对匹配问题的转换，使得支持向量机可以应用于解决用户匹配问题，取得了良好的效果。（2）提出了一种新的推荐系统实现方案。结合Play框架和LensKit推荐库实现的推荐系统具有高度可配置、易于测试并且功能齐全等特点，可以广泛应用于实现各个领域的推荐系统。
[Abstract]:With the rapid growth of the number of users and related data in various Internet communities, it is becoming more and more difficult for users to obtain valuable information for themselves. Computers are needed to help users filter out content that is valuable to each user in a huge amount of information. The classifying list website and search engine developed in earlier period are all to solve this problem. However, these two methods have their own limitations, the content of the classified list website will not be provided separately for different users. It is determined to meet the general needs of most users, so it can not be personalized, and it is not conducive to the mining of users' interests, while search engines rely on the keywords that users input actively each time. Therefore, it is easy to limit the user to a known range, and can not achieve novel content effect. In order to overcome the above limitations, various Internet communities began to combine machine intelligent algorithms to provide users with better services, through machine learning methods to sum up and summarize user behavior habits, to achieve the purpose of understanding user preferences. And according to this learning preference for users to provide personalized services. However, what methods should be used in specific application fields and how to integrate various intelligent algorithms into practical systems are still problems that need to be further studied. This paper aims at providing personalized user matching service and content recommendation service, and summarizes the research status in related fields. Based on the research of linear classifier, support vector machine and collaborative filtering, etc. Based on the matching requirements between teachers and students and the sharing of teaching resources, a user matching and content recommendation system based on play framework is designed and implemented, in which the user matching function is implemented with LIBSVM. The content recommendation function is implemented by LensKit, both of which are well integrated into the application system based on play framework. The advancement of this paper is mainly reflected in the following two aspects: (1) support vector machine theory is used to solve the matching problem between teachers and students. Support vector machine (SVM) theory is generally used to solve classification and regression problems. Through the transformation of matching problem, support vector machine can be used to solve user matching problem. Good results have been achieved. (2) A new recommendation system is proposed. The recommendation system based on play framework and LensKit recommendation library is highly configurable, easy to test and has complete functions, so it can be widely used in the implementation of recommendation systems in various fields.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP391.3

【参考文献】