基于移动互联网交友的个性化推荐系统的设计与实现

发布时间：2018-05-02 02:28

  本文选题：移动互联网 + 个性化推荐　； 参考：《贵州大学》2016年硕士论文

【摘要】：伴随着移动互联网的爆炸式增长与全民社交时代的到来,人们在海量信息中获取有效信息的效率正在下降。面对用户对自身交友需求不明确、用户搜索过滤条件不够丰富、用户搜索结果信息过多等问题,在交友过程中如何简洁高效的让用户找到兴趣相投的好友成为了社交网络的关键问题之一。个性化推荐系统能根据用户的基本信息、用户行为与好友信息,将显性信息与隐性信息相结合,通过推荐系统发现用户的兴趣点,从而引导用户发现自己的交友需求,能极大地降低用户获取有效信息的难度,提高社交网络的用户交友体验。因此,设计一套基于移动互联网交友的个性化推荐系统具有重要的理论价值和实践意义。本文针对目前社交网络在好友推荐中存在的问题,如冷启动、稀疏矩阵等,充分考虑用户对个性化推荐系统的需求,结合基于内容过滤算法和协同过滤算法,设计并实现了一种将基于内容过滤算法与协同过滤算法进行加权综合的个性化推荐系统。本系统首先通过用户数据的训练集对不同权值比的协同过滤与基于内容过滤进行多次训练,得出加权综合性能最佳时的权值比。然后利用TF-IDF算法对目标用户的基本信息数据进行预处理,确定每个特征项在基于内容过滤模块中的权值,并通过余弦相似度公式计算用户的相似度,得到基于内容过滤模块的推荐列表。同时依据目标用户的好友关系,得到协同过滤模块的推荐列表。最后依据之前确定的两算法推荐结果的权值比,对两算法的推荐列表进行加权综合,得到最终的综合推荐列表。本系统在结构上分为服务器和客户端,服务器采用java环境开发,客户端采用iOS平台。
[Abstract]:With the explosive growth of mobile Internet and the arrival of the era of social networking, the efficiency of obtaining effective information in mass information is declining. In the face of the user's unclear need for their own friends, the lack of rich conditions for user search and filtering, and the excessive amount of information about the user's search results, One of the key issues in social networking is how to find friends with similar interests in the process of making friends succinctly and efficiently. According to the basic information of the user, the user behavior and the friend information, the personalized recommendation system can combine the explicit information with the hidden information, and discover the user's interest point through the recommendation system, so as to guide the user to discover his need to make friends. It can greatly reduce the difficulty for users to obtain effective information and improve the experience of social network users making friends. Therefore, it is of great theoretical and practical significance to design a personalized recommendation system based on mobile internet dating. Aiming at the problems existing in friend recommendation of social network, such as cold start, sparse matrix and so on, this paper fully considers the user's demand for personalized recommendation system, and combines the content-based filtering algorithm and collaborative filtering algorithm. A personalized recommendation system based on content filtering and collaborative filtering is designed and implemented. In this system, the cooperative filtering and content-based filtering of different weights and values are trained several times through the training set of user data, and the weight / value ratio is obtained when the weighted synthesis performance is the best. Then the TF-IDF algorithm is used to preprocess the basic information data of the target user to determine the weight of each feature item in the content-based filtering module and to calculate the user similarity by using the cosine similarity formula. Get the list of recommendations based on the content filtering module. At the same time, according to the friend relationship of the target user, the recommendation list of the collaborative filtering module is obtained. Finally, according to the weight / value ratio of the recommended results of the two algorithms, the weighted synthesis of the recommended list of the two algorithms is carried out, and the final comprehensive recommendation list is obtained. The structure of the system is divided into server and client. The server is developed by java environment, and the client adopts iOS platform.
【学位授予单位】：贵州大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP391.3
，

本文编号：1832008