和画图符推荐系统的研究

发布时间：2018-06-27 05:02

本文选题：和画 + 和音　；参考：《西北大学》2017年硕士论文

【摘要】：基于图符的地球村语言——和,目前涉及到5707个图符,而随着即时通讯工具——和音的推广及语料库——和画的完善,用户及图符数量会持续增长,使得从海量的和画信息中找寻用户感兴趣的图符变得非常困难。如何利用图符本身的特征及用户庞大的社交关系为用户提供服务,是本文推荐系统关注的焦点。本文针对和语言,分析了推荐系统、传统推荐算法和地球村语言的发展现状,总结并深入探讨了现有的研究成果,设计并开发了基于和画图符的推荐系统(Recommendation System Based on Harmony,RSoH)。主要的工作内容如下:1.提出了基于和画图符内容的推荐算法(Harmony-Content-based Recommendation Algorithm,HCB)。该算法对和画图符的内容进行了分析,改进了基于内容的推荐,并将其应用在RSoH中,实现了输入框处的推荐功能。2.给出基于综合信任度的推荐算法(Comprehensive-Trust-based Recommendation Algorithm,CTB),并将该算法应用到RSoH中,实现了其通过按钮提供推荐的功能。3.分析了和音及和画的相关数据,实现了对用户信息及和画内容的建模。4.设计并实现基于和画图符的推荐系统RSoH,对RSoH系统进行测试,经过测试验证了 RSoH系统的可用性。RSoH为用户提供个性化的推荐,输入框处的推荐为用户提供快速准确的图符推荐,通过按钮提供推荐则提供更高覆盖度和新颖性的图符列表。RSoH实现了对用户的主动推荐,并且对推荐做出相应解释,提高了用户的采纳率。
[Abstract]:The talk-based global village language-harmony currently involves 5707 symbols, and the number of users and symbols will continue to grow with the promotion of instant messaging tools-harmony sounds and the perfection of corpus-sum paintings. This makes it difficult to find symbols that are of interest to users from large amounts of and drawing information. The focus of this recommendation system is how to make use of the features of chars and the huge social relationship of users to provide services to users. This paper analyzes the development of recommendation system, traditional recommendation algorithm and global village language, summarizes and discusses the existing research results, and designs and develops a recommendation system based on HarmonyRSoH. The main tasks are as follows: 1: 1. The Harmony-Content-based recommendation algorithm (HCB) is proposed. The algorithm analyzes the content of the and drawing symbols, improves the content-based recommendation, and applies it to RSoH to realize the recommendation function at the input box. A comprehensive Trust-based recommendation algorithm CTB (Comprehensive Trust-based recommendation algorithm CTB) is presented, and the algorithm is applied to RSoH. The related data of harmony sound and painting are analyzed, and the modeling of user information and content is realized. 4. Design and implement the recommendation system RSoHbased on the basis of drawing symbol, test the RSoH system, and verify the usability of RSoH system. RSoH provides personalized recommendation for users, and the recommendation at input box provides users with quick and accurate recommendation. The recommendation provided by the button provides a higher coverage and novelty list. RSoH realizes the active recommendation to the user, and explains the recommendation accordingly, which improves the adoption rate of the user.
【学位授予单位】：西北大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.3

【相似文献】