基于多目标免疫算法的网络个性化推荐

发布时间：2018-05-11 04:23

本文选题：个性化推荐系统 + 免疫算法　；参考：《天津工业大学》2017年硕士论文

【摘要】：随着互联网与信息技术的迅猛发展,人类已经进入到了一个数据爆炸的时代。在生活工作中,面对大量的信息,却无法从中获得对自己真正有用的信息,信息的使用效率反而降低了,这无疑会使淹没在信息海洋中的用户感到无比的苦恼。于是,如何能够快速高效地从海量数据中找出对自己有用的信息已经迫在眉睫。推荐系统是解决这一问题的有效手段,它能够很好地为用户提供若干条有用的推荐,被认为是缓解信息过载最有潜力的工具。通常,传统推荐系统的主要目的是使精确度最大化。但是,随着用户数量和产品的规模巨增,只考虑准确性已经完全不能满足用户的需求。个性化推荐系统是为每一个用户都"量身定做"推荐,并且考虑用户的多种需求。在个性化推荐系统中准确性和多样性是两个相互制约的性能指标,提高推荐的准确度无疑会有损推荐的多样性;同样,向用户推荐多样化的项目也会导致推荐准确性的降低。因此,在同时优化推荐准确度和多样性的过程中,需要找到一个适当的平衡。免疫优化算法是一种有效的智能算法,通过模拟免疫系统的原理和功能来解决复杂问题,在局部搜索和全局搜索中显示出优越的性能,为多目标优化问题提供了一种新的求解思路。本文对个性化推荐系统中的推荐技术进行了探索和研究,提出了基于多目标免疫算法的网络个性化推荐。该方法利用了生物免疫系统的基本原理,将要求解的个性化推荐列表建模成一个最大化推荐准确性和多样性的多目标优化问题,设计了适合个性化推荐问题求解的抗体编码方式、克隆算子、变异算子。仿真实验结果表明,该算法能够有效的得到个性化推荐的最佳解,提高了准确性和多样化,达到可以同时为多个用户提供多个不同推荐的需求。
[Abstract]:With the rapid development of Internet and information technology, mankind has entered a data explosion era. In life and work, faced with a large amount of information, but can not get their own real useful information, the efficiency of the use of information is reduced, which will undoubtedly cause inundation of users in the ocean of information. Therefore, how to quickly and efficiently find useful information from massive data is urgent. Recommendation system is an effective means to solve this problem. It can provide users with several useful recommendations and is considered to be the most potential tool to ease information overload. In general, the main purpose of traditional recommendation systems is to maximize accuracy. However, with the increase of the number of users and the scale of products, the accuracy alone can not meet the needs of users. Personalized recommendation system is for each user "tailor-made" recommendation, and take into account the user's various needs. In the personalized recommendation system, accuracy and diversity are two mutually restricted performance indicators, so improving the accuracy of recommendation will undoubtedly damage the diversity of recommendations. Similarly, the diversification of recommendation items to users will also lead to the reduction of recommendation accuracy. Therefore, a proper balance needs to be found in the process of optimizing the accuracy and diversity of recommendations at the same time. Immune optimization algorithm is an effective intelligent algorithm, which can solve complex problems by simulating the principle and function of immune system, and shows excellent performance in local search and global search. It provides a new method for solving multi-objective optimization problems. In this paper, the recommendation technology in personalized recommendation system is explored and studied, and the network personalized recommendation based on multi-objective immune algorithm is proposed. Based on the basic principles of biological immune system, the personalized recommendation list is modeled as a multi-objective optimization problem that maximizes recommendation accuracy and diversity. The antibody coding method, clone operator and mutation operator are designed for solving personalized recommendation problem. Simulation results show that the algorithm can effectively obtain the best solution of personalized recommendation, improve the accuracy and diversity, and can provide multiple different recommendations for multiple users at the same time.
【学位授予单位】：天津工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.3;TP18

【参考文献】