基于情境感知的个性化推荐算法研究与应用

发布时间：2018-06-06 04:34

本文选题：信息过载 + 个性化推荐　；参考：《中北大学》2017年硕士论文

【摘要】：随着互联网产业的蓬勃发展,在带给人们生活巨大便利的同时,也带来严重的信息过载问题。现如今,个性化推荐已成为解决信息过载问题的重要手段之一,其应用也渗透于各个领域中,渗透于人们生活的各个方面。基于情境感知的个性化推荐技术已成为研究的重点,充分考虑情境信息所起的作用,将用户和商品的情境信息融入到推荐算法中,可以使得推荐更为有效,使得用户对推荐结果更为满意。本文将传统推荐算法和情境感知作为研究前提,从实际应用的角度考虑,针对当前在各种情境下推荐系统存在的问题,提出一种推荐精度和用户满意度较高的改进推荐算法并加以应用,本文的主要工作内容有:(1)对已有的推荐算法、情境感知理论和情境感知推荐技术进行了简单介绍,并对它们的研究现状和存在问题做出了具体分析。(2)针对目前推荐算法存在推荐精度不高、用户满意度低等问题,提出一种基于情境相似的协同过滤改进推荐算法。该算法依据物理学点电荷间存在磁力的作用,引入情境因子,构建新的用户-情境-商品模型;然后依据库仑定律,在添加磁力概念后,重新定义一个用户相似度公式;最后根据新的评分聚合函数计算得出更为准确的评分预测,从而进行推荐;最终从理论和实验上验证了改进算法的良好性能。(3)在改进推荐算法基础上,将其应用于系统中,对基于情境感知的就餐推荐系统进行了详细的设计。首先对设计就餐推荐系统进行了包括业务需求和性能需求在内的需求分析;接着对系统功能各模块进行了详细的设计,尤其是核心算法所在推荐模块的设计与实现;最后将传统算法下和改进算法下就餐推荐系统的推荐结果进行了分析对比,从而验证了改进算法的可行性和科学性。
[Abstract]:With the rapid development of Internet industry, it not only brings great convenience to people's life, but also brings serious information overload problem. Nowadays, personalized recommendation has become one of the important means to solve the problem of information overload, and its application has permeated every field and every aspect of people's life. The personalized recommendation technology based on situational awareness has become the focus of the research. Fully considering the role of situational information and integrating the information of users and commodities into the recommendation algorithm can make the recommendation more effective. Make users more satisfied with the recommended results. In this paper, the traditional recommendation algorithm and situational awareness are taken as the research premise, considering from the perspective of practical application, the problems existing in the current recommendation system in various situations are pointed out. An improved recommendation algorithm with high recommendation accuracy and user satisfaction is proposed and applied. The main work of this paper is to introduce the existing recommendation algorithm, the theory of situational awareness and the technology of situation-aware recommendation. The current research situation and existing problems are analyzed. (2) aiming at the problems of low recommendation accuracy and low user satisfaction, a new collaborative filtering improved recommendation algorithm based on situational similarity is proposed. A new user-situation-commodity model is constructed based on the effect of magnetic force between physical point charges, and then a user similarity formula is redefined according to Coulomb's law after adding the concept of magnetic force. Finally, according to the new score aggregation function to calculate a more accurate score prediction, so as to recommend. Finally, the theoretical and experimental results show that the improved algorithm has good performance. (3) on the basis of the improved recommendation algorithm, it is applied to the system. The repast recommendation system based on situational awareness is designed in detail. First of all, the design of the dining recommendation system including business requirements and performance requirements are analyzed, and then the system function modules are designed in detail, especially the design and implementation of the recommendation module where the core algorithm is. Finally, the recommended results of the traditional algorithm and the improved algorithm are analyzed and compared, which verifies the feasibility and scientific nature of the improved algorithm.
【学位授予单位】：中北大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.3

【参考文献】