一种基于效用的个性化文章推荐方法

发布时间：2018-06-09 02:06

本文选题：文章推荐 + 效用函数　；参考：《计算机学报》2017年12期

【摘要】：近年来,个性化推荐方法逐渐成为推荐系统领域的研究热点,文章推荐就是其中的一个分支.然而目前大部分文章推荐方法还存在以下问题:首先,这些方法大多还是基于协同过滤或者基于内容的推荐,而这些方法无法满足用户因自身偏好带来的需求,它们倾向于通过获取文章的各种不同属性来代表文章,这样的获取方式存在着对文章本身理解的偏差.其次,还存在一些文章推荐方法使用搜索代替推荐,这种方法仅仅以用户的偏好为基础,并不考虑文章的内在属性和联系,反馈回的数据依赖于用户做大量的判断.这些都为文章推荐方法的研究提出了挑战与机遇.该文主要聚焦的问题是:(1)已有的文章有多大的可能性成为用户的选择.对于推荐方法来说,过去的文章更多的是通过一些文章属性来获取.如研究领域、作者姓名、文章权威性、文章发表时间、引用数量等,这些属性的数据量庞大且没有量化标准,用户不容易通过这些信息来准确获取自己想要的文章;(2)什么样的推荐方法能将用户需要的文章更好的推荐出来.由于用户的偏好是一种理性和感性的共同选择,因此推荐方法不仅需要考虑文章自身的影响力,还要考虑用户需求的程度和方向.该文通过分析用户与文章的关系,提出一种个性化文章推荐方法,该方法基于效用理论,为了得到效用函数,该文将文章属性和用户偏好作为效用函数的元素,首先分析了文章属性和用户偏好之间的关系,其次通过最优化的效用函数得到用户对文章的评价值,最后通过对不同的评价值进行比较,将相关的文章推荐给用户.这样就可以通过效用函数最终获取用户所需的文章.该文的贡献主要包括以下几方面.(1)建立了一种关于论文属性和用户偏好的组合模型,该模型是通过效用函数建立的,目标是使用户的利益达到最大;(2)效用函数中的变量来自于论文属性和用户偏好的取值,将论文属性和用户偏好同时进行量化,比以往单一的通过文章属性建立的方法更灵活,也比单一通过用户的偏好建立的方法更全面.通过实验对比利用该方法推荐的文章相对于其他方法准确率平均提高了7.17%,召回率平均提高了4.55%,说明该方法推荐的文章质量是可以保证的.
[Abstract]:In recent years, personalized recommendation method has gradually become a hot topic in the field of recommendation system, and recommendation is one of the branches. However, there are still some problems in most of the recommended methods: firstly, most of these methods are based on collaborative filtering or content-based recommendations, and these methods can not meet the needs of users because of their preferences. They tend to represent the article by acquiring different attributes of the article, and there is a deviation in the understanding of the article itself. Secondly, there are some article recommendation methods that use search instead of recommendation. This method is based on the user's preference and does not consider the intrinsic attributes and connections of the article. The feedback data depends on the user to do a lot of judgment. All these provide challenges and opportunities for the study of recommended methods. The main focus of the article is the question of how likely existing articles are to become a user's choice. For recommended methods, the past articles are more through a number of article properties to obtain. Such as the research field, the author's name, the authoritativeness of the article, the time when the article was published, the number of references, and so on. It is not easy for users to accurately obtain the articles they want through this information.) what kind of recommendation method can better recommend the articles that users need. Because the user's preference is a rational and perceptual choice, the recommended method should not only consider the influence of the article itself, but also consider the degree and direction of the user's demand. By analyzing the relationship between users and articles, this paper proposes a personalized article recommendation method. This method is based on utility theory. In order to obtain utility function, the article attributes and user preferences are taken as the elements of utility function. This paper first analyzes the relationship between article attributes and user preferences, and then obtains the evaluation value of the article by optimizing utility function. Finally, by comparing the different evaluation values, the related articles are recommended to the user. In this way, the utility function can be used to obtain the user's desired article. The main contributions of this paper are as follows: (1) A combination model of paper attributes and user preferences is established, which is based on utility function. The goal is to maximize the user's interest) the variables in the utility function come from the values of the paper attribute and the user preference. The paper attribute and the user preference are quantified simultaneously, which is more flexible than the single method established by the article attribute in the past. It is also more comprehensive than a single approach based on user preferences. Compared with other methods, the article recommended by this method has an average accuracy of 7.17 and a recall rate of 4.55, which shows that the quality of the article recommended by this method can be guaranteed.
【作者单位】：华中科技大学计算机科学与技术学院;华中科技大学武汉光电国家实验室;华中科技大学信息存储系统教育部重点实验室;
【基金】：国家“九七三”重点基础研究发展规划项目基金(2011CB302301) 深圳市科技研发基金(JCYJ20170307172447622)资助~~
【分类号】：TP391.3

【相似文献】