基于影响力分析的社交用户推荐方法研究

发布时间：2018-05-29 00:27

本文选题：社交网络分析 + 影响力最大化　；参考：《中国科学技术大学》2015年硕士论文

【摘要】：随着信息技术和互联网的发展与普及,越来越多的人能够在网络上随时随地的发表自己的想法,分享自己的所见所闻；在线社交网络(比如Twitter、Facebook、新浪微博等)在信息的传播、消息的扩散等方面扮演着重要的角色。近年来,利用社交网络进行产品营销或品牌推广受到了广泛的关注。病毒营销作为一种常见的营销方法,主要利用用户之间的信息传播效应来进行产品营销或品牌推广。影响力传播最大化问题因在病毒营销中的重要应用而得到了广泛的研究,然而目前的研究工作大多关注一组用户的整体影响力最大化问题,忽略了用户个体的影响力及其最大化问题。因此,本文首先站在用户自身的角度,对面向影响力增益的社交用户推荐问题进行了研究,提出了用户个体影响力最大化问题,即用户个体如何建立新的连接能使自己在社交网络中的影响力增益最大。另一方面,目前对产品营销中的影响力传播分析,只关注用户之间的影响力,没有对企业与用户之间的影响力进行深入研究。鉴于此,本文分析企业与用户之间的影响力对用户消费行为的影响,研究了多重影响力的潜在用户推荐问题,并提出了一个计算框架及相应算法,能够向产品营销中的企业推荐最具价值潜力的一组用户。本文的研究工作与贡献可以总结如下： (1)对面向影响力增益的社交用户推荐问题进行了研究。本文首先提出并形式化定义了个体影响力最大化问题；然后利用目标函数性质提出了启发式的贪心算法,并结合具体模型计算特性进一步提出了线性时间复杂度的]uBound算法；最后,在四个真实的数据集上的实验结果表明了算法的有效性和高效性。 (2)对基于多重影响力的潜在用户推荐问题进行了研究。本文首先讨论了企业与用户之间的影响力关系及其对用户消费行为的影响；然后提出了一个计算框架及相应算法,来向产品营销中的企业推荐最具有潜力价值的用户(这部分用户对产品感兴趣,但不倾向于选择哪一家公司的产品)；最后,在两个真实的数据集上的实验结果验证了计算框架及相应算法的有效性。
[Abstract]:With the development and popularity of information technology and the Internet, more and more people are able to publish their ideas and share what they have seen and heard at any time and anywhere on the Internet; the spread of online social networks (such as Twitter Facebook, Sina Weibo, etc.) The spread of information plays an important role. In recent years, the use of social networks for product marketing or brand promotion has received widespread attention. As a common marketing method, virus marketing mainly utilizes the effect of information dissemination between users to carry out product marketing or brand promotion. The issue of maximization of influence dissemination has been widely studied because of its important application in virus marketing. However, most of the current research work has focused on the problem of maximizing the overall influence of a group of users. The influence of individual user and its maximization are ignored. Therefore, this paper firstly studies the problem of social user recommendation oriented to influence gain from the point of view of the user itself, and puts forward the problem of maximizing the individual influence of the user. How individual users establish new connections can maximize their influence on social networks. On the other hand, the analysis of influence dissemination in product marketing only focuses on the influence between users, but not on the influence between enterprises and users. In view of this, this paper analyzes the influence of the influence between the enterprise and the user on the consumer behavior, studies the potential user recommendation problem with multiple influences, and proposes a computing framework and corresponding algorithm. A group of users who can recommend the most valuable potential to a company in product marketing. The research work and contribution of this paper can be summarized as follows: 1) the problem of social user recommendation for influence gain is studied. In this paper, we first propose and formalize the problem of maximizing individual influence, then we propose a heuristic greedy algorithm by using the property of objective function, and then we further propose a linear time complexity (uBound) algorithm based on the computational characteristics of a specific model. Finally, the experimental results on four real data sets show the effectiveness and efficiency of the algorithm. The problem of potential user recommendation based on multiple influences is studied. This paper first discusses the influence relationship between enterprise and user and its influence on consumer behavior, and then proposes a computational framework and corresponding algorithm. To recommend the most potential users to companies in product marketing (this group of users are interested in the product but are not inclined to choose which company's product); finally, Experimental results on two real data sets verify the validity of the computational framework and the corresponding algorithm.
【学位授予单位】：中国科学技术大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP391.3

【共引文献】