基于用户行为的水军团体阵营发现技术研究

发布时间：2018-01-22 16:29

本文关键词： 水军阵营行为目的检测模型核心节点　出处：《浙江工商大学》2017年硕士论文　论文类型：学位论文

【摘要】：由于互联网的开放性,网上用户的生成内容缺乏质量控制机制,导致网络虚假信息的泛滥。在电子商务领域中,商品品牌间相互竞争或者合作,雇佣大量水军维护自身品牌口碑或者打击竞争目标品牌。这些水军形成协同阵营,使得个体水军的行为更为隐蔽。水军阵营较单个水军个体而言,其所产生的负面作用更大,危害性更强,严重损害消费者和正当品牌商家的利益,电子商务领域网络秩序遭受严重破坏。因此,迫切需要对水军阵营甄别问题展开研究。当前,对网络水军的研究主要集中于网络水军个体或者水军组织的识别,识别方法的适用性低,识别效果并不理想,难以发现电子商务网站中可能存在于各个品牌下的水军阵营。在基于多特征尺度空间模型的组织发现技术基础上,本文提出基于用户行为的水军阵营发现技术。首先,本文采用基于特定品牌的网络水军组织识别方法,针对单个商品品牌构建目标商品组评论用户关系网,使用多特征尺度模型处理,解决了多特征模型存在的部分不足。其次,从用户行为目的出发,通过对水军行为的分析,构造水军检测模型,用于网络核心节点水军检测。再次,基于产品评分差异以及评论产品目标差异特征,进一步细分好评组织。最后,通过对水军高分评论产品集合与水军低分评论产品集合的分析,根据水军用户间共同受雇程度、共同攻击程度以及相互对立程度,构建水军阵营甄别模型。通过实验分析,实验结果证实基于用户行为的水军团体阵营发现技术较为有效,本文对方法中存在的误差进行相应的讨论。
[Abstract]:Because of the openness of the Internet, the lack of quality control mechanism for the content generated by online users leads to the proliferation of false information on the network. In the field of electronic commerce, commodity brands compete or cooperate with each other. Hire a large number of navy to maintain their own brand reputation or fight the competition target brand. These water forces form a synergistic camp, which makes the behavior of individual water army more subtle. The water army camp is more than the individual water army individual. Its negative effect is bigger, harmfulness is stronger, the interests of consumers and legitimate brand merchants are seriously damaged, and the network order in the field of electronic commerce is seriously damaged. There is an urgent need to study the problem of naval camp screening. At present, the research of network navy mainly focuses on the identification of the individual or organization of the network navy, the applicability of the identification method is low, and the recognition effect is not ideal. It is difficult to find the water army camp that may exist under each brand in the e-commerce website. Based on the organizational discovery technology based on the multi-feature scale space model. In this paper, we propose the discovery technology based on user behavior. Firstly, this paper uses the method of network navy organization identification based on a specific brand to construct the target commodity group comment user relationship network for a single commodity brand. The multi-feature scale model is used to solve the shortcomings of the multi-feature model. Secondly, based on the purpose of user behavior, through the analysis of the behavior of the water army, the detection model of the water army is constructed. Third, based on the product score difference and comment product goal difference characteristic, further subdivides the high praise organization. Through the analysis of the product set of high score and low score of water army, according to the degree of common employment, the degree of common attack and the degree of opposition among the users of water army. The experimental results show that the detection technology based on user behavior is more effective. The errors in the method are discussed in this paper.
【学位授予单位】：浙江工商大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F713.55;TP393.0

【相似文献】