基于隐含狄利克雷分配模型的消费者在线评论复杂网络构建及其应用

发布时间：2018-02-14 22:06

本文关键词： 在线评论隐含狄利克雷分配模型复杂网络皮尔森相似度度分布　出处：《系统工程学报》2017年03期 　论文类型：期刊论文

【摘要】：为研究消费者在线评论的相互关系及整体演化发展,以隐含狄利克雷分配模型对消费者在线评论进行话题挖掘为基础,通过Pearson相似度确定评论间话题关系,构建了以评论为节点的复杂网络模型.以京东等平台上的手机评论语料库为例,对相应复杂网络的拓扑特性进行分析.研究表明,相同产品相同规模的评论网络统计特性相近;而相同平台上同一产品不同规模的评论网络性质差异明显,即小规模评论(产品销售初期)构成的网络是一个无标度网络,大规模评论(产品销售中后期)构成的网络演化为一个度分布为钟形分布的小世界网络.揭示了消费者发表在线评论行为之间相互作用,会影响整体评论网络的拓扑性质和演化规律,并为市场营销和管理实践提出了指导性建议.
[Abstract]:In order to study the relationship and evolution of consumers' online reviews, based on the topic mining of consumers' online comments based on the implicit Dirichlet assignment model, the topic relationship between reviews is determined by Pearson similarity. In this paper, a complex network model with comment nodes is constructed. Taking the mobile phone comment corpus on JingDong and other platforms as an example, the topological characteristics of the corresponding complex network are analyzed. The results show that the statistical characteristics of the same product and the same scale comment network are similar. However, on the same platform, the nature of the comment network of different scale of the same product is obviously different, that is, the network composed of small scale reviews (early product sales) is a scale-free network. The network of large-scale reviews (mid-to-late product sales) evolved into a small-world network with a bell-shaped distribution, revealing the interaction between consumer online reviews, It will affect the topological nature and evolution law of the whole comment network, and give some guiding suggestions for marketing and management practice.
【作者单位】：大连理工大学系统工程研究所;
【基金】：国家自然科学基金资助项目(61471083);国家自然科学基金重点资助项目(71431002) 教育部人文社科研究规划基金资助项目(14YJA630044)
【分类号】：O157.5;TP391.1

【相似文献】