情感词汇共现网络的复杂网络特性分析

The Complexity Analysis of the Emotional Word Co-occurrence Network

本文从情感计算这一热点研究问题出发,分析了情感词汇共现网络的定义与构建原理,阐述了对其小世界效应、无标度特性、网络弹性、度相关性等复杂网络统计学特性进行研究的方法。为了检验这些统计学特性,从22157条网络评论中抽取出1284个情感词汇,并通过统计其在12000条评论语句中的共现情况建立了情感词汇共现网络。经计算,该网络的平均最短路径为2.89,群聚系数为0.19,表明其具有小世界效应;该网络的顶点度和边权重都呈幂律分布,表明其具有无标度特性。研究结果还表明,情感词汇共现网络的顶点度、顶点强度和顶点交互系数之间具有正相关性,是同类混合网络。

Emotional computing is becoming a hot topic.From this,the definition and construction of the Emotional Word Co-occurrence Network（EWCN） is firstly analyzed.Then,the paper elaborate the approach to study the small-world effect, scale-free degree distribution,network resilience,mixing patterns and other statistical properties of EWCN.To validate this,1284 emotional words are extracted from 22157 online customer reviews,and then the EWCN is constructed by counting the distribution of the 1284 words in 12000 sentences.The results show that the average path length is 2.89 and the clustering coefficient is 0.19, which is consistent with the small-world effect.The vertex degree distribution and edge weights distribution obey the power-law distribution,indicating that EWCN has a scale-free structure.The results also show that the degree of a vertex is proportional to its strength and interaction coefficient,and EWCN is an assortative mixing network.