文章摘要
牛奉高,高旭霞,李志欣.基于文献的人工智能领域研究的网络特征与演化分析[J].情报工程,2019,5(4):033-043
基于文献的人工智能领域研究的网络特征与演化分析
Network Characteristics and Evolution Analysis of Artificial Intelligence Field Research Based on Literature
  
DOI:10.3772/j.issn.2095-915X.2019.04.004
中文关键词: 人工智能;加权关键词共现网络(W-KCN);拓扑特征;演化规律;加权创新系数
英文关键词: Artificial Intelligence; weighted keyword co-occurrence network(W-KCN); topological characteristics; evolution law; weighted innovation coefficient
基金项目:山西省基础研究项目“ 加权共现潜在语义向量空间模型及其在文本主题聚类应用中的惩罚性矩阵分解研究” (201801D211002);山西省高等学校创新人才支持计划“ 基于潜在语义的文本信息主题深度聚类研究”(2016052006);国家自然科 学基金项目“ 共现潜在语义向量空间模型及其语义核的构建与应用研究”(71503151)。
作者单位
牛奉高 山西大学数学科学学院 
高旭霞 山西大学数学科学学院 
李志欣 山西大学数学科学学院 
摘要点击次数: 220
全文下载次数: 265
中文摘要:
      现阶段人工智能正处于蓬勃发展时期,挖掘“ 人工智能” 领域的主题特征及演化规律对掌握该 现阶段人工智能正处于蓬勃发展时期,挖掘“ 人工智能” 领域的主题特征及演化规律对掌握该领域的发展动态具有重要意义。为了更高效地掌握其发展动态,进一步了解未来研究趋势,本文定义了一个新的计量指标:加权创新系数。本文首先提取“ 人工智能” 领域文献中的关键词,通过关键词之间的共现关系和共现强度构建加权关键词共现网络(W-KCNs);其次考虑到每年新出现的关键词与往年已有关键词重要度的不同,定义了一个度量加权网络创新度的指标:加权创新系数;然后着重从节点强度分布、平均加权最近邻度和平均加权聚类系数对W-KCNs 进行拓扑特征分析。研究发现:改进后的加权创新系数更能准确描述每年关键词的创新度;W-KCNs 是异配网络,节点的强度分布近似于幂律分布,且网络中度数小的关键词加权聚类系数大,容易形成类簇。
英文摘要:
      At present, artificial intelligence is in the period of vigorous development, thus, it is of great significance to excavate the theme characteristics and evolution law in the field of “artificial intelligence” to grasp its development dynamics.In order to grasp its development dynamics more efficiently and further understand the future research trend, this paper defines a new measurement index: weighted innovation coefficient. This paper first extracts the keywords in the field of “artificial intelligenc”,constructs the Weighted keyword co-occurrence network (W-KCNs) through the co-occurrence relationship and the co-occurrence intensity between the keywords. Considering the differences in the importance of new keywords from previous years, an index to measure the innovation degree of weighted network is defined: the weighted innovation coefficient. Then the topological characteristics of W-KCNs are analyzed from the point of node strength distribution, average weighted nearest neighbor degree and average weighted clustering coefficient. It is found that the improved weighted innovation coefficient can accurately describe the innovation degree of keywords in each year. W-KCNs is a heterogeneous network, the strength distribution of nodes is similar to the power law distribution, and the keyword weighted clustering coefficients with small degrees in the network are large, and it is easy for clusters.
查看全文   查看/发表评论  下载PDF阅读器
关闭

分享按钮