基于关键词语义网络的领域主题演化分析方法研究  

Research on the Domain Theme Evolution Analysis Based on Keywords Semantic Network

作　　者：巴志超 杨子江[1] 朱世伟[1] 王蕾[1] 

机构地区：[1]山东省科学院情报研究所,山东济南250014

出　　处：《情报理论与实践》2016年第3期67-72,共6页Information Studies:Theory & Application

摘　　要：有效揭示领域中主题的浮现机理及发展轨迹,对学科、领域未来发展态势进行预测和战略决策具有重要意义。文章提出一种基于关键词语义网络的领域主题演化分析方法。首先,通过引入浅层神经网络语言模型word2vec对领域文献的题名、摘要进行建模学习,将关键词表示成语义级别的词向量结构;其次,在建模基础上,结合Equivalence共现系数进行关键词语义相似度计算并构建关键词语义网络;最后采用社会网络分析方法从合著网络、共现网络等角度对领域主题进行演化特征分析。实验表明:该方法能够有效地识别领域的热点主题及发展趋势。Reveal of emergence mechanism and development track of theme in the field have important significance in predica- tion of future development trend and strategic decision-making for discipline and domain. This paper presents a new theme evolution analysis method based on keywords semantic network. Firstly, titles and abstracts are modeled to indicate word vector structure for keywords based on neural network language model--word2vee. Secondly, through the study of word distribution and combination of Equivalence co-occurrence coefficient, the semantic association of keywords is calculated to build up the semantic network. Finally, using social network analysis method, the paper analyzes the evolution characteristics of domain theme from collaboration network and co-occurrence network. The experiments demonstrate that the proposed approach can accurately identify the hot topics and devel- opment trend for disciplinary field.

关 键 词：语义网络 神经网络语言模型 词向量 主题演化分析 语义模型 

分 类 号：G254[文化科学—图书馆学]