| 基于语义加权的引文网络社区划分研究 |
| |
| 引用本文： | 刘璐,蔡永明.基于语义加权的引文网络社区划分研究[J].新世纪图书馆,2021(1). |
| |
| 作者姓名： | 刘璐 蔡永明 |
| |
| 作者单位： | 济南大学商学院 |
| |
| 基金项目： | 国家社会科学基金项目“国家战略性新兴产业政策对关键核心技术创新影响机制与路径研究”(项目编号:19BGL038)资助的研究成果。 |
| |
| 摘 要： | 为提高引文网络社区划分的准确性,以文档之间的语义关系以及引文之间的引用关系为基础,结合词汇在文档中的位置关系等信息,构建基于词汇语义加权的引文网络。通过GloVe模型对词汇向量化以充分利用词汇语义信息,结合WMD模型度量文献之间的相似度,把文档相似度的计算转变为在约束条件下求线性规划最优解的问题,结合文本的内容及结构特征对网络中的边进行赋权,以Louvain社区发现算法对加权后的引文网络进行社区划分,并对划分后的社区进行分析与检验,实验证明GloVe-WMD模型可提高引文网络社区划分的准确度。
|
| 关 键 词： | 引文网络 语义加权 社区划分 文本挖掘 自然语言处理 词嵌入 |
Research on the Classification of Citation Network Based on Semantic Weighting |
| |
| Authors: | Liu Lu Cai Yongming |
| |
| Abstract: | To improve the accuracy of citation network community division,citation network with lexical semantic weighting was constructed based on the semantic relationship between documents and the reference relationship between citations and the location relationship of words in documents and other information.The GloVe model was used to vectorize the words in order to make full use of the semantic information of the words.The WMD model was used to measure the similarity between literatures,and the calculation of the similarity of documents was transformed into the problem of finding the optimal solution of linear programming under the constraint condition.The edges in the network were weighted according to the similarity,content and structural features of the text.The citation network was divided into communities by the Louvain community discovery algorithm.The divided community is analyzed and tested.The results show that GloVe-WMD model can improve the accuracy of community division of Citation Network. |
| |
| Keywords: | Citation network Semantic weighting Community discovery Text mining Natural language processing Word embedding |
| 本文献已被 维普 等数据库收录！ |
|
