摘要
[目的/意义]知识图谱已经成为海量信息资源知识组织的新形态。将知识图谱应用于虚拟学术社区用户生成内容知识组织中,对于虚拟学术社区知识发现及服务具有重要意义。[方法/过程]提出一种基于知识图谱的虚拟学术社区用户生成内容知识共聚框架。首先,运用Word2Vec词向量模型对虚拟学术社区用户生成内容数据集进行词向量表示;其次,基于双向长短记忆神经网络模型与条件随机场识别出虚拟学术社区用户生成内容中的命名实体,基于双向门控循环单元与注意力机制进行实体关系抽取;最后,借助Neo4j图数据库对知识共聚的结果进行可视化展示,并采集丁香园心血管论坛的学术交流帖子,对提出的知识共聚框架进行实证分析。[结果/结论]基于知识图谱的虚拟学术社区用户生成内容知识共聚方法能够有效序化重组虚拟学术社区知识资源,细粒度挖掘和揭示不同类型的知识单元和知识关联,有助于为虚拟学术社区智能知识服务提供语义理解和人工智能的基础。
[Purpose/significance] Knowledge graphs have become a new form of knowledge organization for massive information resources.Applying knowledge graphs to the organization of user-generated content(UGC) knowledge in virtual academic communities is of great significance for knowledge discovery and services in virtual academic communities.[Method/process] A knowledge mapping-based knowledge copolymerization framework for user-generated content in virtual academic communities was proposed.Firstly,the Word2vec word vector model was applied to the word vector representation of the virtual academic community UGC dataset.Secondly,named entities in the user-generated content of the virtual academic community were identified based on the bi-directional long short term memory(Bi-LSTM) network model and the conditional random field(CRF),and entity relationship extraction was performed based on the bi-directional gate recurrent unit(Bi-GRU) and the attention mechanism.Finally,the results of knowledge copolymerization were visualized with the help of Neo4j graph database.Academic exchange posts from the Dingxiangyuan Forum were also collected to empirically analyze the proposed knowledge copolymerisation framework.[Result/conclusion] It was found that the knowledge graph-based UGC knowledge copolymerization method for virtual academic communities can effectively sequentially reorganize virtual academic community knowledge resources,fine-grained mining and revealing different types of knowledge units and knowledge associations,which helps provide a basis for semantic understanding and artificial intelligence for intelligent knowledge services in virtual academic communities.
出处
《情报理论与实践》
CSSCI
北大核心
2023年第12期157-166,192,共11页
Information Studies:Theory & Application
基金
国家社会科学基金青年项目“面向突发公共卫生事件的社交媒体用户情感分析与舆情预警研究”(项目编号:22CTQ020)
吉林省科技发展计划项目“移动自媒体平台数智化精准知识服务及关键技术研究”(项目编号:20220508039RC)。
关键词
知识图谱
知识共聚
虚拟学术社区
用户生成内容
实体识别
关系抽取
knowledge graph
knowledge copolymerization
virtual academic community
user-generated content
entity recognition
relation extraction
作者简介
卢恒(ORCID:0000-0002-6680-5915),男,1995年生,博士,讲师,研究方向:知识聚合;陈章杰(ORCID:0009-0005-4412-9265),男,2001年生,硕士生,研究方向:知识服务;通信作者:周知(ORCID:0000-0002-5530-2968),男,1989年生,博士,副教授,研究方向:用户行为。
