摘要
[研究目的]为提高专利知识图谱构建的自动化水平,并实现知识服务与交易服务的融合,提出了面向供需信息挖掘的供需知识图谱(PSD-KG)的构建思路。[研究方法]知识图谱规划方面,对专利交易涉及实体及关系进行了拓展,规划了共由12类实体和14类关系组成的PSD-KG。知识图谱构建方法上,建立专利领域词典以实现语料自动化标注,并提出了基于BERT-BiLSTM-CRF模型的语义实体识别方法。[研究结论]与传统的CRF、BiLSTM-CRF模型对比发现,该文模型的准确率、召回率和F1指数均高于85%,验证了方法的有效性;以燃料电池领域为例构建PSD-KG,通过技术供需热点识别及演化研究,识别出三类技术热点,包括:持续热门技术点、新兴热门技术点和潜在热点技术;并在交易网络分析、供需信息检索等方面挖掘新应用场景。研究成果也为专利交易推荐提供了知识库。
[Research purpose]To improve the automation level of patent knowledge graph construction and realize the integration of knowledge service and transaction service,a construction idea of supply and demand knowledge graph(PSD-KG)for supply and demand information mining is proposed.[Research method]In terms of knowledge graph planning,the entities and relationships involved in patent transactions are expanded,and a PSD-KG consisting of 12 types of entities and 14 types of relationships is planned.In the knowledge graph construction method,a dictionary of patent domain is established to realize automatic corpus labeling,and a semantic entity recognition method based on the BERT-BiLSTM-CRF model is proposed.[Research conclusion]Compared with the traditional CRF and BiLSTM-CRF models,it is found that the accuracy,recall and F1 index of the model in this paper are all higher than 85%,which verifies the effectiveness of the method.Taking the fuel cell field as an example to build a PSD-KG,through the identification and evolution research of technological supply and demand hotspots,three types of technological hotspots are identified,including:persistent hotspots,emerging hotspots and potential hotspots;and new application scenarios are explored in transaction network analysis,supply and demand information retrieval,etc.The research results also provide a knowledge base for patent transaction recommendations.
作者
何喜军
张佑
孟雪
武玉英
He Xijun;Zhang You;Meng Xue;Wu Yuying(School of Economics and Management,Beijing University of Technology,Beijing 100124)
出处
《情报杂志》
CSSCI
北大核心
2023年第3期139-150,共12页
Journal of Intelligence
基金
国家自然科学基金面上项目“异构信息网络下技术供需匹配模型与对接路径研究”(编号:71974009)
国家自然科学基金项目“工程教育中非技术能力的表征及多源定量评价研究”(编号:71774010)
国际科研合作基金项目“基于属性异构网络表示学习的技术交易推荐方法研究”(编号:2021B35)。
