摘要
[目的/意义]实体与实体间的关系可以被看作是知识的基本载体,非结构化数据中实体关系的抽取是进行知识开发利用需要解决的首要问题。[方法/过程]从非结构化文本的研究对象角度出发,利用知识库数据构建知识图谱并结合文物元数据进行实体关系的规则标注,使用深度学习有关方法进行信息资源中实体关系的抽取。[结果/结论]以小样本文物信息资源为例的抽取实验中取得了F1值0.91的结果,同时对信息资源中未标记的时间与位置关系有着较好的发现效果,为文物信息资源中知识发现的有关研究提供了切实有效的数据获取方法。
[Purpose/Significance]The relationship between entities can be regarded as the basic carrier of knowledge.The extraction of entity relationship from unstructured data is the main problem to be solved in the development of knowledge.[Method/Process]From the perspective of the research object of unstructured text,entity relationship was extracted by using the Bert-BiLSTM-ATT model.[Result/Conclusion]In the experiment of entity relationship extraction based on small sample cultural relic information resources,it could discover the unmarked time and location relationship to a certain extent,which provided a practical and effective data acquisition method for the related research of knowledge discovery in cultural relics information resources.
作者
彭博
Peng Bo(School of Information Management,Central China Normal University,Wuhan 430079,China)
出处
《现代情报》
CSSCI
2021年第5期87-94,共8页
Journal of Modern Information
