摘要
经过多年的地质勘探,油藏地质领域积累了丰富的勘探成果以及地质知识。为了实现油藏地质领域知识的共享、传播及对知识进行有效的管理,油藏地质领域知识图谱成为了一种必然的选择。针对目前油藏地质领域知识图谱在实际构建中存在的已有本体不能直接构建知识图谱的模式层,抽取得到的多个实体指称项对应于同一个事实对象,难以从抽取出的多个属性值中得到最优属性值等问题,通过对构建领域知识图谱的方法和技术的研究,该文提出了一种以领域内的业务活动为核心的领域本体构建方法,并且依据此方法构建了油藏地质领域本体。改进了传统的孪生循环神经网络模型,解决了油藏地质领域实体对齐问题;针对不同类别的属性设计了不同的属性值融合方法,实现属性值的最优。
After years of geological exploration,rich exploration results and geological knowledge have been accumulated in the field of reservoir geology.In order to realize the sharing,dissemination and effective management of knowledge in the field of reservoir geology,the knowledge map of reservoir geology has become an inevitable choice.At present,the existing ontology in the actual construction of reservoir geological knowledge map cannot directly construct the mode layer of knowledge map,and the extracted multiple entity references correspond to the same fact object,so it is difficult to get the optimal attribute value from the extracted multiple attribute values.The method and technology of domain knowledge mapping are studied,and a domain ontology construction method with business activities as the core is proposed to realize the construction of reservoir geology domain ontology.The traditional twin neural network model is improved to solve the problem of entity alignment in the field of reservoir geology.Different attribute value fusion methods are designed for different types of attributes to achieve the optimal attribute value.
作者
文必龙
薛广有
WEN Bi-long;XUE Guang-you(School of Computer and Information Technology,Northeast Petroleum University,Daqing 163318,China)
出处
《计算机技术与发展》
2021年第12期204-210,共7页
Computer Technology and Development
基金
中石化科技攻关项目(33550000-20-ZC0613-0098)
中石油科技攻关项目(JDYT-2020-JS-50311)。
关键词
油藏地质领域
领域本体
知识图谱
循环神经网络
实体对齐
属性值融合
reservoir geology
domain ontology
knowledge graph
recurrent neural network
entity alignment
attribute value fusion
