The Domain Knowledge Graph Construction Of Scientific Research Management

With the continuous advancement of industry informatization process,there are more and more data under the scientific research management platform,which comes from different source.It is difficult to comprehensively utilize these data and play its huge value only relying on the existing relational database.However,Google knowledge graph can solve this problem well.Not only can it realize the integration of multi-source data under the scientific research management platform,but also provide data analysis and decision support for scientific research management workers,thus improving the scientific research management level and knowledge service ability of the platform.Therefore,this thesis will make an in-depth study on the multi-source data available inside and outside the platform,and make full use of each data source to extract domain knowledge,and complete the task of knowledge modeling in a top-down way,so as to realize the construction and application of knowledge graph in the domain of scientific research management.In the top-down knowledge modeling task,there is a lack of reusable ontology and expert guidance.In order to effectively make use of each data source,this thesis first implements a semi-automatic ontology construction method,which not only makes full use of the information of relational database with the help of semantic recovery strategy,but also proposes to combine the results of the upper word discovery algorithm based on word vector with the results of Wikipedia concept hierarchy extraction algorithm based on improved VVG.In this way,the domain ontology can be constructed effectively.Then,the constructed ontology is applied to the task of entity knowledge extraction of relational database based on Karma tools and the task of scientific research text relationship extraction based on concept entities,which can effectively extract entity knowledge from multi-source data on the basis of ensuring the consistency of knowledge,so as to better complete the task of knowledge modeling.In the specific deployment implementation,due to the large amount of data and complex construction process.In order to ensure the efficient operation of the whole process,this thesis implements the rapid deployment of the construction process based on the container,and proposes an automatic entity knowledge extraction method to further improve the efficiency of knowledge modeling.Finally,with the help of neo4j graph database to store the knowledge map,the efficient query,management and application of knowledge can be realized on the basis of improving the efficiency of data import.Through the above design,implementation and testing steps,this thesis completes the construction and application of knowledge graph in the domain of scientific research management with the help of multi-source data,which can not only effectively integrate,manage and apply multi-source data in the domain of scientific research management,but also provide better knowledge services for scientific research managers,thus improving the efficiency and level of scientific research management and giving full play to the value of network big data.