Research On An Efficient Approach Of Graph Data Extractin

Along with the coming of Big Data Time, traditional relational database cannot satisfy people’s demands. Graph database hasmany advantages whendealing with a large amount of complex, interconnection and low structured data. The research focuses on how to migraterelational database to graph database.The relatively popular ETL (Extract-Transform-Load) technology currently is not very suitable for translating relational data to graph data and the conversion efficiency is low. So this paper proposes an efficient data migration algorithm based on node merging idea.This paperproposes adata extraction methodbased on the traditional ETL technology, with the research of dataextraction problem from relational database to warehouse and the general extraction technology from relational database to graph database in order to make the data migrationmore efficient. The main content of this paper includes the following aspects:Firstly, analyzing the NoSQL and discussing the principle ofthe relationship transferringto a non-relational database, and existing data extraction method, finally this paper summarizes the problem ofexisting data extraction method, such asthe low conversion efficiency.Secondly, according to the problem mentioned above, a migration algorithmfrom relational data to graph database onnode merging idea was designed. Firstly, classifyingthe relational data tablesaccording to the relationship type into three categories: the relationshipof1:1,1:n and m: n. Then at the relationship tableof1:n, combiningthe same foreign key element as a node, and integrating the other relational tablestoobtain two containers, which is node information container and relationship information containers. Finally, achieving graph data structure through the API function in graph database. The data structure is integral computed by this algorithmand the efficiency is higher than traditional algorithms.This paper implemented a prototype system on the open source of Neo4j system finally, and on which carried on a series of experiments.The experiments results showed thatthis algorithmcan greatly improve the efficiency of data migration compared with these traditional algorithmswithout merging node information, and with the increasing of thetuples with the same foreign keys in relational data, the transformefficiency from relational data to graph data will also increase.