Research And Implementation Of Parallel Analysis And Mining System Based On Graph OLAM

With the rapid development of society,the data scale of all walks of life is increasing.How to extract and mine massive data information efficiently has become a research hotspot in recent years.Among all kinds of data types,graph is widely used in complex relational dataset modeling because of its rich topological structure and dimension information.In the field of large-scale network data analysis,Graph OLAP technology and Graph Mining technology have become important technologies to provide decision support based on graph data.However,the development process of the two is very different,and the existing research is difficult to unify the two fundamentally.In view of the current research and development status,in order to eliminate or reduce this difference and realize the close combination of Graph OLAP and Graph Mining technology,the following contents are studied in this thesis:1.Research and design of Graph OLAM large-scale multi-dimensional network mining framework:it improves the theoretical framework and related concepts of Graph OLAM,and improves the problem that Graph OLAP technology and Graph Mining technology are difficult to combine to a certain extent.Unify and refine the operation mode of Graph OLAM,design metadata management and graph data cube storage model for sequential network,and build a complete Graph OLAM large-scale multi-dimensional network mining framework from two aspects of logical architecture and practical operation.2.Insurance recommendation algorithm and framework based on Graph OLAM Technology:in order to solve the sparse problem of insurance network and improve the network topology,this thesis applies Graph OLAM technology The related technologies in OLAM framework combine the operations of roll up and drill down,aggregation and network representation learning,show the implicit connection relationship of sparse network through high-level aggregation network,and design RU-GOLAM algorithm,so as to effectively solve the recommendation problem of insurance sparse network,which has great practical significance.3.Parallel analysis and mining system based on Graph OLAM:in order to help users quickly use network analysis and mining algorithm for network analysis,so that users do not need to accumulate massive knowledge can also apply this system to solve practical problems,combined with the existing parallel computing framework,this thesis builds a parallel analysis and mining system based on Graph OLAM,which provides users with graphical big data cloud computing application service solutions.