Research And Implementation On Multi-dimensional Network Analysis System Based On Cloud Platform

With the development of internet information technology,large amount of data burst from various of fields.Since the cost of storage device become lower than before and much business demand emerge,plenty of companies store data on database and data warehourse.How to use the data as a guidance of decision analysts become a hot topic.In the meantime,OLAP gradually come to a valid solution,which can support excellent information decision and efficient analysis on multi-dimension and multi-granularity.Therefore,OLAP technology has been applied in many business applications.The popularity of internet makes social network,biology information network and other fields rise immediately,with the appearance of many vast multi-dimensional hetergonous network.However,traditional OLAP only focuses on record type data,how to apply it on analysis of multi-dimensional network and valid information decision is a major challenge.Recently,new research proposes Graph OLAP concept according to the structure of graph and improve the information network model,it is still with limit power and in a first step.In this paper,according to the major problems,the main research can be summarized as follows:(1)We redefine the concept of Aggregate Network and Relation Path Set on the basis of multidimensional hete-rogeneous network,and propose a new Graph Cube Model described as P&D Graph Cube under the guideline of relation path set.(2)Based on Spark,we design and implement the efficient cube computation algorithms.Path related Materialization can be seen as an extension that enriching the relations in the network.And the method is firstly raised up to now.On dimension-related Materialization,we employ the inverted index technique to make a trade-off between space and query efficiency.(3)We redefine the Graph OLAP operations by new cube model,besides,we conclude the different query modes and math model according to the characteristics of their own.(4)We design and implement OLAP Anaylsis System based on the cloud platform,which ultilize spark to compute the cube materialization.Also,we conduct extensive experimental evaluations based on real datasets.The experimental results demonstrate that materialization algorithm is effective and scalable.