The Research Of Multidimensional Analytical Techniques On Heterogeneous Network Based On Meta Path Analysis

A growing number of network analysis techniques are applied to different types of network analysis instances, with more and more complex network research, especially heterogeneous network research meet the real world. Most of the traditional social network analysis is homogeneous network-oriented, focusing on the structural characteristics of the network. But in the real world, abstract network often contains a variety of types of nodes and relations, forming heterogeneous network with rich structure and fertile semantics. Heterogeneous network is not only in line with visual impression of the real world further, but also contains more information than homogeneous network. So the analysis in view of the heterogeneous information network has the potential to find more implicit knowledge which is exacter.As for the multidimensional analysis technology of heterogeneous network, analysis based on fundamental path is an important means of analysis. In heterogeneous network, there may be complicated relation between two nodes formed by relations involved with other nodes and types. This compound relationship can be measured explicitly in meta path. It is of great importance for analysis of heterogeneous network to get hidden semantic information by way of different meta paths. Meta path is formed with node the type of heterogeneous network and the relations between the adjacent node types, which is used to distinguish different semantic similarity between the same node type, so as to understand the specific meaning of network nodes and relations. Meta path analysis has been widely used in a lot of aspects such as child figure extraction, similarity computing and clustering analysis. How to make meta path analysis applied in multidimensional analysis of heterogeneous network? In this paper, the exploration of the following several aspects has been carried on:To the first, focus on the community division in heterogeneous network, purpose a novel algorithm of community division based on the weight in meta path, which has property result in the analyzing test.To the second, describe the Graph OLAP model on heterogeneous network, including the feather, the operation, the strategy of storage and the materialization. At the same time, design a serial kind of test on several different kinds of materialization strategies focus on the function and the relevance of the real scenes.To the third, focus on the non-literature data heterogeneous network, achieve the multi-dimensional analysis target on the roll-up and drill-down operation of special dimension, and finish the experiment of predicting the users’identifier.