Using OLAP queries for data analysis on graph databases

Abstract

Graph Databases (GDB) are an alternative to traditional Relational Databases and allow a better scalability for the system, in addition to representing highly connected data in a more natural way. GDBs also support different kind of network analysis, such as cen- trality measures and community detection algorithms. Despite this, there are still no tools available in the market for multidimensional analysis in graphs, such as existing OLAP systems that operate on Relational DBs. In the academic field, there are some framework proposals that aim at the construction of a multidimensional cube composed by aggre- gate graphs, which are obtained from the combination of vertices and edges of the original graph, according to the dimensions and measures being analysed. However, most part of the researches in this area are focused on the OLAP analysis for homogeneous graphs, while the works dedicated to heterogeneous graphs require an intermediate data model in order to execute the multidimensional analysis. This project proposes a system to execute OLAP queries in a Graph Database without the need to generate an intermediate data model to do multidimensional analysis on heterogeneous graphs. The proposed system is able to answer OLAP queries using aggregate graphs obtained from the original graph, as well as execute analysis about the topology of the graph. In this work, we present experiments showing the effectiveness of the system to answer the analytical queries and some qualitative comparisons between the proposed system and existing solutions.