PREDICTIVE TABLE PRE-JOINS IN LARGE SCALE DATA MANAGEMENT SYSTEM USING GRAPH COMMUNITY DETECTION

A computer-implemented method for identifying pre-join operations, when accessing a database of relational tables, based on a usage history and/or a priority needs, comprises creating a graph of weighted edges and nodes, the nodes represent relational tables and edges represent join operations to be performed on the tables, partitioning the graph into a plurality of graph communities based on graph community densities, with a density indicating a number of edges touching a particular node, with the number of edges being greater than a predetermined edge number threshold, with each edge further including an edge weight indicative of a frequency of referencing within a predetermined recent duration of time and/or indicative of urgency of quick access to the corresponding join result within a predetermined recent duration of time, and generating pre-join results based on the partitioned graph communities and graph community densities.