Query Based Graph Data Reduction Algorithms and Application in Education

Abstract

Graph is a commonly used data structure to store large relational data in today’s education networks. With the growing demand for storing and processing large graph data, graph data compression is becoming more important. By reducing large graph data itself, we will also be able to reduce memory space, processing time and transmission cost. While most existing compression methods compress general graphs by generating an encoded representation, we propose query based graph reduction algorithms. Query based graph reduction algorithms, by taking advantage of structural properties of graph and edge weights, compute reduced graphs that preserves necessary information to answer specific queries through disposing irrelevant nodes and edges. We study graph reduction algorithms based on two typical queries: shortest path queries and minimum spanning tree queries. In this paper, we illustrate our algorithm in detail, provide proof for the correctness of the algorithms and show estimation of their reduction ratio on actual graphs.