Pattern-based visual subgraph query formulation meets query processing

Nowadays, graph is used in many applications because it provides natural ways to represent data. But not all the users who need to work with graph database are IT-experts to use complicated system for querying graph database. PRAGUE is a framework that provides user a friendly GUI and efficient System Response Time (SRT) to query graph database. PRAGUE interleaves visual query graph formulation and processing by exploiting the latency offered by the GUI to remove false matches and pre-fetch partial results. By using SPIG to store all possible connected subgraph of the query, the system can handle both exact and similar subgraph matching. To make framework more useful and friendly, there is a need to enhance it to allow user build queries from given patterns. PRAGUE cannot handle pattern-based query processing efficiently because it cannot simulate the process when user drags and drops in a pattern, which leads to very large SRT. Moreover, the query formulated from pattern is often large, which leads to explosive increase in size of SPIG set and make PRAGUE inefficient in term of memory. By introducing the concepts of pattern-based view of graph, in this report, we overcome the system response time limitation and memory limitation of PRAGUE when processing queries formulated from pattern. Moreover, inspired by the need of selecting the best set of patterns to display on the very limited space of GUI, we propose a novel algorithm to generate patterns by mining from graph database, and then select the best subset from that in order to display on GUI.Bachelor of Engineering (Computer Science