GRAPHMK-an integrated framework for graph computation

Abstract

The rapid change in technological innovation and end user expectation during the past decade resulted in data being an important area for new research and development. In addition, evolvement of internet of things, social graph and communication networks resulted in creation of large-scale real time, rapidly changing datasets. This led to the creation of graph systems supporting various graph analytics to become popular. Meanwhile, relational databases are used in storing and managing data including most advanced systems where graph analytics was never explored on top of a relational schema-based system. This led to the questions whether graph analytics can be done in a relational environment whilst it is still being blindsided or does relational databases even have limitations to execute graph computations. The relational model where data needs to be imagined in a tabular format rather than a graph format with edges and nodes are inefficient in executing iterative graph analysis. Relational systems will end up creating expensive joints in order to execute such computations. Structured Query Language (SQL) queries are difficult in nature to express graph analysis,though we have downsided, relational systems that are composed of great functionalities which includes fault tolerance, integrity constraints, secure transaction and query optimization etc. This thesis writeup reflects an integrated framework for graph analytics that comprise of three main components. Firstly, a data model that is capable of executing graph computation within a relational environment. Secondly, a query language which can be considered as a data log that helps to execute graph specific computations. Finally, a bolt-on solution which comprise of the above two, that executes sitting within a relational query engine and executes queries created from introduced data log language. The tests performed are evident that bolt-on solution introduced achieved better performances in provided scenarios. Subject Descriptors • Graph Theory - Graph Problems, Graph specific algorithms • Logical Design – Models & Schema • Languages – Data log & Query