Efficient query retrieval in Neo4jHA using metaheuristic social data allocation scheme

Abstract

Large amount of data from social networks needs to be shared, distributed and indexed in a parallel structure to be able to make best use of the data. Neo4j High Availability (Neo4jHA) is a popular open-source graph database used for query handling on large social data. This paper analyses how storing and indexing of social data across machines can be carried out by placing all the related information on the same or adjacent machines, with replication. The social graph data allocation problem referred to as Neo4jHA allocation has proved to be NP-Hard in this paper. An integration of Best Fit Decreasing algorithm with Ant Colony Optimization based metaheuristics is proposed for data allocation in a distributed architecture of Neo4jHA. The evaluation of the algorithm is carried out by simulation. The query processing efficiency is compared with other heuristic algorithms like First Fit, Best Fit, First Fit Decreasing and Best Fit Decreasing found in literature. A Skip List index was constructed on Neo4jHA of every machine after the implementation of the proposed allocation strategy for enhancing the query processing efficiency. The results illustrate how the proposed algorithm outperforms other data allocation approaches in query execution with and without an index.