Ontology Mediated Document Retrieval for Exploratory Big Data Analytics

An efficient mechanism for managing and retrieving relevant documents from heterogeneous data sources can enormously enhance the data analytics process. Semantics-based information retrieval approaches have emerged recently to use implicit knowledge for such purposes. Conventional search engines generate results by utilizing the intrinsic linkages included within the underlying data model (with minimal human involvement). However, it is equally essential to consider the necessity for retrieving documents in a private data environment based on domain knowledge. In this paper, we describe IIIT Bangalore’s contribution to building a full-fledged datalake framework founded on semantics-based approaches to host and analyze the data in various government sectors. This paper elaborates (a) an ontology-based big data management system/architecture, (b) a distributed heterogeneous data storage, and (c) a document retrieval solution in which an ontology acts as an intermediary to bridge the gap between what the user wants and what search retrieves. The proposed architecture is found to be promising based on the initial feedback received from the end users of the system.