Knowledge Graph Generation for Research Articles

Abstract

The increasing amount of web data has made accessing and processing information efficiently challenging. One solution is to transform the unstructured data into a machine-readable structured format like a knowledge graph. This paper addresses an information retrieval system that enables users to formulate queries in natural language and obtain pertinent information from a knowledge graph specific to a particular domain. Understanding any unstructured data is tougher than structured data. Inferring knowledge from any research article is difficult for naïve users.  To resolve this, we propose to create a knowledge graph for the same. Our system utilizes natural language processing techniques to analyze user queries, generating SPARQL queries to retrieve pertinent data from the knowledge graph. We leverage state-of-the-art knowledge graph models to assess the system's precision, recall, and F1 score. The result shows that the proposed model can effectively retrieve relevant information with an average precision of approximately 95%.