A Novel Approach to Generate Knowledge Graph from Unstructured Text

Abstract

Intelligent Tutoring System (ITS) is a technology that uses Artificial Intelligence and Machine Learning techniques to provide an adaptive learning environment. ITS records learner’s interactions during the study process in order to provide adaptive real-time feedback. ITS use Domain Models (DM) to map real-world content in a structured format to provide content and feedback to learners. The DM is generated manually by domain experts which becomes a time consuming and challenging task. DMs are used to map real-world concepts and their relationships in a structured format. Knowledge Graphs (KG) are naturally programmed graph structures that are used to depict relationships between entities in a graph form. Therefore, KGs can be used to generate DMs for ITSs. The automatically generated KGs can be used in different fields of studies that includes Question Answering, Financial Market and Semantic Searching. Different probabilistic techniques have been used to build KGs for different purposes. In this research, we propose a new technique for the construction of a KG that will utilize the linguistic structure of English language to generate relationships between entities. The proposed technique is categorized into six stages. These stages extract entities and their relationship from unstructured textual data using machine learning technique and stores them in a database which can be further used for generating questions and answering queries using knowledge graph. Neo4j graph database is used to store data into the database and Java programing language is used to develop the scheme.