Knowledge graph construction from text

Abstract

Being able to organize the massive amount of unstructured data that is produced every day would make analysis simpler and uncover trends and linkages that might not otherwise be seen. Open Information Extraction (OpenIE), a popular tool, extracts semantic triples (Subject -> Relation -> Object) from texts. However, it could lead to ambiguity during semantic triple extraction in information extraction. Pronouns can be thought of as apart from their subject. This project aims to resolve the aforementioned ambiguity by integrating OpenIE systems with Coreference Resolution, thereby allowing the extraction of relations between entities across the entire document. Additionally, new OpenIE systems will be explored and hyperparameter tuning will be done to find the best model.