Domain Specific Fusion of Unstructured Text for Situation Understanding (Poster)
Eric-Olivier Bosse, Julian Falardeau, Isabelle Prevost, Elisa Shahbazian, Olivier Labonte · 2019
This paper presents the initial design and the current and envisaged functionalities of a novel tool for information extraction and reasoning from open source data (OSD), namely, the Open Source Information Collection, Analysis and Reasoning (OSCAR). It has the ability to ingest and process vast amount of OSD to provide situation understanding and decision support about domain specific situations. The data are pre-filtered using a custom created knowledge base (KB) while the information is extracted using the Rule Based Information Extraction (RuBIE), a Natural Language Processing (NLP) and tagging tool. The extracted information is subsequently clustered and transformed into a relation graph of entities of interest. This proof of concept is presented in the context of a use case based on the social crisis in Venezuela in 2019.