An Improved Knowledge Graph Representation for Ransomware Attacks Using LSTM-Based Named Entity Relation Technique and Twitter Data

Abstract

This research aims to examine the challenges associated with ransomware knowledge and develop a knowledge graph of ransomware attacks utilizing Twitter data. Ransomware is a constantly evolving global threat. The three main steps involved in the development of a knowledge graph from informal text are data collection and preprocessing, features extraction, and relation extraction. A ransomware ontology previously proposed has been used in this work for the extraction of the ransomware entities from unstructured data; this is aimed at making it fit the attacks reported on Twitter. The next process is the identification of the existing relationships in the dataset for the knowledge graph construction; the output of the process, which is the developed knowledge graph is evaluated for accuracy using a tracing technique to demonstrate its efficacy.