Towards an Improved Semantic Model: An Ontological Based Framework for Misinformation Detection

Abstract

The scourge of false information across different real life scenarios in this digital age is alarming and worrisome. While open social networks and artificial intelligence have changed the narration of computing by making data available and accessible, on the contrary, they are aiding and abating misinformation. State of the art techniques such as machine learning models, natural language processing, and recently ontology semantic technology have been geared towards addressing this challenge. However, the promising strength of ontology semantic technology in terms of contextual knowledge and handling concept semantic conflict have not been adequately addressed. In view of this development, this research aims to design an ontological semantic framework for the detection of false information considering the domains of politics, science and health. An existing fake news taxonomy data was reused along with the newly harvested domain based dataset. The ontology, dubbed as iFaIDet is developed based on Noy-McGuiness methodology, and the results are promising. The individual OWL component of the ontology far outweigh the results of three other existing related ontologies. Similarly, the results of seven out of eight metrics of its structural based evaluation are promising. For instance, average population and class utilization metrics show 3.45 and 0.88 against the existing 1,413 OWL ontologies’ average values of 1.34 and 0.54 respectively. However, the ontology is still work in progress as standard rule based misinformation database will be fused into it, along with design collaborative algorithm for top level semantic lexical databases.