Abstract

Disinformation, misinformation, mal-information, and "fake news": four concepts that are often used interchangeably, and are still in process of disambiguation in academic literature, closely linked to digital adoption and literacy. Together, these issues have become a strategic topic in IS research, one with urgent needs for digital innovation, along with more agile professional practice in combating these trends. They are also touching all facets of individuals, organizations, and society, and therefore require a transdisciplinary perspective. This TREO Talk Paper is concerned with disinformation against politicians, which is disrupting democratic processes.Given the complexity of dealing with false claims, especially in highly fluid and divisive political debates, political staff needs better tools to address false claims. There is growing hope that Artificial Intelligence (AI), especially since the advent of Generative pre-trained transformers (GPT) models, may enable political support teams to fight disinformation against politicians. We focus our attention on how to leverage "semantic technologies", including ontologies, Knowledge Graphs, and semantic reasoning. We explore how they could help GPT disambiguate disinformation, merging into Graph Neural Network (GNN) models to complement Large Language Models (LLM), e.g., GreaseLM. Other digital artifacts from this research are also considered: general disinformation ontology, digital platform to test detection tactics, and governance guidelines for monitoring the emerging of new campaigns.

Download

Share

COinS