The DIEKB project (Disinformation Identification in Evolving Knowledge) is a project funded by the french Ministry of Finance (DGE, France) under the RAPID convention, of which Mondeca is the coordinator in a consortium made up of Airbus Defense & Space and the higher Institute Jean-Nicod (ENS). EKtimo company will support Mondeca in the project for aspects related to information rating, civilian and military scenarios of disinformation. DIEKB aims at developing capacities for detecting conflicts, contradictions, incompleteness and erroneous facts or “misleading statements”. The developed components must provide the following aspects:
- Reasoning capabilities on dynamic facts
- Rating of multimodal information based on the veracity and reliability of the source
- Detection of abnormal changes in a knowledge graph
- A fine-grained approach of the typology of disinformation actions proposal
- The definition of models and rules for detecting disinformation during the data integration data into a knowledge base.
DIEKB aims to develop capabilities for detecting conflicts, contradictions, incompleteness and mistaken facts or “misinforming statements” by combining work in the humanities, automatic language processing, and artificial intelligence (rules-based system and statistical learning).
One of the challenges of the project will be to define a classification of disinformation in a fine way, as well as to capture in OWL language the dynamic epistemic logic representing it from previous works by Institut Jean-Nicod.
Project Output
DIEKB will provide a generic, multi-domain, multi-language, and easily configurable approach to enable its use in both civilian and military applications. The innovation of the project lies in the resolution of scientific and technical obstacles at the intersection of artificial intelligence, progressive knowledge management, and human and social sciences (SHS).
The project had its official launch in June 2019 at Mondeca premises in the presence of the representative of the DGA and all members of the consortium. The project end date is June, 2024.