As per the stage-1 proposal abstract, Factiverse is addressing a high-impact opportunity:
Factiverse automates detection of fake news and misinformation with cutting-edge AI and NLP, and a tailored fact-checking search engine. The innovation is based on the development of a hierarchical deep attention network that processes a high volume of content from news media and social media to validate the credibility of claims. Contrary to existing solutions, which only apply word-level self-attention, the model learns the latent aspect embeddings for classifying false claims by applying hierarchical attention. This allows for capture of salient vocabulary and complex structures at the document level. The attention weights are more effective at detecting the truthfulness of a claim, providing a credibility score and background evidence. The search engine is multilingual and aware of a range of heterogeneous sources.
Addressing both B2B and (in time) B2C segments, the innovation effort underpins Factiverse’s growth plan, expecting annual recurring revenues (ARR) of €m >50 in 5 years.