Embedded Files
AI-based conversational agent for misinformation fact-checking
AI-based conversational agent for misinformation fact-checking
Principal Investigator: Ciprian-Octavian TRUICĂ
Principal Investigator: Ciprian-Octavian TRUICĂ
Funding Agency: OPTIM Research framework (subsidiary project funded through POCU grant no. 62461/03.06.2022, SMIS code 153735)
Funding Agency: OPTIM Research framework (subsidiary project funded through POCU grant no. 62461/03.06.2022, SMIS code 153735)
Overview
Overview
This project addresses the global surge in digital misinformation, which ranges from malicious election interference campaigns to the accidental sharing of public health falsehoods on social networks. While various fake news detection models exist, they lack user interaction. Conversely, existing chatbot solutions rely on superficial keyword searches that frequently deliver shallow or incorrect answers. To bridge this gap, this project proposes an interactive, Conversational AI-based agent capable of operating in both English and Romanian to guide users through rigorous, dialogue-driven news verification.
Key Objectives
Key Objectives
Dynamic Fact-Checking: Interact with users in real-time dialogue to isolate and verify specific claims within user-submitted articles.
Aggregated Scoring: Compute and deliver an overall, intuitive veracity score for the entire article or set of articles being discussed.
Innovation and Impact
Innovation and Impact
The project innovates by replacing basic, error-prone keyword search chatbots with an advanced Conversational AI agent that provides deep semantic analysis and detailed justification reports. Instead of delivering a simple "True/False" flag, the conversational approach educates users through interactive reporting, building long-term digital literacy and reducing the societal harms of viral misinformation.
Publications
Publications
Journals
Journals
2023
2023
Ciprian-Octavian Truică, Elena-Simona Apostol. It's all in the Embedding! Fake News Detection using Document Embeddings. Mathematics, 11(3):1-29(508), ISSN 2227-7390, January 2023. DOI: 10.3390/math11030508 (Q1 Journal) [pdf]
Conferences
Conferences
2022
2022
Christian Chiarcos, Elena-Simona Apostol, Besim Kabashi, Ciprian-Octavian Truică. Modelling Frequency, Attestation, and Corpus-Based Information with OntoLex-FrAC. International Conference on Computational Linguistics (COLING2022), p. 4018-4027, October 2022 (Rank A Conference) [pdf] [link] [poster]
Software
Software
Document Embeddings: https://github.com/DS4AI-UPB/It-s-all-in-the-Embedding
Page updated
Google Sites
Report abuse