The FF Poirot Project IST 2001-38248

Positioned in the context of Semantic Web endeavour, the project seeks to build computable knowledge resources, ontology of financial forensics, from financial and legal expertise in detecting and preventing fraud in VAT processes, securities exchange, investment and banking and insurance services.
It explores the enabling ontology technology, tools and methodology to represent, store, model, mine, integrate, and use ontology of financial forensics. It focuses on the adaptation of database principles and methodologies for modeling large-scale ontology-based knowledge systems.
It will propose methodology and guidelines in ontology engineering and processing with respect to the technology developed, research conducted, and fieldwork experience in ontology modeling in the consortium.
Since human expertise and knowledge are expressed, structured and conveyed by languages, ontology engineering has close relationship with language engineering. In recognition of this technological kinship and EU application context, the consortium will build up multilingual terminological bases as 'medium' for the concept repositories and their applications. The languages under attention are English, French, Italian and Dutch.
Resultant knowledge resources, methodology and technologies can be exploited as part of information systems or processes to detect, investigate or preventing instances of actual or attempted financial fraud. For example, investigative and monitoring bodies can benefit from intelligent information retrieval on the conceptual level, with the FF POIROT ontology. Financial professionals, such as accountants, auditors, banks, insurance agencies, government departments, regulators and financial experts can use such ontology as an authoritative concept base to make sense of their data cross domains, systems and languages.
The consortium will demonstrate the added value of its ontology contents and technology with showcase applications based on the real user's requirements.


RTV Role

The Artificial Intelligence Group at Tor Vergata University (RTV) is working jointly with Consob and Knowledge Stone (KS) at the design and implementation of the Financial Fraud Ontology as requested by the Consob Consortium. RTV is also taking care of the construction of fraud detection rules able to detect financial fraud web site by applying automatic reasoning techniques to the implemented ontology and analyzing the web using Information Extraction (IE) and Natural Language Processing (NLP) methodologies.
In order to build the ontology an NLP architecture is being set-up: the final architecture will be able to deliver a specific ontological content related to the domain ontology using semi-supervised techniques. For more info about the architecture building framework, visit the Language-centred Ontology Learning page .
The ontological content, together with the detection rules, is expected to be released for the end of October 2004.