Proposal Description

Public reference information is needed for businesses and authorities to ensure safety and efficiency of business process. Does this business still exists and which is its address so that it is safe to contract with it? What are commonly shared key reference information (identifiers, location, activity) to support risk analysis?
For these reasons, authorities have supported the building of public reference information. In 2014, the G20 has created the LEI which now covers 3 million uniquely identified legal entities worldwide, mostly financial, which their relationships. In Europe, implementing regulations to open data directive (implementation 2024) defines a subset of reference information (incl. activity and legal form) as a high-value dataset to be made publicly accessible by public sector bodies through API and bulk download. More recently, Digitalisation directive to be implemented in July 2028 adds key information on relationships.
However, information is still scattered in different data basis, with different identifiers and rules for updates. It is frequently uncomplete, in particular when it comes to parents’ relationships. To facilitate search and improve data quality and completeness, several actors interested in reference information have developed AI approach. Following a survey by its members, the Regulatory Oversight Committee (ROC) has now engaged in exchanges of information about the projects they develop. The objective is to share success and failures, foster possible synergies and facilitate industrialisation and scaling up. This shows that interest concentrates around a small number of objectives and tools: 1) the matching of data basis to improve data quality 2) the extraction of unstructured data on parent relationships mostly from financial statements 3) Chatbot to support data basis search.

The following projects will be approached for presentations (with four selected depending on availability):
• Bank of Spain “Regulatory Oversight Committee exchange of information on the use of AI: survey, workshops and follow ups”
• Bank of France - Data Enrichment and quality improvement of the LEI Database (parent’s information).
• Bank of Spain - Corporate Annual Report Information Extraction Project.
• GLEIF - Unlocking Relationship Insights from Annual Reports – live demo
• Banco de Portugal “Enhancing Entity Data Integrity: Automating LEI integration across internal and external reference databases”
• Bank of France – Conversational assistant / Graph Rag for LEI search.
• Bank of Spain - Matching the National Database with the GLEIF Golden Copy to Assess Corporate Name Consistency (Translations and Transliterations).
• GLEIF - GLEIF AI Search – live demo.