Advanced algorithms for effective compliance

Use case

Classifying and gathering essential information to enable targeted decision-making processes for specific targets

Opportunity

Speeding up the innovation process in financial institutions to provide efficient, safe, sustainable, and inclusive customer-oriented services.

Challenge

Challenge
Simplifying and optimizing the way of working, leveraging search, classification, and clustering algorithms for extracted information (e.g., segments from European regulations, Decrees, internal regulations, etc.) to offer efficiency in the compliance document production process, enabling:

• easier drafting and review of inspection findings
• identification of the most relevant regulations for the specific case
• effective manual classification of findings
• availability of statistical analysis
• immediate access to hard-to-reach information
• classification of main topics.

Solution

A set of regulatory compliance algorithms for searching, classifying, and grouping extracted information, with a prototypical neural model for text classification based on a BERT language model, through which it's possible to achieve state-of-the-art results on various typical Natural Language Processing (NLP) tasks.