NATURAL LANGUAGE PROCESSING IN COMPLIANCE
If you ask most children for the time, they will probably ask Amazon’s Alexa.
Though Alexa, Google Assistant or Apple’s Siri do not presently answer compliance questions, RegTech companies do offer artificial intelligence (AI)-based solutions in a variety of compliance fields. Unfortunately, an AI tool remains a black box for many compliance practitioners, and the resulting controls and governance around the AI tool may seem opaque to them.
This article aims to help practitioners better understand how AI operates by looking at one AI technique commonly used in compliance tools: natural language processing (NLP). The article describes NLP calculations and parameters, and then derives from them elements of governance for the NLP tool.
How NLP works
Wikipedia describes NLP as “a subfield of computer science, information engineering, and artificial intelligence concerned with the interactions between computers and human (natural) languages, in particular how to program computers to process and analyse large amounts of natural language data”.
NLP is a useful tool which enables companies to process written text, to recognise speech and to generate text or speech. Compliance teams use it to read regulations and employees’ emails or to create narratives for alert reviews.
This article’s use case is regulatory tracking. Compliance analysts obtain and analyse regulatory updates from regulators. Filters exist to select only relevant types of updates. Unfortunately, a sizeable number of non-relevant regulatory updates still reaches analysts. This superfluous information distracts firms from analysing the regulatory updates that are relevant and impactful to the organisation.
An NLP tool helps further triage regulatory updates by setting aside non-relevant ones. To do so, an NLP tool uses two main features: standardise text and classify.
Apr-Jun 2020 Issue
Natixis CIB Americas