Case: Setup of a Financial Knowledge Platform to evidence adherence to AMLD

After several regulatory investigations and multiple remediation projects, a large European bank wants to make sure that it comprehensively controls the rules. It wants to be able to be able to show how it knows that it has put everything in place that the rules require.

Being able to capture comprehensively how rules have been adhered to for a given product, a service or for all services being provided by a business line. Our clients wanted a more effective way to be able to organise the necessary information that is needed to capture regulatory adherence. Doing so would allow stakeholders to provide their share of the knowledge puzzle whilst making it easy to consolidate all information to answer regulatory adherence questions without doing cumbersome investigations.

We introduced our proprietary regulatory knowledge platform (Fimantic). The platform allows for machine reasoning over regulatory adherence. It consists of a knowledge base, a data model and a reasoning engine. Based on the information being stored in the knowledge base and through the logic in the data model, the reasoning engine can answer questions on regulatory adherence by connecting digitized regulatory text to business facts in the knowledge base.

The following steps provide an overview of the setup of the Fimantic platform for the client's use.

Step 1: Acquiring a Fimantic API key

The semantic data model and reasoning logic run on a separate Fimantic server. It can be accessed directly through Application Programming Interface (API) requests. However, an authentication is required in order to be able to approach the server. This so-called API-key allows the server to identify who (i.e. what system) is making requests. Once the API key is obtained, the user (or a system) can make requests such as 'Get all articles in AMLD' or 'Get all entity facts needed to determine AMLD relevance'.

Step 2: Setting up a (local) knowledge base

Clients are not allowed to store information in Fimantic's databases. However, in order to be able to store relevant knowledge, a client's knowledge base should be opened. The knowledge base can capture information in a basic (open type data) format, called RDF standard. This knowledge base can be setup in a cloud environment (e.g. AWS or Ms Azure) or at a client's internal server environment. Now, the client can store information in its knowledge base. In the knowledge base, the client will capture the information needed to determine (for example: 'Post that ABC Bank is a credit institution'). Now, the Fimantic platform can be approached with the available knowledge stored in ABC's local knowledge base. This makes it possible to ask the Fimantic platform: Get all AMLD rules applicable to ABC Bank to retrieve a list of all articles applicable to credit institutions.

Step 3: User interfacing

Although systems can already communicate with each other to exchange regulatory data. It is hard for humans to read the data strings that are being send and retrieved. A general user interfacing (or GUI) makes it possible to present information in a human friendly form. It also makes it possible to easily send the right information storage instructions. Web-browsers (e.g. Internet Explorer, Firefox, Chrome) are ideal for running those user interfaces. For Fimantic numerous user interfaces have been built through which users can get all kinds of analytics, retrieve information or store new information.

Step 4: Setting up user profiles

Allowing different users to do different things, get access to different data requires the setup of user profiles. The user profile determines what user interfacing screens are available, and, together with the API key what information can be retrieved and stored. It also facilitates interactions between users to allocate tasks to specific users.

Step 5: Start using the knowledge layer

The organisation could start using the knowledge layer.

With minimal IT involvement, the client has improved its regulatory knowledge capabilities. This makes it possible to easily store all necessary information in a machine readable format that allows to generate analyses over regulatory adherence status. It also enables the bank to find back how regulatory adherence was achieved. When changes occur, automatic alerts signal where things need to be updated.