January 18, 2023 | Tim Ward
Estimated reading time: 5 minutes
Classification is King
There are many different classification systems that are used to group and organize things based on their characteristics. At school, in Chemistry lessons, we are taught about the periodic table that arranges elements into rows and columns based on their atomic structure and chemical properties.
In Biology, you may have been taught the Linnean system that organizes living things into hierarchal classifications with the most general classifications at the top and the most significant classifications at the bottom.
And if you ever ventured into the school library, then perhaps the books were arranged in accordance with the Dewey Decimal System. It organizes books into 10 main classifications with each classification further divided into more specific sub-genres.
Why use a classification system?
Classification helps organize and simplify complex domains. These systems provide a shorthand for describing individual items and allow interoperability between diverse people, places and systems.
Classification helps us find what we are searching for and allows us to compare items that might be similar or relevant.
CUBE’s classification system
At CUBE, we quickly realised that the world of regulatory content would benefit from a clear and comprehensive classification system.
And so, we created the CUBE ontology, a hierarchal classification system with six high-level levels:
Each high-level classification breaks down into a number of levels of sub-classification.
Every regulatory section of every regulatory book is classification using machine learning using concepts included in the CUBE ontology. The same is true for each horizon-scanning article.
One key point is that this classification doesn’t just use word and phrase matching. The machine learning algorithm is able to go one step further and derive the semantic meaning of an article and ensure this meaning is linked to the relevant ontology concepts.
Accurate classification is a game changer for regulatory change management. In the same way, you can easily find an Agatha Christie crime novel in a catalogued library, you can now easily find articles covering ESG, Individual Accountability or Insider Trading within the regulatory universe.
The story doesn’t stop there. The best classification systems evolve. The regulatory landscape is not static. Innovation brings new types of products, new ways of doing things and different types of compliance risks.
The classification system has to reflect these changes, but the art here is to decide which changes are significant enough to add to the system. If every change is added, the system becomes bloated, confusing, and unusable. But missing out on a key development leaves a big hole in the classification system which eventually will result in reduced adoption.
CUBE has a team of experienced regulatory experts and analysts that understand the ebb and flow of regulatory change management. Managing the ontology is a team sport with every proposed change peer reviewed and discussed at length so that it truly earns its place. This is important because it is not ideal to make structural changes later.
By the time the changes are released, companies and users will waste no time integrating them into their own systems and processes.
How is classification integral in the regulatory change management process?
Classification is integral to the future automation of regulatory change management. If we are to get the benefit of using technology to cut down the time it takes to collect, process and act upon regulatory data, then we have to define and understand a common language between humans and machines. We have to have confidence that the classification process is accurate to avoid any compliance gaps.
A missed regulation, obligation or key piece of guidance could result in reputational damage, regulatory action, fines and lost business. The stakes are high and so it is important that appropriate monitoring is in place to ensure that data quality and integrity are being preserved throughout the classification process.
Thankfully, a large part of this monitoring can also be automated. The operational dashboard can be used to ensure transparency and provide further confidence that everything is working as expected.
What makes a great classification system?
Classification systems can also be designed to easily interoperate with other classification systems. It may not be desirable to have a single uber system that describes every concept in the universe.
Instead, systems should be designed to have a sensible defined scope that can be effectively managed by a nominated authoring team. Once understood, these systems can be integrated with other classification systems and used in concert.
Think about the system that manages your projects and tasks. Tasks may be organized in multiple ways. They will certainly have a Status – To Do, In Progress and Done. They may also be assigned to a specific project or may contribute to a business objective. There may be a list of task types. And some of these classifications could be organized into a hierarchy.
When presenting task information, you may choose to group and sort these tasks depending on the audience. Your team may just be interested in what needs doing today or this week, whilst senior management may be more interested in how their strategic objectives are being met.
We can take the same list of items (the tasks) and group them based on the audience using these different classification systems. The same is true for regulatory data. In addition, to CUBE’s ontology, many clients choose to integrate their own taxonomies too. Labeling regulatory data with a specific product, service, business unit or risk rating. The two classifications work hand in hand, serving different purposes for different use cases. Sometimes, these taxonomies are managed at the organization level, and sometimes at the department or business unit level.
The flexibility of implementation is an important characteristic of Enterprise compliance technology. Change can be difficult and costly to implement in organizations with tens of thousands of employees in diverse locations and so it is a huge advantage if the technology can be shaped to work with existing processes and methods.
Being able to integrate an existing customer taxonomy (or indeed several of them) is a great example of how implementation can be accelerated, and adoption increased when implementing a new solution.
Great classification systems stand the test of time, lasting tens or even hundreds of years. They become second nature to use and provide a common language when describing a complex domain. CUBE’s ontology is here to stay after ten years of development and evolution.
Why not find out how it could help you manage regulatory change management easier and more accurately?