Not all data is created equal

Structured vs unstructured data: What you should know

Two very distinct types of data are used today to enhance business intelligence: structured and unstructured. They are both important, but they are not the same. 

Structured data is quantitative data that is easy to decipher. Developed by IBM in the ‘70s, Structured Query Language, or SQL, is an open language, used to query, add to, remove from, and change data in a Relational Database Management System (RDMS). Users can easily search, input and manipulate data. The benefits of using structured data are obvious: ease of use and accessibility. The search button at the top of every database is always very useful, so long as you know exactly what you’re looking for. Databases are full of information that lacks direction, nuance, and context. You might get the dates and numbers relating to specific events. You might get some of the information you need about your subject, but you will never get the complete story, no matter how elaborate the database is. 

Awash in a sea of unstructured data

Today, we find ourselves amidst a sea of data, untamed, unregulated, unstructured. This is data that comes at us in a variety of contexts, and an even bigger variety of formats. Just think about all the news, stories, blogs or articles you can find with a simple search online. And isn’t the prospect of having to weed through all that information a little bit intimidating? Indeed, the task of navigating the world of unstructured data can feel daunting. It can be tricky, tedious, and time consuming, but the rewards are immeasurable. Where structured data provides you with all the information you’re looking for, unstructured data can introduce you to information you didn’t even know was available to you, or valuable and significant to your original query. Just think of all the information you’ve gained inadvertently while browsing the internet, just because some random algorithm saw fit to share it with you based on some of your viewing habits. 

Adverse media is not list-based

When it comes to financial crime risk management, teams conduct due diligence using both structured and unstructured data. Sanctions, watchlists, Politically Exposed Persons (PEP) screening are typically list-based and structured. These screenings are foundational to the detection, mitigation and management of financial crimes. This information can be searched easily within a database.

Adverse media analysis, on the other hand, is much more difficult to tame, because it is largely unstructured data typically found in the public domain, such as articles, blogs, government records, online forums, social media posts, video and audio files, and so on. Unstructured data is difficult to collect, process and analyze, but in today’s world, regulators increasingly expect AML and compliance teams to integrate adverse media within their AML/KYC processes. 

AI-enabled analysis

With billions of historical records already available online and millions more being added daily, there is no shortage of sources from which to find adverse news. This is where artificial intelligence comes into play, specifically the use of Natural Language Processing (NLP). Valital’s cloud-based AI-powered SaaS solution is not a screening tool so much as it is an analysis tool.  With deep and machine learning of millions of misconduct-related articles at its core, Valital’s advanced AI models scour the internet in real-time, meaning a search is conducted at the time a name is submitted.

Enhanced filtering - Valital returns more relevant adverse information so clients don’t need to do the filtering themselves. This is a huge time saver.

Accurate classification - Valital sorts the adverse results into classification pillars that also responds to our clients’ need for ease of use.

Relevant analysis - Valital’s identity matching model, source credibility, relationship extraction and risk assessment of individuals and organizations contribute to the quality of the analysis we provide.

Built-in keyword search - Our proprietary AI models are continuously trained and optimized, allowing users to input minimal data to get maximal results.

And, most important, all of this is delivered in minutes. 

Different but complementary

The tedious, onerous task of sifting through innumerable sources that would take hours if not days to complete manually can be performed in a matter of minutes.  The information you get from a straightforward database can now be contextualized, with added dimension and data enrichment. 

This clarity goes a long way toward helping teams make better, more confident decisions about the people and organizations they choose to do business with.