Corporations need to do more internal investigations than ever before. They need investigations to be able to respond adequately and timely to requests from regulators. And they need to conduct investigations to comply with access requests from customers or former employees.
These requests for information and documentation usually come unannounced and seriously disrupt the daily business routine. With the help of modern eDiscovery technology, they can perform the necessary investigations faster, more thoroughly and accurately at lower costs.
But how does that actually work?
For almost all types of investigations, but especially in legal or criminal cases when you cannot afford to miss anything, "normal" search using search engines and regular search terms is no longer sufficient. Today’s data collections are huge and "just search" hardly solves anything anymore.
The current data sets are simply too complex. There are too many possible questions to be executed and the lists of results are endless.
More advanced techniques utilizing Artificial Intelligence such as Natural Language Processing (NLP) and Machine Learning, therefore, focus on patterns and characteristics. This way, information which would otherwise remain hidden, can be retrieved quickly,
Let's look at specific cases where this type of technology can help.
Investigations for access requests
By means of an access request, a customer or former employee asks you which personal data your organization stores about him or her. You are required to show in a transparent way, which personal data you have stored and how it is used. You must provide insight into:
- What information is involved;
- The purpose of using that information;
- To whom the organization may have provided the data;
- The origin of the data (if known);
- The rights of the customer and to make sure it is clear that the customer has the right to file a complaint with the appropriate authority;
- Whether automated decision-making including profiling are involved;
- Whether you transfer personal data to other countries outside the European Union or to international organizations.
Investigations for regulatory requests
A request for information can come from one of the different agencies. Competition and antitrust violations have traditionally been the main drivers for these request. However, in recent years, regulators have issued similar disruptive requests with regard to financial, bribery, fraud, environmental, healthcare, data privacy, consumer protection, food and drug safety and many other areas.
When the regulator suspects possible misconduct or illegal activities or when they have been contacted by a whistle-blower, they will start by asking broad and open questions such as:
- Who did what?
- Did management know about it or was management involved?
- Has management concealed the irregularities?
- Were there early warning signals and, if so, were they ignored?
- What activities have taken place and when?
- What steps were taken and what was the underlying reasoning?
- Who benefited from the activities?
- Have similar cases ever occurred before?
To prevent is better than to cure, when it comes to corporate misconducts such as fraud, corruption and bribery. But despite increasingly strict regulations and firm compliance programs, occasionally, employees do not follow the rules of your company.
In that case, you obviously want to know as quickly as possible what exactly happened and who might have been involved in the misconduct. You are looking for the "7 Golden W's":
- Who can be associated with the crime?
- What exactly happened?
- Where did the crime take place and were there traces left?
- With what (or how) was the crime committed?
- In what way did the crime occur?
- When did the crime occur?
- Why did the crime occur?
To summarize, there are many types of investigations, conducted by different parties, but all these large scale investigations have one thing in common: in every case, large volumes of data must be collected, analyzed and produced..
By using advanced techniques such as pattern recognition, semantic information extraction, natural language processing and machine learning, the computer identifies the basic elements in a body of text.
This can be people, companies, locations, products, facts, ages and addresses, or more complex patterns such as events, relationships between objects, sentiments or emotions. The computer also analyzes who is referred to in the text and which names, aliases and synonyms appear in the texts.