Analysing vast – and ever-growing – volumes of data necessary to deal with litigation, investigations and/or a regulatory request can be a challenge for legal teams.
Technologies such as artificial intelligence (AI) and machine learning can provide lawyers with invaluable support when it comes to data analysis. Together, tech-driven solutions are helping clients, and their advisers find the relevant data required within documents to successfully carry out their work faster, and at lower cost, than ever before. Technological tools are continuously developing, increasing their analytical power and ability to reflect the ever-evolving way we generate, transfer and store data.
But how can long-term value be realised when data analysis is being delivered to support litigation? And – crucially – where will the processes, and technologies go in the future?
Currently, the data analysis and management process is centred on culling and analysing data post collection after it has already been shared with legal teams. Here, a careful balance must be struck. Lawyers or investigators need to be thorough. They are often navigating vast quantities of information and need to be confident they’ve taken all the reasonable steps and relevant data sources into account; whilst also managing time constraints and limited resource.
Individuals sifting through threads of conversations, across multiple platforms that can go back for years to find one (although vital) document will exacerbate existing time constraints. The challenge is therefore to find efficient ways to gather all relevant data sources; structure and standardise it so that easy, consistent analysis can be applied; and in turn intelligently identify what is and isn’t likely to be relevant.
There are a host of solutions which are currently helping to do this through powerful data analysis software platforms. For example, ‘email threading’ works by identifying relationships between emails in a large data pool and subsequently gathering them together as part of one single, combined conversation, often enabling quick subsequent review.
‘Conceptual searching,’ which refers to automated searching that uses an analytics index to derive meaning, allows investigators to use a sample block of existing text to find other documents that have similar conceptual content, rather than simply looking for fixed words or phrases.
And ‘continuous active learning,’ which refers to models which learn and evolve based on the input of human reviewers can be applied to new data as additional sources are added, while retaining previously learned information. This approach ensures that lawyers’/ subject matter expertise and human insight, is applied in the most cost effective and timely manner to potentially huge document volumes.
While this demands a degree of time investment from the case team, the overall time saving can be significant. Once up and running, the technology can run entirely independently – extrapolating from the human’s behaviour and continued inputs to generate its own recommendations on what is likely to be of interest, and what is not.
These tools and data discovery processes are delivering outstanding results today. But there are areas where we think we’ll see future change and development. One is data governance.
As mentioned, today’s discovery processes are overwhelmingly centred on managing data once it has been collected. It’s common that businesses don’t always fully understand where information is kept in their operations, what it refers to and who owns it. Consequently, delaying the start of the data analysis element of the matter, as the data has to be first identified and mapped prior to commencement of any collection exercise.
A key step to delivering more efficient processes will be encouraging more stringent data management at the pre-collection stage. This could require an initial cost and time investment from businesses but makes it far easier to source information when the time eventually comes.
More widely, the challenges of data management are growing – driven by changes in how individuals communicate. People no longer reliably conduct interactions on one single device or platform. More and more, we’re seeing discussions start on one channel, such as email, and then shift to instant messaging, or telephone.
This has been exacerbated by the COVID-19 pandemic. Alongside an accelerated shift to doing business on digital devices, people are now using a new range of data-generating tools to conduct their work – from instant messaging (IM) platforms such as WhatsApp, Signal or Telegram to name a few, as well as video calls.
Capturing and combining all of this information is a challenge legal teams will need to manage. Technological solutions can help with this, and the process might become easier to manage as more and more information is held digitally on the cloud – work can begin on data stored in the cloud before it even reaches the analysis platform, again saving further time – and money.
Currently, there are real limitations in how well data review platforms can handle key pieces of information such as numbers. Manual analysis can be required where the data contains lots of figures, for example, a company’s general ledger – adding to a project’s cost. Developing this capability will be a must for more efficient analysis processes.
Progress in emotional analytics – the ability to develop an emotive ‘profile’ of an individual in their communications, and then identify emotions in their messages – is another exciting area that could significantly enhance machine’s analytical power. And, it will be important to see development in audio analysis capability, and consider the growing role of video.
There are both sources that have seen a surge in data volumes in the last year or so as the pandemic has forced business online. While the volume of audio and video recordings from platforms like Microsoft Teams or Zoom currently at play in litigation is small, it will only be a matter of time before this increases, reflecting the manner in which business has been conducted during the pandemic.
Reviewing audio quickly – whether purely on the phone, or as part of a video call – is difficult, for the simple reason that there is a limit to how quickly you can listen to sounds. There are automated voice to text solutions which are imperfect and may well misinterpret a nuanced conversation that a human reviewer would pick up on.
Using computers to reliably and consistently review multiple pieces of audio or video material simultaneously, while being able to account for natural vocal variations, factors like accents and dialects, would deliver huge time savings to legal teams. Ultimately, it would enable them to spend more time focussing on the parts of the analysis computers simply can’t handle.
Solutions are in place now to help legal teams more efficiently tackle data challenges, and further development of technology will only enhance the results. However, it’s important to remember humans are at the heart of these processes.
Ultimately, nothing is a replacement for a well-organised legal team, supported by skilled analysts who understand how to get the most out of the tools at hand. An ever-increasing reliance on technology means that the sampling and validation of machine derived results will play a key part in any project. This not only provides the legal team with the confidence that the AI is doing its job, but also allows them to validate their approach to the other side.