It is generally accepted that the single largest cost in addressing any eDiscovery exercise is neither the technology nor the litigation support teams employed, but rather it is the cost of lawyer review time. And that makes perfect sense. After all, you are paying for the lawyers’ understanding of the law and experience of its application, as they review and then build or defend the case, together with the provision of their professional assurances.
And the same holds true in an internal investigation should you need to engage external resources for content review.
However, with ever-expanding volumes of Electronically Stored Information (ESI), the scale of the task and the associated costs also grow proportionally, such that paring down the original volumes becomes a key focus of attention, to keep the cost of review in check. But how do you reconcile maintaining control of the costs with ensuring that salient evidence is not overlooked through overly ‘enthusiastic’ culling?
By the time a matter reaches the legal department or the law firm, there is a suspicion of some wrongdoing, the outline of which forms the basis of what we might call the fact patterns. It is these fact patterns, including but not limited to such elements as persons of interest, date ranges and the traditional disclosed keywords, that can be applied to the overall data set to reduce the volumes to something manageable for review.
There has been much written about the application of AI technologies such as TAR and CAL to drive more efficient review. However, all of this presupposes that the fact patterns are firstly correct and secondly complete. What if they are not a comprehensive understanding of the case, or indeed you are still in a more investigatory phase of the matter? If those ‘unknown unknowns’ are not serendipitously exposed later in the review process, or worse yet, not at all, what impact will that have on the outcome and overall costs?
Cognitive Analytics and Leading Indicators
With the sheer volumes in play, it is perhaps apparent that AI has a role to play, given the computer’s ability to process and assimilate data at a scale that is simply beyond the reach of mere mortals. However, rather than relying solely on TAR and CAL late on in the overall process, at Salient, we believe there is a strong case for applying robust AI tools far earlier on in the process, to intelligently cull the overall volumes, long before the review takes place.
Think of it as casting the net wider in the first instance. Then by applying the AI, along with an investigative mindset, you stand a better chance of exposing the leading indicators, which in turn yield a more comprehensive (or at the very least, validated) set of fact patterns. Automatically surfacing entities (such as people, places and companies), topics, patterns of behaviour, the sentiment in communications, even un-seen communicators, are all examples of readily consumable, graphically presented cognitive analytics capabilities, which can be extremely inciteful when formulating the crucial dimensions of the case.
Culling Data with Cognitive Analytics
So, in conclusion, the combination of AI, together with the ‘human in the loop’ early on in the overall process, should result in greater confidence in the subsequent culling of data. Which in turn will yield reduced volumes being passed to the most costly activity, review, at which point CAL and TAR can be applied as necessary. The approach results in a further by-product which is the reduction in the risk that critical evidence has been overlooked and has not made it to the human review stage.
If you’d like to find out how Salient can assist you with adopting cognitive analytics and building more robust fact patterns in your investigative and litigation workflows, please contact us here.