Machine learning speeds up virus coverage insurance reviews

May 7. 2020
Cover image

Written by Alison Meredith. As featured in The Lawyer's Daily.

A key task for lawyers representing their corporate clients over the next few months will be to review insurance policies for ways in which those clients can recover from their insurers. This means that we can all unfortunately expect to spend a lot more time deeply immersed in those lengthy documents.

Keyword searching long documents for a comprehensive list of pandemic related search terms is a task that simultaneously requires high attention to detail and is, at scale, mind numbingly dull. However there are tools that can help lawyers with this type of task. The tools won’t do the legal work for you, but they will help reduce the manual part. Things like scrolling through documents, copying and pasting, retyping paragraphs, indexing provisions and filling out templates can be eliminated or significantly reduced.

The combination of the right features together will typically save between 10 to 20 minutes per document and often more. Over the course of a month this sort of time saving could become significant. Utilising simple tools like this can help teams stay on top of their caseload.

Here are the key questions you should ask when evaluating whether a contract review tool could be configured to help with your insurance policy project.

What is machine learning, and is it right for my use case?

Machine learning searches that contract review platforms use are typically built to find provisions or paragraphs rather than words. This type of search will save you the most time when your documents use keywords many times in many different contexts, and only some of those are important.

For lawyers reviewing insurance policies for pandemic coverage this could be a key time saver. Keyword searching an insurance policy for ‘virus’ would likely return many results about both pathological viruses and computer viruses, only some of which would be important. Using a machine learning search will distinguish between those two contexts. This will ultimately reduce the amount of hits that need review per document which means less time reviewing irrelevant parts of the document and more time analyzing the important parts.

Will the tool allow me to keyword search within provisions?

Keyword searching is useful because it is very precise and inflexible, and in some cases, that is exactly what you need. A simple way that many contract review tools exploit the power of keyword searching is by giving users the option to search for keywords in targeted provisions rather than the whole document. By splitting the document up into it’s different provisions and then searching each provision, you can create a search that looks for a keyword only when it appears in the context of a provision. This will be extremely useful for any lawyers who are searching things like definitions of ‘micro-organism’ for the word ‘virus’ or ‘force majeure’ provisions for pandemic-related triggers.

What pre-built machine learning models does the tool have included?

Many contract review platforms will have a library of machine learning searches that are built and maintained by their lawyers. Using these will mean less setup time for you, so if they cover your use case then these are an excellent resource that should be taken advantage of where possible.

Can I use the tool to train my own machine learning models?

Some platforms also offer options for you to train your own new searches. This can be a great option where you have a large set of documents that you’re looking for something very specific in. The training time to build your own search can be as little as 20 to 30 minutes with the right tool. If you’re reviewing 200 insurance policies that’s an investment of less than 10 seconds per policy. Look for tools that have a purpose-built training component to their platform for best efficiency.

Can I export the results to a report?

Almost all contract review platforms will allow you to export a report of the key findings in your contracts or policies. An automatic report might need some human checking before it is client ready, but will give you a decent head start. This will be useful for all types of reviews, but most useful where you want to analyze a large set of documents, or you need to provide an abstract to the client.

Can I see a demo?

You should always ask for a demonstration of a tool’s key functionality before you start a trial. Explain your use case so the vendor can personalise a demo to you. Seeing how a tool works in real time will help to understand how it will fit into your process and whether it is a good fit for your practice.

Originally Published by The Lawyer's Daily, part of LexisNexis Canada Inc. Written by Alison Meredith.