To many of us the legal industry doesn’t first come to mind when thinking of the application of technology such as artificial intelligence (AI) and machine learning (ML).

Law is a field where abstract reasoning and problem solving is routinely applied in environments of legal and factual uncertainty. At first this does not sound like the type of environment where the quantitative techniques of data science such as AI and ML would be applicable. However just like in many other professional services, law is being impacted and improved by the pervasiveness of data science.

Document Analysis

A routine legal task where data science is having a profound impact is the improvement of document analysis. The process of reviewing and analysing, often large amounts of data and documents to identify which are relevant to a case, is often the most labour-intensive stage of legal work. Law practices often rely on outsourced billable resources to carry out these tasks and in some cases, outsourcing tasks to other countries where fees are lower.

New technologies can speed up and improve document analysis. Once a certain type of document is denoted as relevant, ML algorithms can search for other documents that are similarly relevant. During this process, algorithms can also learn which documents are marked as important by human operators and to suggest which ones are useful on their own. Machines are much faster at sorting through documents than humans and can produce output and results that can be statistically validated.

While the final decision of whether documents are used still belongs to experienced legal practitioners, the use of automated systems can reliably filter out large swathes of documents that are likely to be irrelevant. As with any technologies, there is still room for improvement – some difficult cases may still prove challenging for automated systems, but the time saved during the initial research / filtering phases will avoid using limited and often expensive cognitive resources.

Automated Case Search

The use of automated search systems can also support the search for records of similar and relevant past cases. This can be useful to identify case law judgements where court decisions are made by applying precedents defined by past rulings.

Documents are increasingly being stored and viewed in digital format in the cloud. In addition, there are many case law databases available online. Both of these factors helps facilitate automated searches and the collection of training large amounts of data for developing AI models.

Document processing, including text searching and topic modelling, can be tuned for legal terminology, whereby relevance is assessed based on the textual content. Automated systems can also extract information from scanned documents, like old case records that have not been converted to text documents.

Prediction

Another application of data science to law practice it the prediction of likely legal outcomes. Some academics have suggested that the combination of automated systems and human judgement will perform better than human analysis alone for a variety of legal prediction tasks. Such systems however require considerable input from human experts to categorise information about the cases into a series of machine-readable quantities. This is the most challenging component of this solution however once obtained, this information can then be combined with the likes of previous case outcomes to predict the probability of a decision or liability.

Generalised models that perform for a wide variety of cases (e.g. from divorce to litigation to traffic) are difficult to implement. Specialised systems focusing on certain types of cases are easier to implement. These models can provide outputs such as “based on previous outcomes there is a 78% change of a guilty verdict”. This information has the opportunity to provide a valuable addition to standard legal counselling approaches.

As in ever expanding areas where these types of technologies are applied, data science is helping law firms become more efficient. Data science techniques and models save costs and free up law professionals from repetitive and time-consuming tasks, such as document searching, allowing them to focus on value-add activities such as advising and decision making – tasks which for the timebeing are best left to humans.