Our client is a legal process outsourcing leader in providing Income Tax related legal updates to law firms in the US. There are about 2600 federal, state and local websites that regularly publish relevant content for income tax analysts. The process of extracting relevant content from various websites involved a team of 15 legal experts manually do downloading websites that have changed, analysing the changes and deciding if the change is relevant and then summarizing the change for consumption by the clients.
Challenges with this process involved the following
- Repetitive process susceptible to manual errors
- Labour and time intensive process often resulting in delays in availability of final output
- Poor job satisfaction to highly skilled legal experts due to non-core work, resulting in attrition issues
Solution: Ellicium Gadfly framework for Legal Discovery
Ellicium worked with the legal experts at Mind crest to increase the efficiency of the process using artificial intelligence. Ellicium Gadfly framework used for this purpose.
The work of monitoring the websites and downloading changed content was automated using Python based Data Robots, part of Gadfly framework. Next the focus was on automating the highly complex task of classifying the documents and websites. Inputs gathered from legal experts on patterns used to classify the documents. The patterns configured in the machine learning algorithms and various algorithms tested for accuracy. Readily available algorithms in the Gadfly framework made the task of testing various algorithms quick and easy.
Manual document classification and automated classification using artificial intelligence executed in parallel for one month to benchmark the accuracy of the process.