Cognitive solution for Legal

• Languages: Java
• Tech Stacks: Rest, IBM Watson, IBM Watson content analytics
• Client: Brazil
• Duration: 2 months

Project goal The Superior Court of Justice also known as STJ is the highest appellate court in Brazil for non-constitutional questions of federal law. The STJ also has original jurisdiction over some cases. The goal of project was to proof the cognitive capabilities to understand the unstructured data using machine learning technology and score the legal process based on same similarity.

I led MVP project for cognitive solution for Legal area, working with customer and a multidisciplinary team of 10 people. The goal of project was to proof the cognitive capabilities to understand the unstructured data using machine learning technology and score the legal process based on same similarity. The jurisdiction historical base was near 250.000 legal processes that we used as base for training, and we analyze 12.000 new legal processes using NLP and ML classify methods to understand the similarity and provide a scoring. we used IBM Watson APIs on cloud and IBM Watson content analytics for the cognitive solution and the customer was very satisfied and confident about the results. The main challenge was the short period of time to deliver the solution and the large quantity of information to analyze. IBM Watson NLU (Natural Language Processing) can analyze text to extract metadata from content such as concepts, entities, keywords, categories, sentiment, emotion, relations, semantic roles, and syntax.