Legal Decision Support: Exploring Big Data Analytics Approach to Modeling Pharma Patent Validity Cases

This exploratory research examines the potential for applying a big data analytic framework to the modeling and analysis of cases in pharmaceutical patent validity brought before the U.S. Court of Appeals of the Federal Circuit. We start with two specific goals: one, to identify the key issues or reasons the Court uses to makes validity decisions and, two, to attempt to predict outcomes for new cases. The ultimate goal is to support legal decision-making with automation. The legal domain is a challenging one to tackle. However, current advances in analytic technologies and models hold the promise of success. Our application of Hadoop MapReduce in conjunction with a number of algorithms, such as clustering, classification, word count, word co-occurrence, and row similarity, is encouraging, in that the results are robust enough to suggest these approaches have promise and are worth pursuing. By utilizing larger case data sets and sample sizes and by using deep machine learning models in text analytics, more breakthroughs can be achieved to provide decision support to the legal domain. From an economic standpoint, the potential for litigation cost reduction is another objective of our study. Synergies are obtained in applying lessons to the computational field and vice versa, leading to acceleration in our understanding.