Title: From JD to Data Science: A Case Study in AI and Law
Thomson Reuters Westlaw has long integrated machine learning and AI into its product offerings. However, in the last few years, deep neural networks have been responsible for a sea change in NLP capabilities, making possible what, until recently, would have been considered science fiction. Zac Kriegman will describe his path from a Harvard JD to a data science career focused on deep learning, and demonstrate a case study illustrating how some of these new techniques were applied to a legal annotation task to produce legal summaries on par with human annotators, allowing Thomson Reuters to improve quality, expand coverage and reduce costs.
Bio: Zac Kriegman is a Director of Data Science in Thomson Reuters Labs where he heads the Deep Learning Team that is building neural networks to understand, analyze, and generate legal language by training on Thomson Reuters’ vast repositories of legal, tax and news data. Before heading the Deep Learning Team, Zac co-created Thomson Reuters' Singapore Lab focused on developing novel financial applications with international banking customers. Prior to starting his Data Science career at Thomson Reuters, he was an economist doing econometric analysis for high stakes anti-trust litigation, a business lawyer at a top national law firm, and a software engineer at a Boston area startup (not all at the same time). He holds a B.A. in Economics from the University of Michigan and a J.D. from Harvard Law School.
This is a joint seminar with Berkman Klein Center for Internet and Society.