Hidden in plain sight: Using law to summarize the law
summary
Case law gets automatically summarized by harvesting judge-written parenthetical summaries and stitching frequently quoted text into key passages, using deterministic parsing (context-free grammars and Antlr) plus machine indexing with Apache UIMA. Pipeline runs on Spark/EMR, orchestration via Airflow, storage in S3 with Postgres metadata. Initial run yields about 2.5 million harvested summaries across ~8 million opinions (roughly 1.5 million cases summarized), favoring high-precision extraction over abstractive generation.
AI-written summary · about
chapters (7)
- 0:00Introduction to CaseText
- 1:44System architecture overview
- 3:53Data processing and NLP pipeline
- 8:02Extracting judicial summaries
- 13:46Key passages and summaries
- 15:34Demo of judicial summary extraction
- 17:44Q&A on data and summarization
related talks
20mKnowledge Scout: From Talks and Slides to Smart Summaries
49mNext-Gen Documentation with AI
16mInteract with Your Technical Documentation via Generative AI
43mA journey from static documentation to generative AI
28mA Retrieval Augmented Generation system to query the scikit-learn documentation
29mI feel like writing software documentation is like doing my taxes - help!
24mNet2Text: Query-Guided Summarization of Network Forwarding Behaviors
10mCrowdsourcing Scholarly Discourse Annotations