Reasoning Efficiently From Self-Organization of Unstructured Data (Resound)
Abstract
During the two years since its effective start date (28 Aug 2006), the HNC IARPA CASE project has brought us closer to the goal of a universal and optimal approach to information extraction. Building on the earlier IARPA NIMD project, new algorithms were developed for unsupervised learning of hierarchical feature sets for text and imagery, and the Text Analysis Engine (TAE) SOA component of the CASE Integrated Architecture was extended to several languages and more thoroughly hardened and tested. Most importantly, we have clarified our understanding of universal abstract principles that can guide future research on information extraction and organization directly to its greatest potential payoff.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 01, 2009
- Accession Number
- ADA510848
Entities
People
- Richard Rohwer