LANGUAGE COMPUTER CORPORATION — Department of Defense SBIR Phase II: DTRA152-005
LANGUAGE COMPUTER CORPORATION — SBIR Phase II award from Department of Defense.
- Amount
- $999,964
- Agency
- Department of Defense · Defense Threat Reduction Agency
- Program / Phase
- SBIR · Phase II
- Topic
- DTRA152-005
- Solicitation
- 2015.2
- NAICS
- —
- Place of performance
- TX
- Period
- 2017-12-13 → 2019-12-12
Description
In Phase II of JOLTER, we will develop several extensions and improvements upon the probabilistic graphical models developed in Phase I for discovering topics, entities, and relations jointly. The system will be able to (1) Discover and extend hierarchies of topics, entities, and relations; (2) Discover categories for events, multiway relations, and their roles; (3) Conditionalize the topics, entities, and relations on metadata available about the document collection; (4) Display the learned categories and assignments of categories to documents using learned labels; and (5) Learn optimal hyperparameter settings such as the number of categories and perform probabilistic inference in an efficient manner.