CWIC Communicating Intelligently with Computers

Abstract

The goals of DARPA Communicating with Computers program are to advance the state of the art in text and video analytics to the extent that the machine and the human operator have the same mental model. This requires that the machine be able to understand the human intent, and that can explain back, to the human, in ways that makes use of the context and prior knowledge. Three challenge problems were chosen: Block world, Bio curation and Story generation. Furthermore, an Offeror could choose to address five different task areas: TA1 Ð apparatus for the block world, TA2 Ð Elementary Composable Ideas, TA3 Ð Composition of Ideas that takes context into account, TA4 Ð What to say or Do and TA5 -- Evaluation. The OfferorÕs task is to address TA2, TA3 and TA4. The proposed work of the Offeror will be based on seven innovations in (a) a unified framework for representing context, language, images, video, relations, actions and goals; (b) a novel multi-modal data-structure for grounding language and actions, intentions and goals; (c) a large scale simulation environment for representing and evaluating use cases; (d) a stochastic scenario generator that could be used in story generation; (e) enhancement and integration of currently available implementations with proposed work; (f) exploration of new techniques for checking textual and knowledge coherence; (g) compositional algorithms for creation of coherent stories. All of the work will be evaluated in the context of program wide metrics.

Document Details

Document Type
DoD Grant Award
Publication Date
Jan 12, 2017
Source ID
W911NF1510543

Entities

People

  • Kevin Knight

Organizations

  • Army Contracting Command
  • Defense Advanced Research Projects Agency
  • University of Southern California

Tags

Fields of Study

  • Computer science

Readers

  • Computational Linguistics
  • Distributed Systems and Data Platform Development
  • Research Science/Academic Research