Enhanced Streaming Text Summarization (ESTESA)

Period of Performance: 07/03/2012 - 01/02/2013

$80K

Phase 1 SBIR

Recipient Firm

Language Computer Corp.
2435 N. Central Expressway Array
Richardson, TX 75080
Principal Investigator

Abstract

In this SBIR effort, called Enhanced Streaming TExt SummArization (ESTESA), Language Computer Corporation (LCC) will generate summaries of streaming texts by relying on an elaborate topic representation, called topic themes, that will be used to design and implement a novel probabilistic generative model of ranking and extracting sentences coherent with the topic. The topic themes will also inform two textual inference systems that will select novel sentences for the update summary: a textual entailment and a textual contradiction system. The textual inference system shall be further enhanced by a temporal inference model that will rely on event chronologies.