Enhanced Streaming Text Summarization (ESTESA)

In this Phase II SBIR effort, called Enhanced Streaming TExt SummArization (ESTESA), Language Computer Corporation (LCC) will generate summaries of streaming texts by enhancing the Phase I prototype system to (a) process large volumes of streaming data, (b) process natural language topic formulations entered by the users, and (c) combine multiple summarization strategies into a single topically coherent and non-redundant abstract. In addition, we propose to provide a source of supporting information for the summary in the form of dossiers on the entities and events presented in the summary. In order to achieve these goals we plan to combine our existing natural language processing tools with novel techniques that shall enhance the user experience such as (a) emerging topic detection, (b) topic decomposition and expansion, (c) targeted retrieval for summarization, (d) domain customization, (e) automatic template induction and population, (f) search space reduction for redundancy elimination, and (g) sentence compression and ordering.