Enhanced Summarizations of Streaming Text Navy SBIR FY2012.1 Sol No.: Navy SBIR FY2012.1 Topic No.: N121-078 Topic Title: Enhanced Summarizations of Streaming Text Proposal No.: N121-078-0951 Firm: Securboration Inc 1050 W NASA Blvd Suite 155 Melbourne, Florida 32901 Contact: Josh Powers Phone: (571) 338-6957 Web Site: www.securboration.com Abstract: Streams of time-stamped, unstructured text content represent both a valuable information resource for analysts and a daunting volume of input which must be sifted and sorted according to multiple perspectives. Securboration proposes Topical Summarization Over Time (TSOT), which treats summaries as objects which have lifecycles developing as more reports about their real-world events and entities become available. TSOT will allow a human user to explicitly request summaries for a population of documents of interest. More interestingly, TSOT will monitor streaming data sources for topics of interest which form over time, creating summaries as necessary. Such topics arise quickly in message or news traffic, but can very often be aligned to existing Critical Information Requirements (CIR). When a group of documents describes some real world event or situation which matches the criteria of a CIR, TSOT will produce a summary which ties the content directly to the CIR's specification, orienting the analyst to the important source document statements. Metadata about summary statements will be presented, including a strength of signal, the time signature of the particular information and a measure of reliability of the information. The summary and its metadata will change over time as new developments occur. Benefits: The Topical Summaries Over Time (TSOT) platform will be a strong support tool for DoD analysts who must monitor large volumes of streaming data sources, looking for signals that an event or situation is arising which corresponds to a Critical Information Requirement (CIR). By associating dataset contents with existing CIR definitions and structured domain vocabularies, TSOT will quickly produce summaries of current events and keep tracking them through time. Once an important event has occurred, there are bound to be multiple, redundant reports about it. TSOT suppresses this redundancy, but retains enough provenance metadata to allow source attribution when requested. TSOT also emphasizes new content as it enters the summary. Outside the DoD, investigative government functions such as law enforcement and communities that use analytical processes such as the pharmaceutical, legal and financial industries would realize similar benefits, particularly when their analytical functions are spread over multiple streams of data. Finally, a version of TSOT tailored to small businesses and personal users would allow any individual to select a template domain context and a dataset, and discover documents pertaining to them, and be able to monitor them as the develop. Return