Our single and Multiple-Document Summarization technology offers yet another way of tackling the current information overload experienced by users and enterprises online. Summaries compress the information available in a long document by producing a much shorter text that can be read and interpreted rapidly. Our Summarization technology complements our PowerAnswer Question Answering approach, and it allows us to answer complex questions that cannot be answered by a single fact or entity.

Summarization solutions are ideally suited for customers interested in retrieving rapidly multiple perspectives on the same topic. We do not rely on simple sentence extraction or keyword matching, instead, we provide summaries based on topics' semantic representations. Our summarization technology, implemented in the GISTexter system, produces abstracts and/or extracts for virtually any topic of interest, such as news coverage on natural disasters, e.g. hurricane Gilbert, business analysis, e.g. McDonald's in Seoul or China.

The summarization technology incorporated in GISTexter was evaluated in the Document Understanding Conference (DUC), a test bed for major advances in summarization technology, both in English and other languages. Technical details about our technology have been published in the DUC-2002 Conference materials and our complete paper is available here.

Our novel single-document and multiple-document summarization solutions are based on high-accuracy methods of identifying relevant information across documents. Our topic representation, indexing and extraction methods enable the generation of coherent and informative short summaries. Instead of relying on keyword-based approaches, our unique summarization technology combines information extraction approaches with link analysis of lexico-semantic resources. Language Computer Corporation has developed a highly robust single and multiple document summarization framework which includes customized solutions for topics of high interest as well as robust solutions for virtually any new or developing topic.