Summarizing with Roget's and with FrameNet

We submitted runs from two different systems for the update summary task at TAC 2009. The first system refined its use of Roget’s Thesaurus, moving beyond 2008’s semantic relatedness to compute an entropy-based uniqueness measure, with improved results in summary construction. The other system, our first use of deeper semantic knowledge, represents sentences as FrameNet types in a conceptual graph. Pairwise similarity comparisons identify the sentences most central to the document collection content and best candidates for a summary. Our AESOP submission suggests that together, the group of TAC participants tend to select summaryworthy sentences.