Colander: Sifting Documents for Special Terms

We propose to demonstrate Colander, a set of lightweight approaches, and a tool, to mine “special terms” from a corpus of documents. In this proposal, we give an overview of three highlevel approaches to extracting special terms; introduce some metrics to measure the performance of the approaches; and outline an evaluation methodology. We also provide a high-level description of the tool and its features. In this proposal, we illustrate the approaches and the tool in an application to mine both known and candidate trademarks used in documents.