Topic : learning algorithm Preference : oral presentation Presenter : Anni Coden 1 A Statistical Method for Learning the Residual Quickly , Accurately , and Language Independently

Mining knowledge from various data sources is rapidly increasing. A critical first step in this process is to determining semantically equivalent concepts, in effect creating and maintaining up-to-date semantic lexicons. In the medical domain, many concepts must be gleaned from unstructured clinical text, whose format, language, abbreviations and punctuation may vary within and across institutions, as well as by specialty.