CS G224 Natural Language Processing Assignment 2 Spring 2006, Prof. Hafner Due Date: Wednesday, January 25, 2006 During the past 15 years, the NLP research community has increasingly been engaged in coordinated activities to develop large-scale computational resources and evaluation frameworks that benefit the entire NL field, including: -- large corpuses of naturally occurring text and speech data, some of which are marked for part-of-speech (tagged corpuses), and some of which are further marked for syntactic structure (treebanks); -- computational lexicons, concept networks, and/or ontologies -- tools for performing important NLP tasks: such as tagging, parsing, and content extraction -- methodologies and test collections for empirical evaluation and comparison of NLP systems Working with a partner, prepare a 25-minute talk to be presented to the class on January 25. The talk will be an overview of a particular resource category or evaluation activity: its purpose, its organization, its size or scope, and its contributions and/or outcomes to date. Hand in an annotated outline of your talk, including references*. This can be a written document or a printout of PowerPoint slides used in your talk. ******* Use of technology for presentations. If you wish to use computer projection in your presentation you may: -- bring your own laptop computer -- send PowerPoint files or other displayable material to Prof. Hafner by email before 3:30 on January 25. -- bring displayable files on USB stick or CD to class or to Prof. Hafner's office before class -- Plan to access the Web for your presentation (*** risky ***) Teams and selected topics: Alan Feuer, Dan Kunkle TREC evaluation activities Jun Gong, Dan Schulman Corpora and Treebanks Kevin Bloomquist, Dan Pratt Lexicons, Concept Networks, and Ontologies (esp. WordNet) John Kennedy, Victor Ortenberg MUC evaluation activities Guruprasad, Kham Nguyen Tools Lisa Norman, Jason Blind SENSEVAL evaluation activities *References to a Web site should include: -- One or more of: Author Title Sponsoring Organization (include all of these unless they are the same or unknown) -- URL -- Date visited