Alistair Willis

Lecturer in computing at the Open University.

Contact: Alistair Willis, Department of Computing, The Open University, Walton Hall, Keynes, Keynes, MK7 6AA, UK. Tel: 01908 652890 Fax: 01908 652140 Mail: A.G.Willis @ open.ac.uk

Research Interests My research interests focus mainly upon natural language processing and Artificial Intelligence. At some point I decided that writing computer programs was not going to teach us anything about natural intelligence, so I see the study of AI as an attempt to model those processes that would normally be thought to require intelligence in humans or animals. My particular area of interest is how to identify when stylistically different texts have the same meaning. For example, articles from journal papers are often reported in popular science periodicals or national newspapers. However, a New Scientist report on a journal article usually has very different linguistic properties from those of the original. Length, structure, and especially vocabulary are different in the two cases, reflecting the documents' different target audiences. I am interested in how to align the content of documents which are on the same subject, but are taken from such different genres. Being able to match up the content from a generalist article with that from a specialist article should improve access to the information in digital libraries for interested, but non-specialist, audiences. I currently hold a COLMSCT teaching fellowship, in which I am investigating methods to support automatic marking of student answers to short questions in free text (20 words or so). A system that can automatically mark responses and provide immediate feedback is a valuable tool for students' self-assessment and revision. Other Projects ABLE: Automatic Biodiversity Literature Enhancement. Investigating how to improve the identification of species names from biodiversity documents, in the face of errors introduced through imperfect Optical Character Recognition (funded by JISC). MaTREx: Making Tacit Knowledge in Requirements Explicit. Investigating Natural Language Processing techniques to identify where tacit knowledge and ambiguity in requirements documents might lead to incorrect implementations in later stages of the software development lifecycle (funded by EPSRC). PhD Students I am currently co-supervising Tom Collins on "Stochastic Modeling for Computer Music Generation" and Lin Ma on presupposition analysis, as part of the MaTREx project. Teaching  Joint chair of M801, the MSc Research Project and Dissertation. On the production course team for WM890, a joint course with the Centre for Law on IT and Law. On the presentation course team for M450, the undergraduate computing project. On the presentation course team for M359, the level 3 relational databases course. Other Research Interests I am interested in how we can automate the process of tesing software. Software testing is a very interesting area of software engineering that I think deserves a higher status. I am interested in how we address the questions: How do we measure how good a set of test cases is? Finding bugs in well written software is harder than finding bugs in buggy software. Given a suitable description of a piece of software, and a requirement of a certain level of testing, can we automatically generate a optimal test set to achieve that level of testing (that is, the smallest set of test cases that will achieve the trequired level of coverage)? Quick Biog I graduated in physics and philosophy from Cambridge, then took an MSc in Artificial Intelligence at Edinburgh, and a DPhil in Computational Linguistics at York. After leaving York, I worked for Philips Research Laboratories for about three years on automated software testing before returning to academia and natural language processing with the Open University. Publications Alistair Willis, Francis Chantree and Anne De Roeck. Automatic Identification of Nocuous Ambiguity. Research on Language & Computation 6(4). Springer. 2008. Alistair Willis, NP Coordination in Underspecified Scope Representations. 7th International Workshop on Computational Semantics, Tilburg, the Netherlands, January 2007. Francis Chantree, Alistair Willis, Adam Kilgarriff and Anne de Roeck, Detecting Dangerous Coordination Ambiguities Using Word Distribution. In Current Issues in Linguistic Theory IV: Recent Advances in Natural Language Processing, John Benjamins. 2007. Francis Chantree, Adam Kilgarriff, Anne de Roeck and Alistair Willis, Disambiguating Coordinations Using Word Distribution Information. RANLP 2005, Borovets, Bulgaria, September 2005. Francis Chantree, Bashar Nuseibeh, Anne de Roeck and Alistair Willis, Identifying Nocuous Ambiguities in Requirements Specifications. In Proceedings of 14th IEEE International Requirements Engineering conference (RE'06) Minneapolis/St. Paul, Minnesota, U.S.A., September 2006.