The Centre for the New Oxford English Dictionary and Text Research serves as a focal point for studying text management within the context of the Database Research Group within the School of Computer Science, Faculty of Mathematics at the University of Waterloo.
In January 1985, the University of Waterloo established the Centre for the New OED to fulfill its obligations under an agreement with the Oxford University Press to computerize the OED. The fundamental goal of the Centre remains to support innovative research through the development of application-driven text management software.
In 1989, the Oxford University Press published the 20-volume Oxford English Dictionary, second edition. As a joint venture with the Press, the University of Waterloo designed an on-line dictionary database that is suitable for editors charged with maintaining the OED, lexicographers working on other dictionaries, and researchers who wish to consult the OED interactively.
The project adopted the philosophy of modern text markup systems that a computer-processable version of text is well-represented by interleaving ``tags'' with the text of the original document, still leaving the original words in proper sequence. In its simplest interpretation, by suppressing the tags we are left with the original text, and by converting the tags to formatting instructions we can produce a suitable presentation of the text.
The most visibly successful aspect of our research is embodied in the flexible and efficient search and display software. LECTOR is a general purpose browser that takes as input a stream of arbitrarilty tagged text and formats it to the screen through user-defined typography reflecting its structure. As a complementary software component, the PAT text search engine efficiently retrieves all occurrences of words or phrases appearing in the 570-megabyte OED in less than 1 second, and allows users to restrict queries and results to arbitrary structured text fragments. Used together, PAT and LECTOR form a powerful query facility for text-dominated databases.
While the research continues at the University, the tangible results are further developed and commercialized by Open Text Corporation, a spin-off company located in Waterloo.
The Centre was an active member of CSSC, the Canadian Strategic Software Consortium, formed in 1994 to conduct pre-competitive research leading to database management systems that integrate structured text and relational data.
Our prototype distributed federated database system allowed integrated data access via the Oracle (Oracle) and DB2 (IBM) relational database systems and the SearchServer (Fulcrum), PAT (Open Text), and MultiText (UW) text engines. A fundamental goal of this research is the creation of good standards for managing and manipulating text. Such standards encourage and facilitate software interoperability and thus cooperation between between software vendors, developers, and integrators.
Principal research interests include, but are not limited to:
This project explored the use of powerful retrieval engines that combine fast text search with the power of structured data retrieval to support resource discovery.
Currently users of the World Wide Web issue queries against one or more indexing sites and browse through long lists of page references. As the internet grows in size and as more and more data is not converted to HTML until requested by users via query forms, monolithic indexing of the Web becomes less effective in meeting users' needs. Locating information is better achieved by characterizing resources (that is, extracting profile templates that describe data sources and indexed data), distributing and communicating resulting descriptions, and flexibly searching multiple, diverse template databases.
A simple user query can be transmitted to several sites (chosen by an appropriate site selection process) by a program that sits behind the browser. At each site a backend program transforms the query into several queries specific to the particular index, issues the queries, and packages the query responses into an appropriately rich structure. The frontend then provides facilities to the browser for examining the structured response and preparing subsequent refinements to the queries. The goal is to help users to find specific query forms to be used in conjunction with specific data resources to serve their information needs.
The resulting infrastructure provides scalability in terms of sites, data volume and heterogeneity of document collections.
Professor Frank Tompa
(519) 888-4567 ext. 4675
fax: (519) 885-1208