TRESTLE project

trestle project

home

"[My task] is a battle against the bewitchment of our intelligence by means of language" - Wittgenstein

Welcome to the TRESTLE project home-page

"Text Retrieval, Extraction and Summarisation for Large Enterprises"

The application of state-of-the-art techniques for information extraction to a large-scale, real-world situation, namely some newsfeeds used by GlaxoSmithKline. From the academic point of view this gives a view on which techniques stand a chance of real use, and which directions pragmatic research could go in. From the commercial point of view a state-of-the-art system gives new utility to existing information, which is a competitive edge.

This is the version for browsers and operating systems that don't support the Javascript and/or frames used in the main version. Please use the browser's "back" button to get back here from these sub-pages.

2001-01-24	ph	change to GSK's name & logo
2000-12-20	ph	the whole site moves to a new machine outside the DCS firewall, so that IS and Glaxo can see CGI-generated pages. NetMind registrations should migrate from the old pages.
2000-04-07	ph	arranged for a version of the site which doesn't use frames or JavaScript
2000-03-17	ph	added mockup of system showing named-entity results. Sensitive information now in a secure directory: email me for the password.
2000-02-25	ph	add several menu items from IR, courtesy of hf; use overLIB to get help text on mousing over a link
2000-01-07	ph	change menu system from FolderTree to Joust rather than fix FolderTree to do what I want.
1999-12-22	ph	draft version for comment; needs frames & javascript; the right-hand side will have 'trestle' superimposed at some point; additional graphics by dg. The Trestle icon is the Japanese character for torii, which are shrine gates between the secular & sacred, or form & meaning perhaps; it looks not unlike a trestle and TRESTLE runs on GATE (OK OK, try this).

Be emailed with changes: * Powered by NetMind *