next up previous contents
Next: Lexicon and Terminology Up: Spoken language terminology Previous: A hypergraph-based approach

A re-usable terminological text base

The task of automatically extracting terminology from enormous quantities of documents in the field of human language engineering, even limiting it to spoken language, is not envisaged here; this would require large-scale project resources. The strategy to be taken involves the `pre-digested' meta-level material contained in the text and the elementary terminological resources of the Handbook of Standards and Resources for Spoken Language Systems and the projected Supplement.

This approach has the following main advantages:

  1. The body of text is pre-selected by experts.
  2. The texts have pre-negotiated consensual character.
  3. There are no copyright problems.
  4. Elementary terminological work has already been done on the text.
  5. All the material is in well-defined computer-readable form.

The EAGLET concept combines a number of traditional and state-of-the-art techniques into a new kind of theoretically well-founded and practical resource for the speech community. The explicit definition of terminological micromodel and micromodel as specifications for text structuring and presentation in an implementation, together with the re-use of existing resources, is designed to result not only in efficient consensus-building but also in maximally economical use of available project resources.



Dafydd Gibbon
Wed Apr 15 14:31:25 MET DST 1998