Topic: | Re:Formats for documenting NLP patterns |
Posted by: | Carmen Bostic St. Clair and John Grinder |
Date/Time: | 20/09/2002 17:34:11 |
Jon We find your suggestions very attractive. In general, at this stage of development of NLP, any set of reasonably formal representations would add great value to the enterprise. Even a review of the miscommunications that have occurred on this website indicates a serious need for more explicit representations to ensure we are referring to the same phenomena - see Whispering, pages 351 - 352 for an elaboration of an alternative method for accomplishing this. The GEM and GLIF formats have value - whether they are appropriate for reporting and representing patterning in NLP is an open question and one worth considering. The most interesting portion of your posting for us was the statement, "As I've been reading WITW, part of me has been wondering "How could one represent the concepts as state-diagrams, guidelines, flowcharts, or similar"." We are quite hopeful that your suggestion will take in the community. As you likely noticed in the section called "a recognition of the value of formalization and explicit" pages 127 and 128 in Whispering, we sense that an increase in the formal quality and explicit representation of pattening in NLP would constitute a major advance for the field. Indeed, Automata Theory is the underlying model for the linguistic portions (at least) of NLP patterning and is simultaneously the source for formal objects such as state diagrams (from Turing machines to finite state automata). We believe the fit is excellent. Would you be willing to work up a relatively simple example of how you might take a well-known NLP pattern and represent it in state diagrams or flow charts? It would introduce some readers to what you are proposing and for those already familiar with such techniques, it would offer a model of how such representation might be done. If you want to confer off-line during the preparation of such an example, it would be a pleasure to collaborate. All the best, Carmen and John |