Dear Philip, Steve and Robert Thanks for your comments. I think all comments are welcome, both detailed and general. The grammar has intellectual interest, but my main aim is to see whether we can produce something that'll help standard names to be agreed more quickly and with less effort, as Philip suggests.
* "In the phenomenon section, you list both 'due_to' and 'from'. However, from examining the patterns, the only time 'from_phenomenon' appears is as a secondary phenomenon." That's right, and actually the primary phenomenon is always the same one! So I should really have just put the from phrases as part of the due_to and omitted the from category. Thanks. * "Is it worth adding usage comments to the lexicon?" Yes, absolutely. In fact, we could put explanatory text in the lexicon that could automatically be used to synthesise the full description of the standard name, by stringing together the text for the constituent phrases. The lexicon would have to be an xml file to make this manageable, I suppose, which is a pity because xml files are much less readable than plain files in columns! * "could/should the "water_source" vocabulary be subsumed under "phenomenon"?" That's a good idea. The from_(water_source) phrases really have the same kind of function as due_to_(phenomenon) and those types could be merged. * "wondering whether the "untyped" names might deserve to be treated a little differently ..." Quite right. I made some related comments in the "Directions" section about what could be done in future. Since there is only one level of parsing at present, the grammar cannot recognise that product_of X and Y is a higher-level pattern, in which "product_of" and "and" are essential syntactic elements. Several of the untyped phrases are used for such purposes. Others are used to generate scalar standard names from species etc. These would be recognised at a lower level of parsing. Best wishes Jonathan _______________________________________________ CF-metadata mailing list [email protected] http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata
