Rogets Data
Roget's needs to be built as a web site of "Subject Indicator Pages" defining
Semantic class models
The PSIs we establish should take dynamic parameters for category and
POS data (stuff we can get out of Scanner & Transcriber) from text.
Optionally, they should include INGLISH semantic summaries for the
class, which if not present explictly will be used implicitly in the page
data:
-
the SIs will represent classes of THINGS modeled in INGLISH
ontology
-
such models can be embedded as "coded strings" in the URI used
-
they go into a local class-topic, created with a fixed basename
-
that topic should be treated as a local class within the TM
-
that rule helps proclaim (?) all classes "expected" within it
-
that in turn helps the Scanner select lexemes for parser
-
by publishing their URIs, Lexikos can standardize usage of
"lexicon" TMs
-
browsing to the URI (as help text) cites what each class models
-
the rules formalize what is under #1 in human-readable terms
-
it also explains: if two "lexicon TMs" merge, a legal third results!!
A "lexicon TM" is (a) CORELEX, or (b) an extension holding extra vocabulary
for a context.
Must ensure that one can build, combine and use them easily to help
create DISCOURSE.