[Corpora-List] WordNet ignores function words ...

Albretch Mueller lbrtchx at gmail.com
Wed May 3 23:02:16 CEST 2017


Excellent links and suggestions Ken!

Going over your links and thinking about your suggestions made me rethink/reinforce my own bias ;-), which some of you may have thought or know about.

In my own research and simulations I have come naturally to what I call "rml"-grammars (rml: "referent-modifier-link"). Do you/does anyone in our list know of ways to associate "links" (what they call "stop words") to syntax and just use indexes and then two way indexes for searches (which also helps the purpose of data compression and encryption)?

Spaces (in the languages which have them), articles, punctuation and all morphological aspects relating to syntax would be "links". Grammars can be fully described with those three elements only.

referents (having some actual semantic grounding) are substantives and verbs.

modifiers (respectively "modify" referents): adjectives and adverbs

"Mary had a little lamb":= {

[[r: Mary] [m:] [l:" "]]

[[r: to have] [m: 3rd person singular/past tense][l:" "]

[[r: lamb] [m:little] [l:" a ", " "]]

}

"Colorless green ideas sleep furiously":= {

[

[r:[r: ideas] [m:"green"] [l:" "]] [m:"Colorless"][l:" "]

]

[[r: sleep] [m:"furiously"] [l:" "]]

}

Sorry, I keep asking stuff which to you may be obvious. I am what they call an "independent researcher". I once read that Arabic grammar was studied in such a way millennia ago

lbrtchx



More information about the Corpora mailing list