FuzzieDice,
snoz0r has hypothesized correctly. The XTF Brain extracts known synonyms and meronyms from the Wordnet data base when a new topic is created. Some topic words have just meronyms, some just synonyms, some have both. The "related" file for that topic is created the first time a new topic is introduced. The Wordnet data is placed between the "exist" and "end here" flags just for reference purposes. New "related" words that Hal learns from the user are added in after the Wordnet data.
By definition all Wordnet meronyms and synonyms are "True". The user taught "related" words can be "True" or "False" depending on how the user responds to Hal's inquiries about two words being "related". You can see from my file example below that some user taught words are "False". Upon examination I think you can see why some words aren't considered "related" to the main topic of beer.
Here is my "XTF_BEER_Related.brn" file:
" [SYNONYMS_EXIST_FLAG] ","True"
" BREW ","True"
" BREWAGE ","True"
" [SYNONYMS_END_HERE] ","True"
" ALE ","True"
" LAGER ","True"
" MAKES ","False"
" MAKE ","False"
" FLAVORING ","False"
" BREWERY ","True"
" BREWERIES ","True"
" FLAVOR ","False"
" DAY ","False"
" RAT ","False"
" RATS ","False"
" ELEMENT ","False"
" ALUMINUM ","False"
" OCTOBERFEST ","True"
" FUN ","False"
" SUDS ","True"
" LIKE ","False"
" LOT ","False"
" GO ","False"
" FAVORITE ","False"
" YOU'LL ","False"
" WORK ","False"
" BUDWIESER ","True"
" COORS ","True"
" BUDWEISER ","True"
Please note these files can be manually edited, but the uppercase, lowercase and capitalization formats must be exact. The file *must* end in a "blank" line like so many other Hal .brn files.
Whew! Another XTF Brain mystery solved.
=vonsmith=