yakushima/nsm-dalia

Parser for Natural Semantic Metalanguage

nsm-dalia's Issues

No "another", apparently

E.g. "another kind" fails but "some other kind" succeeds. "Other kind" also succeeds.

"All of ...." noparse

"something happens somewhere" - noparse

But "something happens in a place" works

"[I know that ... = it is like this: …, I know it]"

Not clear what this means

"say something with ..." doesn't work

"say something to people" works. Unfortunately, so does "say something to something."

"I don't want this" - noparse

but "I do not want this" works

"the same someone" noparse, "the same person" succeeds

Not sure how Zamblera is handling "the"

No "to <VP>" nominalizer?

In Goddard's chart here https://intranet.secure.griffith.edu.au/__data/assets/pdf_file/0019/346033/NSM_Chart_ENGLISH_v19_April_12_2017_Greyscale.pdf

"I want to ___" yields noparse

But it's not that simple. ps("I want you to do something") works fine. Maybe just an issue with the definition of WANT? More likely KNOW. "I want you to do/say something" parses, but not "I want you to know something".

Little and Few seem entirely missing

We do have "small" though

"I want it very much" - noparse

again, noparse has "muC" in it

"at another time" - no partse

General problem with "another". "at some other times" works

"moment" examples completely broken

it happens in one moment
a moment before, a moment after
at this moment

"at the time when ..." - no parse

but "when I do this" works, so does "at this time"

"someone wants someone else to know/think something" - noparse

But works with "do" and "say"

"someone happens ...." - weird

"someone happens somewhere" - no parse. But "someone happens" and "someone happens happens in a place" both parse

"much/many more" doesn't parse by itself

"much more" gives noparse("muC more"). "Many more things" works, but not "much more things" -- sam "muC" noparse

""this word says something" - fails, but maybe OK?

"these words say something" DOES work.

"it is good if someone lives like this"

Another infinite loop, or very long noparse?

"during this time" - noparse

Not sure if this should be included

"people can/cannot see well in this place" - noparse

"few/no parts" is noparse

"this someone" works

Inconsistent with plural agreement being enforced elsewhere, maybe no big deal otherwise.

"before this" - no parse

but "before this time" works

tutorial mentions non-appearing file sentences.txt

$ find . -name '*sentences*'
./demo/eng_sentences.txt
./demo/tpi_sentences.txt

Suggestion: Pick correct sentences file per context in tutorial.txt

"someone says the same" noparse

"someone says the same thing" works. So does "someone says the same words". Same for feels, wants, knows, does, etc.

Say something in words - noparse

Goddard's chart has "say something in other words", for example. "Say something with words" also noparse

"someone's body" fails, though "body of someone" succeeds

Probably a general problem with "'s" possessive though

"someone wants to know something" - noparse

but "someone wants to say/do something" works

"after this" works but "before this" noparse

"much of this something" fails

srangely, noparse reports it as "muC of this something". "Many of these things also fails, though with no mangling of "Many".

"a short/long time before" - noparse

Prolog windows version trouble

If you do a ps("I am good") then copy-paste the result into gs(...), NSM-DALIA bombs out. Perhaps it's because starting the windows GUI initializes differently.

"someone moves somewhere" - noparse

Maybe not even a problem?

"this thing is mine" - noparse

"a short time before/after" - no parse

"someone dies at this time" - noparse

But "someone died at this time" works

comment out style_check

#1

Batch of errors from english parse (possibly already submitted)

I welcome comments like, you're barking up the wrong tree!

noparse("a other kind").
noparse("a other part").
noparse("a other time").
noparse("a other place").
noparse("one of these people").
noparse("one of these things").
noparse("two of these people").
noparse("two of these things").
noparse("many of these people").
noparse("many of these things").
noparse("this thing is above that thing").
noparse("this thing is below that thing").
noparse("this thing is inside that thing").
noparse("this thing is on the other side of that thing").
noparse("I saw someone in that place").
noparse("at that time").
noparse("before that time").
noparse("after that time").
noparse("a long time after").

"this person" - OK, "this someone" - noparse

I guess it depends on whether we want to enforce syntactic sugar

"someone says something with words" - noparse

"Some of ..." - no parse

"One of these things/people" noparse

"Two of these ..." also fails

"someone does something with part of the body" - noparse

works if you leave out "the"

"something of this kind" works (!) but not "something not of this kind"

Need to check valency options. "Not something" doesn't work. "This is something", "This is not something", "This is not one thing" ... various combinations also fail.

Solve the user:run issue

I found a run in bin/cline_interface.pl. I am not sure if its user:run or not.

$ grep run bin/cline_interface.pl
run/0,
the running interface.
%% run is det.
% Procedure run calls init/0, then enters the main loop.
run :- init,loop.

We could get lucky by running init,loop by ourselves.

"people can feel something bad when they hear this word"

"they" -> "these people" and it still doesn't work. Seems to go into a very deep search or even (is this possible in Prolog?) an infinite loop

"it happens like this for some time" - noparse

but "someone does this for some time" works

"someone feels something good/bad in part of the body" - noparse

doesn't parse even without "the"