APE: DRS-string to clause ready for assertion?

Help! Data Structure
1

How do you convert a string to Prolog fact ?

?- ape('Peter owns a book.',R).
R = '[A,B]\nobject(A,book,countable,na,eq,1)-1/4\npredicate(B,own,named(Peter),A)-1/2\n'.

The APE returns a Discourse-representation-structure as a string.
I want to convert it to Prolog fact and store it in the db, so I can later query it .
Cleaner view :

?- ape('Peter owns a book.').
[A,B]
object(A,book,countable,na,eq,1)-1/4
predicate(B,own,named(Peter),A)-1/2

FYI:

ape(T,Res) :- get_ape_results([text=T, solo=drspp],_, Res).
ape(T) :- ape(T,Res), writeln(Res).

drs instead of drspp returns this string, which cleaner :

?- ape('Peter owns a book.',R).
R = 'drs([A,B],[object(A,book,countable,na,eq,1)-1/4,predicate(B,own,named(\'Peter\'),A)-1/2])'.
2

You can recover the term with

?- read_term_from_atom('drs([A,B],[object(A,book,countable,na,eq,1)-1/4,predicate(B,own,named(\'Peter\'),A)-1/2])',T,[variable_names(VarNames)]).

In alternative, when you have decided the most useful interface for your task, it’s possible to reuse inner module’ predicates (that is, not exported) just qualifying them with the module. This is - usually - discouraged, but for a so general interface like APE seems to be pretty useful.

When analyzing, from your CLI, just enter

?- use_module(library(ape)).
true.
?- edit(get_ape_results).

and then explore the library by means of right clicks on appropriate goals. Guess you already found this trick…

2 Likes
3

thanks for the trick.

I’m wondering how do you assert() facts with variables, so you can query them ?

In this case how would I assert() DRS… what I cant wrap my head around is how do you store them… and then map the query to find the fact ?

?- ape('What does Peter own ?',R).
R = 'drs([],[question(drs([A,B],[query(A,what)-1/1,predicate(B,own,named(\'Peter\'),A)-1/4]))])'.

Using variables not pointing at anything, but used as part of the structure as “placeholders/hooks”, confuses me.

4

Just assert(FactWithVariables) :slight_smile: Variables are the most general term in Prolog, and works in facts the same they do in rules. I think these are a big payload of APE, providing ready (and easy) to use semantic data, waiting for binding to application data.

5

got it ;), just confused of how to use this in Dialog or Q/A app.
But getting there …

Still trying to get around creating correct sentences …
Straight questions seem doable, but when they become hierarchical will be mess :wink:

Is there some Formalism to extract info from DRS-db … something in the line of SQL/Xpath/?? for DRS.

Something that will permute the Q and the A so they can be matched, even if there is no exact match !! just wondering

?- listing(drs).
drs([A, B], [object(A, book, countable, na, eq, 1)-1/4, predicate(B, own, named('Peter'), A)-1/2]).
drs([], [question(drs([A, B], [query(A, what)-1/1, predicate(B, own, named('Peter'), A)-1/4]))]).
drs([], [question(drs([A, B], [object(A, book, countable, na, eq, 1)-1/5, predicate(B, own, named('Peter'), A)-1/3]))]).
6

Usually one can use drs2fol/2

exists(  What,
                   exists( Own,
                     ( query(Query,What,what) &
                     predicate(Query,Own,own,named('Peter'),What)))).

Next you will need to use something akin to PTTP or s(CASP) to convert the FOL to the prolog database.

7

Thanks, @logicmoo . I’m also struggling with DRS over here: Attempto, Prolog, and OWL - #7 by d-vyd