% Discourse 2009 go(S) :- retractall(tel(_)), assert(tel(1)), %blank slate DRS = drs([], []), analyse([S], DRS, _). gokat :- retractall(tel(_)), assert(tel(1)), %blank slate DRS = drs([], []), S = [ [ik,zie,een,rode,kat], [ik,zie,een,blauwe,mat], [de,kat,zit,op,de,mat] ], analyse(S, DRS, _). go :- retractall(tel(_)), assert(tel(1)), %blank slate DRS = drs([], []), S = [ ['in', 'het', 'midden', 'van', 'het', 'hartje', 'van', 'het', 'binnenste', 'van', 'het', 'muiskleurig', 'gebergte', 'woonde', 'de', 'weerwolf'], ['hij', 'was', 'groot', 'en', 'verschrikkelijk'], ['zijn', 'ogen', 'sproeiden', 'vuur', 'en', 'zijn', 'tong', 'had', 'karteltjes'] ], analyse(S, DRS, NewDRS), %S1 = ['iedere', 'avond', 'om', 'zes', 'uur', 'stiet', 'hij', 'een', 'gebrul', 'uit', 'waarvan', 'het', 'hele', 'muiskleurige', 'gebergte', 'sidderde'], interpret(S1, M1, NewDRS), %this sentence is not interpreted correctly yet, so here is a hardcoded version M1 = [ lex(hij, [], [find(33), pronoun(1), male(33), singular(1), animate(33)], 1), lex(had, [1, 3], [verb(2), pastTense(2), eat(1), eat(3), singular(1), noun(1), before(1), noun(3), after(3)], 2), lex(lange, [4], [adj(3), eat(4), noun(4), next(4), same_features(3, 4)], 3), lex(witte, [5], [adj(4), eat(5), noun(5), next(5), same_features(4, 5)], 4), lex(scherpe, [6], [adj(5), eat(6), noun(6), next(6), same_features(5, 6)], 5), lex(tanden, [], [noun(6), plural(6), inanimate(6)], 6), lex(en, [2, 18], [conn(7), eat(2), eat(18)], 7), lex(iedere, [9, 13], [univ(8), eat(9), noun(9), next(9), singular(9), same_features(8, 9), verb(13), eat(13)], 8), lex(avond, [], [noun(9), inanimate(9), singular(9), male(9)], 9), % lex(om, [8, 11], [prep(10), eat(8), before(8), noun(8), eat(11), noun(11), after(11)], 10), % lex(zes, [12], [det(11), eat(12), noun(12), next(12), same_features(11, 12)], 11), % lex(uur, [], [noun(12), neuter(12), inanimate(12)], 12), lex(stiet, [14, 15, 17], [verb(13), pastTense(13), eat(14), eat(15), eat(17), singular(14), animate(14), noun(15), after(15), particle(17)], 13), lex(hij, [], [find(33), pronoun(14), male(33), singular(14), animate(33)], 14), lex(een, [16], [det(15), bind(15), eat(16), next(16), noun(16), singular(16), same_features(15, 16)], 15), lex(gebrul, [], [noun(16), singular(16), neuter(16), inanimate(16)], 16), lex(uit, [], [particle(17)], 17) , lex(waarvan, [16, 23], [prep(18), eat(10), eat(23), prep(10), verb(23)], 18), lex(het, [20], [det(19), find(11), eat(20), next(20), neuter(20), same_features(19, 20)], 19), lex(hele, [21], [adj(20), eat(21), noun(21), next(21), same_features(20, 21)], 20), lex(muiskleurige, [22], [adj(21), noun(22), next(22), same_features(21, 22), eat(22)], 21) , lex(gebergte, [], [noun(22), neuter(22), singular(22)], 22), lex(sidderde, [19], [verb(23), pastTense(23), singular(19), before(19), noun(19), eat(19)], 23)], docontext(M1, M1, NewDRS, NNewDRS), analyse([ ['dat', 'is', 'de', 'weerwolf'] ], NNewDRS, _NNNewDRS). analyse([], DRS, DRS). analyse([Sentence | Rest], DRS, Result) :- interpret(Sentence, Meaning, DRS), writeln('Interpretation: '), writelist(Meaning), nl, %generatie check? docontext(Meaning, Meaning, DRS, drs(Ref, Cond)), writeln('Sentence: '), writeln(Sentence), nl, writeln('Meaning: '), writelist(Meaning), nl, writeln('Context: '), % 'pretty-print' DRS writeln(Ref), writeln('--------------'), writelist(Cond), nl, analyse(Rest, drs(Ref, Cond), Result). writelist([]). writelist([Head | Tail]) :- writeln(Head), writelist(Tail). % Groep 1: Lexicon %lex(word, VariablesToEat, Features). %Features: % - eat, find, bind % - part of speech (det, pron, neg, prep, adj, noun, verb) % - after, before, next (difference between before and next is that next % only matches the word directly next to it) % - male, female, neuter % - plural, singular % - animate % % All features take one argument. The argument is anonymous (_) if it is about % the word itself, or it specifies a constraint for eat/find/bind. % in het midden van het hartje van het binnenste van het muiskleurig gebergte woonde de weerwolf lex(in, [X, Y], [prep(This), eat(X), eat(Y), verb(X), prep(Y), after(Y)], This). lex(het, [X], [det(This), find(This), eat(X), next(X), neuter(X), same_features(This, X)], This). lex(midden, [], [noun(This), neuter(This)], This). lex(van, [X,Y], [prep(This), eat(X), eat(Y), noun(X), prep(Y), before(X), after(Y)], This). lex(van, [X,Y], [prep(This), eat(X), eat(Y), noun(X), noun(Y), before(X), after(Y)], This). lex(hartje, [], [noun(This), neuter(This)], This). lex(binnenste, [], [noun(This), neuter(This)], This). lex(muiskleurig, [X], [adj(This), eat(X), noun(X), next(X), same_features(This, X)], This). lex(gebergte, [], [noun(This), neuter(This),singular(This)], This). lex(woonde, [X], [verb(This), singular(This), past(This), eat(X), noun(X), animate(X)], This). lex(de, [X], [det(This), find(_), eat(X), noun(X), next(X), male(X), same_features(This, X)], This). lex(de, [X], [det(This), find(_Y), eat(X), next(X), female(X), same_features(This, X)], This). lex(de, [X], [det(This), find(_Y), eat(X), next(X), plural(X), same_features(This, X)], This). lex(weerwolf, [], [noun(This), animate(This), male(This), singular(This)], This). % hij was groot en verschrikkelijk lex(hij, [], [pronoun(This), find(X), male(X), animate(X), singular(This)], This). lex(was, [X,Y], [verb(This), pastTense(This), singular(This), eat(X), eat(Y)],This). lex(groot, [], [adj(This)], This). lex(en, [X,Y], [npconj(This), eat(X), eat(Y), before(X), after(Y)], This). % een ‘en’ zoals in een opsomming lex(verschrikkelijk, [], [adj(This)], This). % zijn ogen sproeiden vuur en zijn tong had karteltjes lex(zijn, [X], [pronoun(This), eat(X), noun(X), next(X), find(This), male(This), same_features(This, X)], This). lex(ogen, [], [noun(This), plural(This), inanimate(This)], This). lex(sproeiden, [X,Y], [verb(This), pastTense(This), eat(X), eat(Y), plural(X), noun(X), before(X), noun(Y), after(Y)], This). lex(vuur, [], [noun(This), inanimate(This), singular(This), neuter(This)], This). lex(en, [X,Y], [conn(This), eat(X), eat(Y), before(X), after(Y), verb(X),verb(Y)], This). % een ‘en’ die twee complete zinnen verbindt lex(tong, [], [noun(This), inanimate(This), singular(This), female(This)], This). %lex(had, [X,Y], [verb(This), pastTense(This), eat(X), eat(Y), singular(X), noun(X), before(X), noun(Y), after(Y)], This). lex(had,[X,Y],[verb(This), singular(This),pastTense(This),eat(X),eat(Y), noun(Y)],This). lex(karteltjes, [], [noun(This), plural(This)], This). % hij had lange witte scherpe tanden en lex(lange, [X], [adj(This),eat(X),next(X),same_features(This,X)], This). lex(witte, [X], [adj(This),eat(X),next(X),same_features(This,X)], This). lex(scherpe,[X],[adj(This),eat(X),next(X),same_features(This,X)], This). lex(tanden, [], [noun(This),plural(This),inanimate(This)], This). % iedere avond om zes uur stiet hij een gebrul uit waarvan het hele muiskleurige gebergte sidderde lex(iedere, [X,Y], [det(This), univ(This),eat(X),noun(X),next(X),singular(X),same_features(This,X),verb(Y),eat(Y)], This). lex(avond, [] , [noun(This),inanimate(This),singular(This),male(This)], This). lex(om, [X,Y], [prep(This),eat(X),before(X),noun(X),eat(Y),noun(Y),after(Y)], This). lex(zes, [X], [det(This),eat(X),noun(X),next(X),same_features(This,X)], This). lex(uur, [], [noun(This),neuter(This),inanimate(This)], This). lex(stiet,[X,Y,Z], [verb(This),pastTense(This),eat(X),eat(Y),eat(Z),singular(X),animate(X),noun(Y),after(Y),particle(Z)], This). lex(een, [X], [det(T), eat(X), bind(_Y), next(X), same_features(T, X)], T). lex(gebrul, [], [noun(This),singular(This),neuter(This),inanimate(This)], This). % deze ‘uit’ hoort bij ‘stiet’ (uitstoten) % ik maak er daarom een particle van en geen preposition % zie http://en.wikipedia.org/wiki/Preposition_and_postposition#Particles % -- Jeroen lex(uit, [], [particle(This)], This). lex(waarvan, [X,Y], [prep(This),eat(X),eat(Y),prep(X),verb(Y)], This). % ‘waarvan’ is een betrekkelijk voornaamwoord lex(hele, [X], [adj(This),eat(X),noun(X),next(X),same_features(This,X)], This). lex(muiskleurige, [X], [adj(This),noun(X),next(X),same_features(This,X),eat(X)], This). lex(sidderde, [X], [verb(This),pastTense(This),singular(X),before(X),noun(X),eat(X)], This). % dat is de weerwolf lex(dat,[X],[pronoun(This),noun(X),find(X)],This). lex(is, [X,Y],[verb(This),presentTense(This),singular(This),noun(X),eat(X),eat(Y)],This). % ik zie een rode kat en een blauwe mat. de kat zit niet op de mat. lex(ik, [], [pronoun(T), find(Y), animate(Y), animate(T), male(Y)], T). lex(zie, [X, Y], [eat(X), eat(Y), before(X), verb(T), animate(X), noun(Y)], T). lex(rode, [X], [eat(X), adj(T), next(X), same_features(T, X)], T). lex(kat, [], [noun(Y), animate(Y), male(Y)], Y). lex(blauwe, [X], [adj(T), eat(X), next(X), same_features(T, X)], T). lex(zit, [X], [eat(X), before(X), verb(T)], T). lex(niet, [X], [eat(X), before(X), verb(X), neg(T)], T). lex(mat, [], [noun(T), male(T)], T). lex(op, [X,Y], [prep(T), eat(X), eat(Y), before(X), next(Y), verb(X)], T). :- dynamic tel/1. % Groep 2 interpret(S, M, _):- def(S, M), eat_in_order(M). % Get word definitions and assign unique word IDs def([], []). def([H|T], [lex(H, A2, A3, Tel)|M]) :- tel(Tel), lex(H, A2, A3, Tel), NewTel is Tel + 1, retractall(tel(_)), assert(tel(NewTel)), def(T, M). %apply predicate to all elements of the list and retain unifications eat_in_order(Sentence):- eat_in_order(Sentence, [], Breakfast, adj), eat_in_order(Sentence, Breakfast, Snack1, pronoun), eat_in_order(Sentence, Snack1, Lunch, det), eat_in_order(Sentence, Lunch, Snack2, verb), eat_in_order(Sentence, Snack2, Dinner, conn), eat_in_order(Sentence, Dinner, _, prep). eat_in_order(Sentence, Stomach, Meal, Type) :- findall(lex(A1,A2,A3,A4), (member(lex(A1, A2, A3, A4), Sentence), Pred=..[Type,A4], strict_mem2(Pred, A3)), Lexs), ( Type = adj, reverse(Lexs, OrderedLexs) ; OrderedLexs = Lexs ), eat_list(OrderedLexs, Stomach, Meal, Sentence). eat_list([], Meal, Meal, _). eat_list([H|T], Stomach, Meal, Sentence):- eat(H, Stomach, MealPart, Sentence), eat_list(T, MealPart, Meal, Sentence). % succeed if there is nothing to eat: eat(lex(_,_,Feat,_), Meal, Meal, _Meaning) :- \+ member(eat(X), Feat), var(X). eat(lex(_,_,Feat, Id), Stomach, Meal, Meaning) :- strict_sel(eat(_), Feat, eat(Food), Rest), %find all constraints specified for the referent we need to find: findall(Formula, (strict_mem([P,Food], Rest), Formula=..[P, Food]), Constraints), %instantiate X with word satisfying constraints getFood(Food, Id, Constraints, Stomach, Meaning), ( member(lex(_,_,Feat,Id), Meaning),! ; true ), eat(lex(_,_,Rest, Id), [Food|Stomach], Meal, Meaning). strict_sel(M, [H|T], H, T):- M = H. strict_sel(M, [H|T], X, [H|Rest]):- \+ M = H, strict_sel(M, T, X, Rest). strict_mem([Pred,X], [H|_]):- H =..[Pred,Y], X == Y. strict_mem([P,X], [_|T]):- %H =..[_, Y], %\+ X == Y, strict_mem([P,X], T). strict_mem2(Pred, [H|_]):- Pred == H. strict_mem2(Pred, [H|T]):- \+ Pred == H, strict_mem2(Pred, T). getFeatures(lex(_, _, Feats, ID), _, Feats):- \+ member(same_features(ID, _), Feats). getFeatures(lex(_, _, Feats, ID), M, TotalFeats):- select(same_features(ID, FoodID), Feats, TempFeats), nonvar(FoodID), member(lex(_, _, Feats2, FoodID), M), getFeatures(lex(_, _, Feats2, FoodID), M, Feats3), replaceIDFeats(ID, FoodID, Feats3, Feats4), append(TempFeats, Feats4, TotalFeats). replaceIDFeats(_, _, [], []). replaceIDFeats(ID, FoodID, [H|T], [IDFeat|Rest]):- H =.. [Pred, FoodID], IDFeat =.. [Pred, ID], replaceIDFeats(ID, FoodID, T, Rest). replaceIDFeats(ID, FoodID, [H|T], Rest):- \+ H =.. [_, FoodID], replaceIDFeats(ID, FoodID, T, Rest). getFood(X, Id, Constraints, Stomach, Meaning) :- ( member(before(_), Constraints), reverse(Meaning, OrderedM) ; OrderedM = Meaning ), member(lex(_, _, Feats, FoodID), OrderedM), \+ Id = FoodID, %don't eat yourself. % don't eat something which has already been eaten: (causes problems with current lexicon) \+ member(FoodID, Stomach), getFeatures(lex(_, _, Feats, FoodID), Meaning, TotalFeats), select_relevant_feats(FoodID, TotalFeats, RelevantFeats), forall(member(C, Constraints), checkconstraint(C, RelevantFeats, Id, FoodID)), X = FoodID. select_relevant_feats(_, [], []). select_relevant_feats(ID, [H|T], [H|RelFeats]):- H =..[_,X], X == ID, select_relevant_feats(ID, T, RelFeats). select_relevant_feats(ID, [H|T], RelFeats):- H =..[_,X], \+ X == ID, select_relevant_feats(ID, T, RelFeats). checkconstraint(C, Feats, Id, FoodId) :- (C = before(_), FoodId < Id) ; (C = after(_), FoodId > Id) ; % words directly next to each other (C = next(_), FoodId =:= Id + 1) ; ( \+ C = after(_), \+ C = before(_), \+ C = next(_), member(C, Feats)). sentenceFind(Meaning, [], _, Meaning, _). sentenceFind(Meaning, [H|T], DRS, NewMeaning, Refs):- H = lex(_,_,Feat, _), \+ member(find(_), Feat), sentenceFind(Meaning, T, DRS, NewMeaning, Refs). sentenceFind(Meaning, [H|T], DRS, NewMeaning, Refs):- H = lex(_Word, _OtherRefs, Feat, _Id), member(find(_), Feat), find(H, Meaning, DRS, Ref, OldRef), writeln(Ref), writeln(Refs), \+ strict_mem2(Ref, Refs), replaceId(Meaning, OldRef, Ref, TempNewMeaning), sentenceFind(TempNewMeaning, T, DRS, NewMeaning,[Ref|Refs]). replaceId([], _,_,[]). replaceId([H|T], OldId, NewId, [NewLex|Rest]):- H = lex(Word,Refs,Feats, Id), Id = OldId, replaceLexId(Feats, OldId, NewId, NewFeats), replaceRefs(Refs, OldId, NewId, NewRefs), NewLex = lex(Word, NewRefs, NewFeats, NewId), replaceId(T, OldId, NewId, Rest). replaceId([H|T], OldId, NewId, [NewLex|Rest]):- H = lex(Word,Refs,Feats, Id), \+ Id = oldId, replaceLexId(Feats, OldId, NewId, NewFeats), replaceRefs(Refs, OldId, NewId, NewRefs), NewLex = lex(Word, NewRefs, NewFeats, Id), replaceId(T, OldId, NewId, Rest). replaceRefs([], _, _,[]). replaceRefs([OldId|T],OldId, NewId, [NewId|Rest]):- replaceRefs(T, OldId, NewId, Rest). replaceRefs([H|T], OldId, NewId, [H|Rest]):- H \= OldId, replaceRefs(T, OldId, NewId, Rest). % Replace the old referent by a referent that has been found replaceLexId([], _, _,[]). replaceLexId([H|T], Id, NewRef, [NewFeat|Rest]):- H =.. [Pred, Id], NewFeat =.. [Pred, NewRef], replaceLexId(T, Id, NewRef, Rest). replaceLexId([H|T], Id, NewRef, [H|Rest]):- \+ H =.. [_Pred, Id], replaceLexId(T, Id, NewRef, Rest). % Find a referent for one item % Finding words with "de", "het", and "een" is different find(lex(Word,_, Features, Id), Meaning, DRS, Ref, X):- member(Word, [de, het, een]), getMostInnerRef(lex(Word,_, Features, Id), Meaning, X), member(lex(OtherWord,OtherRefs, EatenWordFeats, X),Meaning), find(lex(OtherWord, OtherRefs, EatenWordFeats, X), Meaning, DRS, Ref, _). find(lex(Word,_, Features, Id), Meaning, DRS, Ref, Id):- \+ member(Word, [de, het, een]), DRS = drs(_Referents, Table), findLastRef(Table, Ref), checkFeatures(lex(Word, _, Features, Id),Meaning, DRS, Ref). find(lex(_,_,_,Id), _, _, _,Id). getMostInnerRef(lex(_, _, Features, Id), _, Ref):- \+ member(eat(_), Features), Ref = Id. getMostInnerRef(lex(_, _, Features, _), Meaning, Ref):- member(eat(X), Features), member(lex(A, B, C, X), Meaning), getMostInnerRef(lex(A,B,C,X), Meaning, Ref). %Input is a DRS predicate checkFeatures(lex(Word, _, Features, Id),Meaning, DRS, Ref):- \+ DRS = [_|_], DRS =.. [_Pred, [_|_], [H|T]], checkFeatures(lex(Word, _, Features, Id),Meaning, [H|T], Ref). %Input is a list checkFeatures(lex(Word, _, Features, Id),Meaning, DRS, Ref):- DRS = [_|_], checkFeatsList(lex(Word, _, Features, Id),Meaning, DRS, Ref). checkFeatures(lex(Word, _, Features, Id),Meaning, DRS, Ref):- member(X, DRS), X =.. [_Pred, [_|_]], checkFeatures(lex(Word, _, Features, Id),Meaning, X, Ref). checkFeatsList(lex(Word, _, _Features, _Id), _Meaning, List, Ref):- Pred =.. [Word, Ref], member(Pred, List). checkFeatsList(lex(Word, OtherRefs, Features, Id),Meaning, List, Ref):- member(Pred, List), Pred =.. [PredWord, Ref], lex(PredWord, _, DRSFeats, _), removeIrrelevantFeats(DRSFeats, RelDRSFeats), getFeatures(lex(Word, OtherRefs, Features, Id), Meaning, AllFeatures), removeIrrelevantFeats(AllFeatures, RelFeats), allMember(RelDRSFeats, RelFeats). relevantFeat(X):- \+ member(X, [noun(_), adj(_), prep(_), eat(_), find(_), bind(_), before(_), after(_), next(_), previous(_)]). removeIrrelevantFeats([], []). removeIrrelevantFeats([H|T], Rest):- \+ relevantFeat(H),!, removeIrrelevantFeats(T, Rest). removeIrrelevantFeats([H|T], [H|Rest]):- relevantFeat(H), removeIrrelevantFeats(T, Rest). %Input is a predicate findLastRef(X, Ref):- \+ X = [_|_], X =.. [_Pred|[H|T]], \+ H = [_|_], reverse([H|T], Reversed), member(Ref, Reversed). %Input is a DRS predicate findLastRef(DRS, Ref):- \+ DRS = [_,_], DRS =.. [_Pred, [_|_], [H|T]], reverse([H|T], Reversed), findLastRef(Reversed, Ref). %Input is a list findLastRef(Table, Ref):- Table = [_|_], reverse(Table, ReversedTable), member(X, ReversedTable), findLastRef(X, Ref). allMember([], _). allMember([H|T], B):- member(H, B), allMember(T, B). % Groep 3: context model gocontext :- retractall(tel(_)), assert(tel(1)), DRS = drs([], []), Meaning = [lex(hij, [], [find(1), pronoun(1), male(1), singular(1), animate(1)], 1), lex(had, [1, 3], [verb(2), pastTense(2), eat(1), eat(3), singular(1), noun(1), before(1), noun(3), after(3)], 2), lex(lange, [4], [adj(3), eat(4), noun(4), next(4), same_features(3, 4)], 3), lex(witte, [5], [adj(4), eat(5), noun(5), next(5), same_features(4, 5)], 4), lex(scherpe, [6], [adj(5), eat(6), noun(6), next(6), same_features(5, 6)], 5), lex(tanden, [], [noun(6), plural(6), inanimate(6)], 6), lex(en, [2, 18], [conn(7), eat(2), eat(18)], 7), lex(iedere, [9, 13], [univ(8), eat(9), noun(9), next(9), singular(9), same_features(8, 9), verb(13), eat(13)], 8), lex(avond, [], [noun(9), inanimate(9), singular(9), male(9)], 9), % lex(om, [8, 11], [prep(10), eat(8), before(8), noun(8), eat(11), noun(11), after(11)], 10), % lex(zes, [12], [det(11), eat(12), noun(12), next(12), same_features(11, 12)], 11), % lex(uur, [], [noun(12), neuter(12), inanimate(12)], 12), lex(stiet, [14, 15, 17], [verb(13), pastTense(13), eat(14), eat(15), eat(17), singular(14), animate(14), noun(15), after(15), particle(17)], 13), lex(hij, [], [find(14), noun(14), male(14), singular(14), animate(14)], 14), lex(een, [16], [det(15), bind(15), eat(16), next(16), noun(16), singular(16), same_features(15, 16)], 15), lex(gebrul, [], [noun(16), singular(16), neuter(16), inanimate(16)], 16), lex(uit, [], [particle(17)], 17) , lex(waarvan, [16, 23], [prep(18), eat(10), eat(23), prep(10), verb(23)], 18), lex(het, [20], [det(19), find(19), eat(20), next(20), neuter(20), same_features(19, 20)], 19), lex(hele, [21], [adj(20), eat(21), noun(21), next(21), same_features(20, 21)], 20), lex(muiskleurige, [22], [adj(21), noun(22), next(22), same_features(21, 22), eat(22)], 21) , lex(gebergte, [], [noun(22), neuter(22), singular(22)], 22), lex(sidderde, [19], [verb(23), pastTense(23), singular(19), before(19), noun(19), eat(19)], 23)], /* Meaning = [ lex(in, [14, 4], [prep(1), eat(14), eat(4), verb(14), prep(4), after(4)], 1), lex(het, [3], [det(2), find(2), eat(3), next(3), neuter(3), same_features(2, 3)], 2), lex(midden, [], [noun(3), neuter(3)], 3), lex(van, [2, 7], [prep(4), eat(2), eat(7), noun(2), prep(7), before(2), after(7)], 4) , lex(het, [6], [det(5), find(5), eat(6), next(6), neuter(6), same_features(5, 6)], 5) , lex(hartje, [], [noun(6), neuter(6)], 6) , lex(van, [5, 10], [prep(7), eat(5), eat(10), noun(5), prep(10), before(5), after(10)], 7) , lex(het, [9], [det(8), find(8), eat(9), next(9), neuter(9), same_features(8, 9)], 8) , lex(binnenste, [], [noun(9), neuter(9)], 9) , lex(van, [8, 11], [prep(10), eat(8), eat(11), noun(8), noun(11), before(8), after(11)], 10) , lex(het, [12], [det(11), find(11), eat(12), next(12), neuter(12), same_features(11, 12)], 11) , lex(muiskleurig, [13], [adj(12), eat(13), noun(13), next(13), same_features(12, 13)], 12) , lex(gebergte, [], [noun(13), neuter(13), singular(13)], 13) , lex(woonde, [15], [verb(14), singular(14), past(14), eat(15), noun(15), animate(15)], 14) , lex(de, [16], [det(15), find(15), eat(16), noun(16), next(16), male(16), same_features(15, 16)], 15) , lex(weerwolf, [], [noun(16), animate(16), male(16)], 16)], */ docontext(Meaning, Meaning, DRS, NewDRS), writeln(NewDRS). gocontext1 :- M=[lex(hij, [], [find(1), noun(1), male(1), singular(1), animate(1)], 1), lex(was, [1, 4], [verb(2), pastTense(2), eat(1), eat(4), singular(1), noun(1), before(1), prep(4), after(4)], 2), lex(groot, [], [noun(3)], 3), lex(en, [3, 5], [npconj(4), eat(3), eat(5), noun(3), before(3), noun(5), after(5)], 4), lex(verschrikkelijk, [], [noun(5)], 5)], docontext(M,M,drs([2, 5, 8, 11, 15], [in(15, 2), midden(2), van(2, 5), hartje(5), van(5,8), binnenste(8), van(8, 11), gebergte(11), muiskleurig(11), woonde(15),weerwolf(15)]),ND), writeln(ND). %bind new referents dobind([], _, DRS, DRS). dobind([lex(Word, Vars, Feats, Id) | Rest], M, drs(Ref, Cond), drs(NewRef, Cond)) :- dobind(Rest, M, drs(Ref, Cond), drs(R, Cond)), ((member(bind(V), Feats) ; (member(find(V), Feats), var(V))) -> V = Id, %Add variable to M: member(lex(Word, Vars, Feats, Id), M), %Add new variable to Ref: union(R, [V], NewRef) ; % ?! nonvar(Ref), NewRef = R). % wrapper docontext(Meaning, Meaning, DRS, NewDRS) :- dobind(Meaning, Meaning, DRS, Tmp), Tmp = drs(Ref, Cond), %handle negation ((select(lex(_, _, C, _), Meaning, Rest), member(neg(_), C)) -> docontext(Rest, Rest, drs([], []), drs(NewRef, TempCond)), subtract(TempCond, Cond, NewCond), NewDRS = drs(Ref, [not(drs(NewRef, NewCond)) | Cond]) ; % handle sentence conjunction ((select(lex(_, _, D, _), Meaning, _Rest), member(conn(_), D)) -> append(A, [lex(_,_,D,_)|B], Meaning), docontext(A, A, DRS, drs(NewRef1, TempCond1)), docontext(B, B, DRS, drs(NewRef2, TempCond2)), % merge DRSes union(NewRef1, NewRef2, NewRef3), union(NewRef3, Ref, NewRef), union(TempCond1, TempCond2, TempCond3), union(TempCond3, Cond, NewCond), NewDRS = drs(NewRef, NewCond) ; % handle universal quantification ((select(lex(_, _, E, _), Meaning, Rest), member(univ(_), E)) -> select(eat(A), E, ER), %variable select(eat(B), ER, _), %sentence select(lex(W,_V,_F,A), Rest, RRest), %one-word hack for variable: Pred =.. [W, A], DRS1 = drs([A], [Pred]), %AM = [lex(W,V,F,A)], %plus more words? %eg. iedere donkere avond %docontext(AM, AM, drs([],[]), DRS1), docontext(RRest, RRest, drs([],[]), DRS2), DRS2 = drs(Ref2, Cond2), subtract(Ref2, Ref, NRef2), subtract(Cond2, Cond, Cond3), union(Cond, [forall(DRS1, drs(NRef2, Cond3))], NewCond), NewDRS = drs(Ref, NewCond) ; %normal sentence, stop recursing context(Meaning, Meaning, Tmp, NewDRS)))). context([], _, NewDRS, NewDRS). context([lex(Word, Vars, Feats, Id) | Rest], M, drs(Ref, Cond), Result) :- % add conditions to DRS % normal words, ie., neither a determiner or pronoun, has arguments: (\+ (member(det(_), Feats) ; member(npconj(_), Feats) ; member(pronoun(_), Feats) ; Vars = []) -> % handle NP conjunctions and copula (member(eat(X), Feats), member(lex(_,_,XFeats,X), M), member(npconj(X), XFeats) -> select(eat(A), XFeats, XRest), select(eat(B), XRest, _), (Vars = [Y, X] -> XVars1 = [Y, A], XVars2 = [Y, B] ; XVars1 = [A, Y], XVars2 = [B, Y]), ((deref(XVars1, M, DerefVars1), deref(XVars2, M, DerefVars2)) -> Pred1 =.. [Word | DerefVars1], Pred2 =.. [Word | DerefVars2], union(Ref, DerefVars1, Tmp), union(Tmp, DerefVars2, NewRef) ; %copula, verb has no meaning member(lex(Word1, _,_, A), M), member(lex(Word2, _,_, B), M), Pred1 =.. [Word1, Y], Pred2 =.. [Word2, Y], union(Ref, [Y], NewRef)), union(Cond, [Pred1, Pred2], NewCond) ; %resolve word IDs to discourse referents (deref(Vars, M, DerefVars) -> Pred =.. [Word | DerefVars], union(Ref, DerefVars, NewRef) ; %copula, verb has no meaning % eg. hij was groot: % get ref from 'hij', condition from 'groot' select(eat(A), Feats, XRest), select(eat(B), XRest, _), member(lex(_W,_,AF,A), M), member(lex(Word,_,_,B), M), member(bind(A), AF), Pred =.. [Word, A], union(Ref, [A], NewRef)), union(Cond, [Pred], NewCond)), context(Rest, M, drs(NewRef, NewCond), Result) ; %handle determiners ((member(det(_), Feats) ; member(pronoun(_), Feats)) -> (member(find(F), Feats) ; member(bind(F), Feats)), %recurse until word which hasn't eaten (noun) findconditions(F, Feats, M, Preds), union(Cond, Preds, NewCond), context(Rest, M, drs(Ref, NewCond), Result) ; % a word which hasn't eaten anything but is % eaten directly by a verb should bind a new referent % and be added as condition. ((member(lex(_,_,VFeat,_), M), member(eat(Id), VFeat), (member(verb(_), VFeat) ; member(prep(_), VFeat))) -> Pred =.. [Word, Id], union(Cond, [Pred], NewCond), union(Ref, [Id], NewRef), context(Rest, M, drs(NewRef, NewCond), Result) ; % other words without arguments are ignored: context(Rest, M, drs(Ref, Cond), Result)))). % eg. de -> rode -> kat should yield [kat, rode] findconditions(_F, Feats, _Meaning, []) :- \+ member(eat(_X), Feats). findconditions(F, Feats, Meaning, [Pred | Tail]) :- member(eat(X), Feats), select(lex(AteWord, _, NFeats, X), Meaning, Rest), Pred =.. [AteWord, F], findconditions(F, NFeats, Rest, Tail). % dereference word IDs to their original discourse referents % (ie., word ID from when referent was introduced) deref([], _Meaning, []). deref([V | Vars], Meaning, [R | RVars]) :- deref(Vars, Meaning, RVars), %find word referred to: select(lex(_, _, Feats, V), Meaning, Rest), %either the discourse referent has been found: (member(find(R), Feats) ; %or it has or will be bound: member(bind(R), Feats) ; % implicit bind: bare nouns (member(lex(_,_, Feats2, _), Meaning), member(eat(V), Feats2), member(verb(_), Feats2), R = V) ; %recurse (determiner contains find/bind). (\+ member(bind(_), Feats), \+ member(find(_), Feats), ((member(verb(_), Feats) ; member(prep(_), Feats) ; member(npconj(_), Feats) ; member(adv(_), Feats)) -> % resolve verb to one of its arguments %(which argument is part of ambiguity of language...) member(eat(Foo), Feats) ; %backwards: eg. kat -> rode -> de member(lex(_,_, Feats2, Foo), Meaning), member(eat(V), Feats2)), % prevent cycles by passing Rest, ie., without the word we just tried: deref([Foo], Rest, [R]))).