This repository has been archived on 2023-08-20. You can view files and clone it, but cannot push or open issues or pull requests.
yap-6.3/packages/cplint/lemur/revise_lemur.pl
Fabrizio Riguzzi b25c9e5b61 lemur
2014-10-15 15:15:46 +02:00

637 lines
13 KiB
Prolog

/*
EMBLEM and SLIPCASE
Copyright (c) 2011, Fabrizio Riguzzi, Nicola di Mauro and Elena Bellodi
*/
:- use_module(library(terms)).
:- use_module(library(lists)).
:- use_module(library(random)).
:- set_prolog_flag(discontiguous_warnings,on).
:- set_prolog_flag(single_var_warnings,on).
:- [dv_lemur].
theory_revisions_op(Theory,TheoryRevs):-
setof(RevOp, Theory^revise_theory(Theory,RevOp), TheoryRevs),!.
theory_revisions_op(_Theory,[]).
filter_add_rule([],[]).
filter_add_rule([add(Rule)|R],R1):-
!,
filter_add_rule(R,R1).
filter_add_rule([A|R],[A|R1]):-
!,
filter_add_rule(R,R1).
theory_revisions_r(Theory,TheoryRevs):-
theory_revisions_op(Theory,TheoryRevs1),
% filter_add_rule(TheoryRevs11,TheoryRevs1),
( TheoryRevs1 == [] ->
TheoryRevs = []
;
length(TheoryRevs1,L),
random(0,L,K),
nth0(K, TheoryRevs1,Revision),
apply_operators([Revision],Theory,TheoryRevs)
).
theory_revisions(Theory,TheoryRevs):-
theory_revisions_op(Theory,TheoryRevs1),
apply_operators(TheoryRevs1,Theory,TheoryRevs).
apply_operators([],_Theory,[]).
apply_operators([add(Rule)|RestOps],Theory,[NewTheory|RestTheory]) :-
append(Theory, [Rule], NewTheory),
% nl,write(NewTheory),
apply_operators(RestOps,Theory,RestTheory).
apply_operators([add_body(Rule1,Rule2,_A)|RestOps],Theory,[NewTheory|RestTheory]) :-
delete_matching(Theory,Rule1,Theory1),
append(Theory1, [Rule2], NewTheory),
% nl,write(NewTheory),
apply_operators(RestOps,Theory,RestTheory).
apply_operators([remove_body(Rule1,Rule2,_A)|RestOps],Theory,[NewTheory|RestTheory]) :-
delete_matching(Theory,Rule1,Theory1),
append(Theory1, [Rule2], NewTheory),
% nl,write(NewTheory),
apply_operators(RestOps,Theory,RestTheory).
apply_operators([add_head(Rule1,Rule2,_A)|RestOps],Theory,[NewTheory|RestTheory]) :-
delete_matching(Theory,Rule1,Theory1),
append(Theory1, [Rule2], NewTheory),
% nl,write(NewTheory),
apply_operators(RestOps,Theory,RestTheory).
apply_operators([remove_head(Rule1,Rule2,_A)|RestOps],Theory,[NewTheory|RestTheory]) :-
delete_matching(Theory,Rule1,Theory1),
append(Theory1, [Rule2], NewTheory),
% nl,write(NewTheory),
apply_operators(RestOps,Theory,RestTheory).
apply_operators([remove(Rule)|RestOps],Theory,[NewTheory|RestTheory]) :-
delete_matching(Theory,Rule,NewTheory),
% nl,write(NewTheory),
apply_operators(RestOps,Theory,RestTheory).
revise_theory(Theory,Ref):-
specialize_theory(Theory,Ref).
revise_theory(Theory,Ref):-
generalize_theory(Theory,Ref).
/*
generalize_theory(Theory,Ref):-
Theory \== [],
choose_rule(Theory,Rule),
generalize_rule(Rule,Ref).
*/
generalize_theory(Theory,Ref):-
length(Theory,LT),
setting(max_rules,MR),
LT<MR,
add_rule(Ref).
generalize_rule(Rule,Ref):-
generalize_head(Rule,Ref).
generalize_rule(Rule,Ref):-
generalize_body(Rule,Ref).
add_rule(add(rule(ID,Head,[],Lits))):-
setting(specialization,bottom),!,
database(DB),
sample(1,DB,[M]),
get_head_atoms(O),
member(A,O),
functor(A,F,N),
functor(F1,F,N),
F1=..[F|Arg],
Pred1=..[F,M|Arg],
A=..[F|ArgM],
keep_const(ArgM,Arg),
findall((A,Pred1),call(Pred1),L),
sample(1,L,LH),
generate_body(LH,[rule(ID,Head,[],Lits)]).
add_rule(add(SpecRule)):-
findall(HL , modeh(_,HL), HLS),
length(HLS,L),
L1 is L+1,
P is 1/L1,
generate_head(HLS,P,Head),
get_next_rule_number(ID),
Rule0 = rule(ID,Head,[],true),
specialize_rule(Rule0,SpecRule,Lit).
generate_head([H|_T],_P,[H1:0.5,'':0.5]):-
H=..[Pred|Args],
length(Args,LA),
length(Args1,LA),
H1=..[Pred|Args1],
check_for_constants(Args,Args1).
check_for_constants([],[]).
check_for_constants([+X|R],[V|R1]):-
!,
check_for_constants(R,R1).
check_for_constants([-X|R],[V|R1]):-
!,
check_for_constants(R,R1).
check_for_constants([X|R],[X|R1]):-
check_for_constants(R,R1).
generate_head([_H|T],P,Head):-
generate_head(T,P,Head).
generalize_head(Rule,Ref):-
Rule = rule(ID,LH,BL,L),
generalize_head1(LH,LH1,NewAt),
Ref = add_head(Rule,rule(ID,LH1,BL,L),NewAt).
generalize_head1(LH,LH1,NH):-
findall(HL , modeh(_,HL), HLS),
generalize_head2(HLS,LH,LH1,NH).
generalize_head2([X|_R],LH,LH1,PH) :-
X =.. [P|A],
length(A,LA),
length(A1,LA),
PH =.. [P|A1],
\+ member(PH:_, LH),
(setting(new_head_atoms_zero_prob,true)->
delete_matching(LH,'':PNull,LH0),
append(LH0,[PH:0.0,'':PNull],LH1)
;
length(LH,NH),
add_to_head(LH,NH,PH,LH1)
).
generalize_head2([_X|R],LH,LH1) :-
generalize_head2(R,LH,LH1).
add_to_head(['':PN],NH,At,[At:PA,'':PN1]):-!,
PN1 is PN*NH/(NH+1),
PA is 1/(NH+1).
add_to_head([H:PH|T],NH,At,[H:PH1|T1]):-
PH1 is PH*NH/(NH+1),
add_to_head(T,NH,At,T1).
get_module_var(LH,Module):-
member(H:_,LH),!,
H=..[_F,Module|_].
generalize_body(Rule,Ref):-
Rule = rule(ID,LH,BL),
delete_one(BL,BL1,A),
remove_prob(LH,LH1),
delete(LH1,'',LH2),
linked_clause(BL1,LH2),
Ref = remove_body(Rule,rule(ID,LH,BL1),A).
specialize_theory(Theory,Ref):-
Theory \== [],
choose_rule(Theory,Rule),
specialize_rule(Rule,SpecRule,Lit),
Ref = add_body(Rule,SpecRule,Lit),
SpecRule = rule(_,_,B,_).
/*,
\+ (member(b_rel11(X1,Y1),B), member(b_rel11(Z1,Y1),B), Y1 \== Z1),
\+ (member(b_rel12(X2,Y2),B), member(b_rel12(Z2,Y2),B), Y2 \== Z2),
\+ (member(b_rel13(X3,Y3),B), member(b_rel13(Z3,Y3),B), Y3 \== Z3).*/
specialize_rule(Rule,SpecRule,Lit):-
setting(specialization,bottom),
Rule = rule(ID,LH,BL,Lits),
delete_one(Lits,RLits,Lit),
\+ lookahead_cons(Lit,_),
append(BL,[Lit],BL1),
remove_prob(LH,LH1),
% check_ref(LH1,BL1),
delete(LH1,'',LH2),
append(LH2,BL1,ALL2),
extract_fancy_vars(ALL2,Vars1),
length(Vars1,NV),
setting(max_var,MV),
NV=<MV,
linked_clause(BL1,LH2),
\+ banned_clause(LH2,BL1),
SpecRule=rule(ID,LH,BL1,RLits).
specialize_rule(Rule,SpecRule,Lit):-
setting(specialization,bottom),
Rule = rule(ID,LH,BL,Lits),
delete_one(Lits,RLits,Lit),
append(BL,[Lit],BL0),
(lookahead(Lit,LLit1);lookahead_cons(Lit,LLit1)), % lookahead_cons serve a dire che rating(_A,_B,_C) e' aggiunto solo insieme ai letterali indicati nella lista, mai da solo.
copy_term(LLit1,LLit2),
specialize_rule_la_bot(LLit2,RLits,RLits1,BL0,BL1),
remove_prob(LH,LH1),
% check_ref(LH1,BL1),
delete(LH1,'',LH2),
append(LH2,BL1,ALL2),
extract_fancy_vars(ALL2,Vars1),
length(Vars1,NV),
setting(max_var,MV),
NV=<MV,
linked_clause(BL1,LH2),
\+ banned_clause(LH2,BL1),
SpecRule=rule(ID,LH,BL1,RLits1).
specialize_rule(Rule,SpecRule,Lit):-
setting(specialization,mode),!,
% findall(BL , modeb(_,BL), BLS),
mcts_modeb(BSL0),
Rule = rule(ID,LH,BL,_),
( BL \= [] ->
%last(BL,LastLit),
%LastLit =.. [Pred|_],
%filter_modeb(BSL0,LastLit,BSL)
BSL = BSL0
;
BSL = BSL0
),
specialize_rule(BSL,Rule,SpecRule,Lit).
filter_modeb([],_Pred,[]).
filter_modeb([Modeb|RestModeb],Pred,[Modeb|RestBSL]):-
Modeb =.. [PredMode|_],
Modeb @>= Pred,
!,
filter_modeb(RestModeb,Pred,RestBSL).
filter_modeb([_|RestModeb],Pred,RestBSL):-
filter_modeb(RestModeb,Pred,RestBSL).
skolemize(Theory,Theory1):-
copy_term(Theory,Theory1),
term_variables(Theory1,Vars),
skolemize1(Vars,1).
skolemize1([],_).
skolemize1([Var|R],K):-
atomic_list_concat([s,K],Skolem),
Var = Skolem,
K1 is K + 1,
skolemize1(R,K1).
banned_clause(H,B):-
skolemize([H,B],[H1,B1]),
banned(H2,B2),
mysublist(H2,H1),
mysublist(B2,B1).
mysublist([],_).
mysublist([A\==B|T],L):-
!,
A\==B,
mysublist(T,L).
mysublist([H|T],L):-
nth(_,L,H,R),
mysublist(T,R).
check_ref(H,B):-
copy_term((H,B),(H1,B1)),
numbervars((H1,B1),0,_N),
(ref(H1,B1)->
fail
;
assert(ref(H1,B1))
).
specialize_rule([Lit|_RLit],Rule,SpecRul,SLit):-
Rule = rule(ID,LH,BL,true),
remove_prob(LH,LH1),
append(LH1,BL,ALL),
specialize_rule1(Lit,ALL,SLit),
append(BL,[SLit],BL1),
(lookahead(SLit,LLit1);lookahead_cons(SLit,LLit1)),
specialize_rule_la(LLit1,LH1,BL1,BL2),
append(LH1,BL2,ALL2),
extract_fancy_vars(ALL2,Vars1),
length(Vars1,NV),
setting(max_var,MV),
NV=<MV,
\+ banned_clause(LH1,BL2),
SpecRul = rule(ID,LH,BL2,true).
specialize_rule([Lit|_RLit],Rule,SpecRul,SLit):-
Rule = rule(ID,LH,BL,true),
remove_prob(LH,LH1),
append(LH1,BL,ALL),
specialize_rule1(Lit,ALL,SLit),
% \+ member(SLit,LH1), %%%%
\+ lookahead_cons(SLit,_),
append(BL,[SLit],BL1),
append(LH1,BL1,ALL1),
% dv(LH1,BL1,DList), %var,depth list DList in output
extract_fancy_vars(ALL1,Vars1),
length(Vars1,NV),
setting(max_var,MV),
NV=<MV,
setting(maxdepth_var,MD),
% exceed_depth(DList,MD), %fallisce se una sottolista eccede MD
\+ banned_clause(LH1,BL1),
SpecRul = rule(ID,LH,BL1,true).
specialize_rule([_|RLit],Rule,SpecRul,Lit):-
specialize_rule(RLit,Rule,SpecRul,Lit).
specialize_rule_la([],_LH1,BL1,BL1).
specialize_rule_la([Lit1|T],LH1,BL1,BL3):-
copy_term(Lit1,Lit2),
modeb(_,Lit2),
append(LH1,BL1,ALL1),
specialize_rule1(Lit2,ALL1,SLit1),
append(BL1,[SLit1],BL2),
specialize_rule_la(T,LH1,BL2,BL3).
specialize_rule_la_bot([],Bot,Bot,BL,BL).
specialize_rule_la_bot([Lit|T],Bot0,Bot,BL1,BL3):-
delete_one(Bot0,Bot1,Lit),
append(BL1,[Lit],BL2),
specialize_rule_la_bot(T,Bot1,Bot,BL2,BL3).
remove_prob(['':_P],[]):-!.
remove_prob([X:_|R],[X|R1]):-
remove_prob(R,R1).
specialize_rule1(Lit,Lits,SpecLit):-
Lit =.. [Pred|Args],
exctract_type_vars(Lits,TypeVars0),
remove_duplicates(TypeVars0,TypeVars),
take_var_args(Args,TypeVars,Args1),
SpecLit =.. [Pred|Args1],
\+ member_eq(SpecLit,Lits).
convert_to_input_vars([],[]):-!.
convert_to_input_vars([+T|RT],[+T|RT1]):-
!,
convert_to_input_vars(RT,RT1).
convert_to_input_vars([-T|RT],[+T|RT1]):-
convert_to_input_vars(RT,RT1).
member_eq(X,[Y|_List]) :-
X == Y.
member_eq(X,[_|List]) :-
member_eq(X,List).
remove_eq(X,[Y|R],R):-
X == Y,
!.
remove_eq(X,[_|R],R1):-
remove_eq(X,R,R1).
linked_clause(X):-
linked_clause(X,[]).
linked_clause([],_).
linked_clause([L|R],PrevLits):-
term_variables(PrevLits,PrevVars),
input_variables(L,InputVars),
linked(InputVars,PrevVars),!,
linked_clause(R,[L|PrevLits]).
linked([],_).
linked([X|R],L) :-
member_eq(X,L),
!,
linked(R,L).
input_variables(\+ LitM,InputVars):-
!,
LitM=..[P|Args],
length(Args,LA),
length(Args1,LA),
Lit1=..[P|Args1],
copy_term(LitM,Lit0),
modeb(_,Lit1),
Lit1 =.. [P|Args1],
convert_to_input_vars(Args1,Args2),
Lit2 =.. [P|Args2],
input_vars(Lit0,Lit2,InputVars).
input_variables(LitM,InputVars):-
LitM=..[P|Args],
length(Args,LA),
length(Args1,LA),
Lit1=..[P|Args1],
modeb(_,Lit1),
input_vars(LitM,Lit1,InputVars).
input_variables(LitM,InputVars):-
LitM=..[P|Args],
length(Args,LA),
length(Args1,LA),
Lit1=..[P|Args1],
modeh(_,Lit1),
input_vars(LitM,Lit1,InputVars).
input_vars(Lit,Lit1,InputVars):-
Lit =.. [_|Vars],
Lit1 =.. [_|Types],
input_vars1(Vars,Types,InputVars).
input_vars1([],_,[]).
input_vars1([V|RV],[+_T|RT],[V|RV1]):-
!,
input_vars1(RV,RT,RV1).
input_vars1([_V|RV],[_|RT],RV1):-
input_vars1(RV,RT,RV1).
exctract_type_vars([],[]).
exctract_type_vars([Lit|RestLit],TypeVars):-
Lit =.. [Pred|Args],
length(Args,L),
length(Args1,L),
Lit1 =.. [Pred|Args1],
take_mode(Lit1),
type_vars(Args,Args1,Types),
exctract_type_vars(RestLit,TypeVars0),
!,
append(Types,TypeVars0,TypeVars).
take_mode(Lit):-
modeh(_,Lit),!.
take_mode(Lit):-
modeb(_,Lit),!.
take_mode(Lit):-
mode(_,Lit),!.
type_vars([],[],[]).
type_vars([V|RV],[+T|RT],[V=T|RTV]):-
!,
type_vars(RV,RT,RTV).
type_vars([V|RV],[-T|RT],[V=T|RTV]):-atom(T),!,
type_vars(RV,RT,RTV).
type_vars([_V|RV],[_T|RT],RTV):-
type_vars(RV,RT,RTV).
take_var_args([],_,[]).
take_var_args([+T|RT],TypeVars,[V|RV]):-
!,
member(V=T,TypeVars),
take_var_args(RT,TypeVars,RV).
take_var_args([-T|RT],TypeVars,[_V|RV]):-
atom(T),
take_var_args(RT,TypeVars,RV).
take_var_args([-T|RT],TypeVars,[V|RV]):-
member(V=T,TypeVars),
take_var_args(RT,TypeVars,RV).
take_var_args([T|RT],TypeVars,[T|RV]):-
T\= + _,(T\= - _; T= - A,number(A)),
take_var_args(RT,TypeVars,RV).
choose_rule(Theory,Rule):-
( setting(mcts_covering,true) ->
mcts_restart(Restart),
nth(K,Theory,Rule),
K >= Restart
;
member(Rule,Theory)
).
%last(Theory,Rule).
add_rule(Theory,add(rule(ID,H,[],true))):-
new_id(ID),
findall(HL , modeh(_,HL), HLS),
length(HLS,NH),
P is 1/(NH+1),
add_probs(HLS,H,P),
\+ member(rule(_,H,[],true),Theory).
add_rule(Theory,TheoryGen):-
findall(HL , modeh(_,HL), HLS),
add_rule(HLS,Theory,TheoryGen).
add_rule([X|_R],Theory,TheoryGen) :-
new_id(ID),
X =.. [P|A],
length(A,LA),
length(A1,LA),
PH =.. [P|A1],
TheoryGen = add(rule(ID,[PH:0.5,'':0.5],[],true)),
\+ member(rule(_,[PH:_,'':_],[],true),Theory).
add_rule([_X|R],Theory,TheoryGen) :-
add_rule(R,Theory,TheoryGen).
add_probs([],['':P],P):-!.
add_probs([H|T],[H:P|T1],P):-
add_probs(T,T1,P).
extract_fancy_vars(List,Vars):-
term_variables(List,Vars0),
fancy_vars(Vars0,1,Vars).
fancy_vars([],_,[]).
fancy_vars([X|R],N,[NN2=X|R1]):-
name(N,NN),
append([86],NN,NN1),
name(NN2,NN1),
N1 is N + 1,
fancy_vars(R,N1,R1).
delete_one([X|R],R,X).
delete_one([X|R],[X|R1],D):-
delete_one(R,R1,D).
remove_last([_X],[]) :-
!.
remove_last([X|R],[X|R1]):-
remove_last(R,R1).
delete_matching([],_El,[]).
delete_matching([El|T],El,T1):-!,
delete_matching(T,El,T1).
delete_matching([H|T],El,[H|T1]):-
delete_matching(T,El,T1).