diff --git a/evidence.pl b/evidence.pl
index b594d23ff672a809c52c45e2335c072101b5c413..b967db319521c2808984882b15e155712ad0f763 100644
--- a/evidence.pl
+++ b/evidence.pl
@@ -64,16 +64,24 @@ combine([Mass1|Masses1], Masses2, RMasses, Opts) :-
append(RMasses1, RMasses2, RMasses).
combine1(_, [], [], _).
-combine1((Set1, Mass1), [(Set2, _)|L], Result, Opts) :-
+combine1((Label1, Set1, Mass1), [(_, Set2, _)|L], Result, Opts) :-
+ % empty intersection case
ord_intersect(Set1, Set2, []), % empty intersection case
trace('~w inter ~w = emptyset~n', [Set1, Set2], Opts, 2),
- combine1((Set1, Mass1), L, Result, Opts).
-combine1((Set1, Mass1), [(Set2, Mass2)|L], [(RSet, RMass)|Result], Opts) :-
+ combine1((Label1, Set1, Mass1), L, Result, Opts).
+combine1((Label1, Set1, Mass1), [(Label2, Set2, Mass2)|L], [(RLabel, RSet, RMass)|Result], Opts) :-
ord_intersect(Set1, Set2, RSet),
RSet \= [], % nonempty intersection case
RMass is Mass1 * Mass2,
+ ( memberchk(0, Label1)
+ -> RLabel = Label2
+ ; ( memberchk(0, Label2)
+ -> RLabel = Label1
+ ; ord_union(Label1, Label2, RLabel)
+ )
+ ),
trace('~w inter ~w = ~w : ~w~n', [Set1, Set2, RSet, RMass], Opts, 2),
- combine1((Set1, Mass1), L, Result, Opts).
+ combine1((Label1, Set1, Mass1), L, Result, Opts).
%! normalize(+BBA, -NormBBA) is det
%
@@ -94,22 +102,24 @@ normalize(M1, M2) :-
%
% Merge duplicate sets and add their masses.
merge_unique([], []).
-merge_unique([(Set, M1)|B1], [(Set, M2)|B3]) :-
- collect_masses_unique(Set, B1, Masses, B2),
+merge_unique([(Label, Set, M1)|B1], [(Label, Set, M2)|B3]) :-
+ collect_masses_unique((Label, Set), B1, Masses, B2),
M2 is M1 + Masses,
merge_unique(B2, B3).
-%! collect_masses_unique(+Set, +BBas, -Masses, -B2) is det
+%! collect_masses_unique(+LabelSet, +BBas, -Masses, -B2) is det
%
% Masses is the sum of all masses of be belief assignments in BBAs
-% concerning set Set. B2 containts the remaining assignments.
+% concerning (Label, Set). B2 containts the remaining
+% assignments.
collect_masses_unique(_, [], 0, []).
-collect_masses_unique(Set, [(Set, M1)|BBAs], Masses, FinalBBAs) :-
- collect_masses_unique(Set, BBAs, M2, FinalBBAs),
+collect_masses_unique((Label, Set), [(Label, Set, M1)|BBAs], Masses, FinalBBAs) :-
+ collect_masses_unique((Label, Set), BBAs, M2, FinalBBAs),
Masses is M1 + M2.
-collect_masses_unique(Set, [(S1, M1)|BBAs], Masses, [(S1, M1)|FinalBBAs]) :-
- Set \== S1,
- collect_masses_unique(Set, BBAs, Masses, FinalBBAs).
+collect_masses_unique((Label, Set), [(Label1, S1, M1)|BBAs], Masses, [(Label1, S1, M1)|FinalBBAs]) :-
+ ( Set \== S1 ; Label \== Label1
+ ),
+ collect_masses_unique((Label, Set), BBAs, Masses, FinalBBAs).
%! add_all(+L, -Sum) is semidet
%
@@ -124,7 +134,7 @@ add_all([N|L], S) :-
%
% Sum is the sum of the masses in IMasses BBA
sum_mass([], 0).
-sum_mass([(_,M)|L], S) :-
+sum_mass([(_, _, M)|L], S) :-
sum_mass(L, S1),
S is S1 + M.
@@ -132,7 +142,7 @@ sum_mass([(_,M)|L], S) :-
%
% Multiply a BBA L1 by a factor F.
mult(_, [], []).
-mult(F, [(S1,N1)|L1], [(S1,N2)|L2]) :-
+mult(F, [(Lb1, S1, N1)|L1], [(Lb1, S1, N2)|L2]) :-
N2 is N1 * F,
mult(F, L1, L2).
@@ -171,27 +181,30 @@ global_bba2(B, [BBa|BBas], GBBA, Opts) :-
%
% Bel is the set of beliefs for each Bi, given the global BBA GBBA.
beliefs([], _, []).
-beliefs([Bi|Bis], GBBA, [(Bi, Bel)|Bels]) :-
- belief(Bi, GBBA, Bel),
+beliefs([Ni/Bi|Bis], GBBA, [(Ni, Bi, Bel)|Bels]) :-
+ belief(Ni/Bi, GBBA, Bel),
beliefs(Bis, GBBA, Bels).
belief(_, [], 0).
-belief(Bi, [(Bj, M)|GBBA], Bel) :-
+belief(Ni/Bi, [(Nj, Bj, M)|GBBA], Bel) :-
ord_subset(Bj, Bi),
- belief(Bi, GBBA, Bel1),
+ memberchk(Ni, Nj),
+ belief(Ni/Bi, GBBA, Bel1),
Bel is M + Bel1.
-belief(Bi, [(Bj, _)|GBBA], Bel) :-
- \+ ord_subset(Bj, Bi),
- belief(Bi, GBBA, Bel).
+belief(Ni/Bi, [(Nj, Bj, _)|GBBA], Bel) :-
+ ( \+ ord_subset(Bj, Bi) ; \+ memberchk(Ni, Nj) ),
+ belief(Ni/Bi, GBBA, Bel).
-%! print_beliefs(+Bels) is semidet
+
+%! print_beliefs(+Bels, +Assoc) is semidet
+%
+% essentially used for debugging.
print_beliefs([]).
-print_beliefs([(Bi, Bel)|Bels]) :-
- format('Bel(~w)=~w~n', [Bi, Bel]),
+print_beliefs([(Li, Bi, Bel)|Bels]) :-
+ format('Bel(~w={~w})=~w~n', [Li, Bi, Bel]),
print_beliefs(Bels).
-
/**********************************************************************/
/* test data */
/**********************************************************************/
diff --git a/wbel.pl b/wbel.pl
index fe5b68bf9cc0e9e7c0ffc8e9601ec78bd2abeaea..17516956883a47d7923d18101df105a17c89cdf4 100644
--- a/wbel.pl
+++ b/wbel.pl
@@ -152,14 +152,15 @@ print_raw_beliefs([(Bi, Bel)|Bels]) :-
format('Bel(~w)=~w~n', [Bi, Bel]),
print_raw_beliefs(Bels).
-%! print_beliefs(+Bels,+Assoc) is semidet
+%! print_beliefs(+Bels, +Assoc) is semidet
print_beliefs([], _).
-print_beliefs([(Bi, Bel)|Bels], Assoc) :-
- expand_formulae(Bi, Assoc, FBi),
- format('Bel(~w)=~w~n', [FBi, Bel]),
+print_beliefs([(Li, Bi, Bel)|Bels], Assoc) :-
+ expand_formulae(Bi, Assoc, Forms),
+ format('Bel(~w={~w})=~w~n', [Li, Forms, Bel]),
print_beliefs(Bels, Assoc).
+
%! print_wincl(+Bels,+Assoc) is semidet
%
% print sets of formulas.
@@ -201,8 +202,8 @@ generate_b1(N, CountDown, [T|L], Func) :-
%
% collect ASP answer sets
% Each answer set represent subbase of B which is maximaly consistent
-% with mu. Number each base, that is, the results is a list of
-% i/List terms.
+% with mu. Number each base (starting from 1), that is, the results is
+% a list of i/List terms.
collect_results([], [], _) :- !.
collect_results([as(L)|L1], [N/SL|L2], N) :-
!,