I want to sum in SWI-Prolog depending on certain user answers - prolog

Question

I need to sum in a variable depending on certain user answers and I'm starting to know Prolog syntax and paradigm.
Right now, I can read user data, and print it too, but I lack a way to accumulate the results, because right now the results are inconsistent.
What I have now is:
inicio :-
write('¿You have overweight yes/no?'),
read(R1),
write('¿Are you a smoker yes/no?'),
read(R2),
write('¿Do you have some direct relative with diabetes?'),
read(R3),
Risk is 0,
( R1 = yes -> Risk is Risk + 2 ; Risk is Risk + 0 ),
imprimir(['The result is ', Risk]),
( R2 = yes -> Risk is Risk + 1 ; Risk is Risk + 0 ),
imprimir(['The result is ', Risk]),
( R3 = yes -> Risk is Risk + 3 ; Risk is Risk + 0 ),
imprimir(['The result is ', Risk]).
imprimir([]).
imprimir([Term| Terms]) :-
write(Term),
imprimir(Terms).

Answer 1

I'm going to show you a fundamentally different way of approaching this program that leverages Prolog a little better. First, let's make a table of the penalties. Making tables for your program's configuration is often a useful thing to do:
risk_penalty(overweight, 2).
risk_penalty(smoker, 1).
risk_penalty(diabetes, 3).
Now that we have a uniform way of thinking about the problem, let's see if we can make a uniform way of getting information from the user. Let's use the dynamic store to keep track of what the user has told us, because it will simplify querying later:
:- dynamic risk/2.
ask(Prompt, Fact) :-
format('~a [yes/no]> ', [Prompt]),
read(Response),
assertz(risk(Fact, Response)).
Now we have a little predicate we can use to interview the user. This kind of print-read-assert function is pretty common in small expert systems like yours, because it helps you separate the logic of the system from its front-end. When you do ask('Do you have X?', has_x), the dynamic store will either receive risk(has_x, yes) or risk(has_x, no) depending on which the user has entered. It also gives you a natural place to make the user input more robust, by checking it and re-asking if you get something weird.
Now we can do your initial loop a little more cleanly:
inicio :-
ask('Are you overweight?', overweight),
ask('Are you a smoker?', smoker),
ask('Do you have some direct relative with diabetes?', diabetes).
This just does the interview portion. Now if you run through it once, say answering "yes", "no", "yes", then the database will contain these facts:
?- risk(Factor, Response).
Factor = overweight,
Response = yes ;
Factor = smoker,
Response = no ;
Factor = diabetes,
Response = yes.
What we need to do now is select out "yes" factors and then look up their penalties and add them up. To do this, we can use findall/3, which takes a Template, a Goal, and gives back a result list:
?- findall(risk(Factor, Response), risk(Factor, Response), Responses).
Responses = [risk(overweight, yes), risk(smoker, no), risk(diabetes, yes)].
As you can see, I used the same template and goal here, just to see all the results, but we can put "yes" in to filter it down to just the risk factors we care about:
?- findall(risk(Factor), risk(Factor, yes), Responses).
Responses = [risk(overweight), risk(diabetes)].
Now you can see that the Template (first argument) is just some arbitrary structure populated with the variables that findall/3 found by running Goal, the second argument. So we could also just obtain the list of penalty values, if we look them up inside the Goal query. Like this:
?- findall(Penalty, %% <- template
(risk(Factor, yes), risk_penalty(Factor, Penalty)), %% <- goal
Penalties). %% <- result
Penalties = [2, 3].
We can then follow this with just sumlist/2 to add everything up:
?- findall(Penalty,
(risk(Factor, yes), risk_penalty(Factor, Penalty)),
Penalties),
sumlist(Responsa, Score).
Responsa = [2, 3],
Score = 5.
Now we can finish the inicio/0 predicate:
inicio :-
retractall(risk(_, _)),
ask('Are you overweight?', overweight),
ask('Are you a smoker?', smoker),
ask('Do you have some direct relative with diabetes?', diabetes)
findall(Penalty,
(risk(Factor, yes), risk_penalty(Factor, Penalty)), Penalties),
sumlist(Penalties, Score),
format('The result is ~a~n', [Score]).
This now looks like this when run:
?- inicio.
Are you overweight? [yes/no]> yes.
Are you a smoker? [yes/no]> |: no.
Do you have some direct relative with diabetes? [yes/no]> |: yes.
The result is 5
true.
I hope you find the result pleasing to look at, much less procedural, and easier to modify and maintain.