# Understanding Bratko \`translate\` predicate **URL:** \ **Category:** General\ **Created:** [April 20, 2023, 2:02pm UTC](https://swi-prolog.discourse.group/t/understanding-bratko-translate-predicate/6484 "2023-04-20T14:02:12Z")\ **Posts on this page:** 20\ **Page:** 1
**Author:** ![unlocked2412](https://avatars.discourse-cdn.com/v4/letter/u/e47c2d/32.png) [@unlocked2412](https://swi-prolog.discourse.group/u/unlocked2412)\ **Post date:** [April 20, 2023, 2:02pm UTC](https://swi-prolog.discourse.group/t/understanding-bratko-translate-predicate/6484/1 "2023-04-20T14:02:12Z")
As a follow-up to @EricGT and @anon95304481 recommendation, I am studying: > Figure 25.15 `Translating a propositional calculus formula into a set of (asserted) clauses.`, page 638 > Book: _Prolog Programming for Artificial Intelligence_ (4th Edition) by Ivan Bratko **Goal:** Rewrite a propositional formula like: `p => q => r` using the `translate` predicate explained in the book. **Source code:** ```prolog :- op( 100, fy, ~). % Negation :- op( 110, xfy, &). %Conjunction :- op( 120, xfy, v). % Disjunction :- op( 130, xfy, =>). % Implication translate( F & G) :- !, % Translate conjunctive formula translate( F), translate( G). translate( Formula) :- transform( Formula, NewFormula), !, % Transformation step on Formula translate( NewFormula ). translate(Formula):- % No more trasnformation possible assert( clause( Formula )). % Transformation rules for propositional formulas transform(~(~X), X). transform( X => Y, ~X v Y). transform(~ (X & Y), ~X v ~Y). transform(~ (X v Y), ~X & ~Y). transform( X & Y v Z, (X v Z) & (Y v Z)). transform( X v Y & Z, (X v Y) & ( X v Z)). transform( X v Y, X1 v Y) :- % Transform subexpression transform( X, X1), !. transform( X v Y, X v Y1) :- % Transform subexpression transform( Y, Y1), !. transform( ~ X, ~ X1 ) :- % Transform subexpression transform( X, X1). ``` However, I noticed `translate/1` predicate has arity 1, so I can’t use it to unify a variable with the transformed expression. - How can I use that code to rewrite an arbitrary formula like `p => q => r` ? Thank you for any guidance. ---
**Author:** ![EricGT](https://avatars.discourse-cdn.com/v4/letter/e/f1d935/32.png) [@EricGT](https://swi-prolog.discourse.group/u/EricGT)\ **Post date:** [April 20, 2023, 2:16pm UTC](https://swi-prolog.discourse.group/t/understanding-bratko-translate-predicate/6484/2 "2023-04-20T14:16:40Z")
> [@unlocked2412](#): > > How can I use that code to rewrite an arbitrary formula like `p => q => r` ? If you rewrite translate/2 as ```prolog translate(F0 & G0,F & G) :- !, % Translate conjunctive formula translate(F0,F), translate(G0,G). translate(Formula0,Forumula) :- transform(Formula0,Formula1), !, % Transformation step on Formula translate(Formula1,Forumula). translate(Formula,Formula). % Pass final formula back. ``` it should work or give you some idea. I did not run this and thus did not test it. Notice the names are changed to the classic accumulator passing notation. For a very simple example of accumulator passing see: Learn Prolog Now! [5.3 Arithmetic and Lists](https://lpn.swi-prolog.org/lpnpage.php?pagetype=html&pageid=lpn-htmlse20) ---
**Author:** ![unlocked2412](https://avatars.discourse-cdn.com/v4/letter/u/e47c2d/32.png) [@unlocked2412](https://swi-prolog.discourse.group/u/unlocked2412)\ **Post date:** [April 20, 2023, 2:35pm UTC](https://swi-prolog.discourse.group/t/understanding-bratko-translate-predicate/6484/3 "2023-04-20T14:35:58Z")
Thank you so much ! Works great, so far. Could you please explain what these definitions mean ? Why do I need `!` cut operator ? ```prolog transform( X v Y, X1 v Y) :- % Transform subexpression transform( X, X1), !. transform( X v Y, X v Y1) :- % Transform subexpression transform( Y, Y1), !. ``` and give any tips to learn prolog code like this one ? How do you read it ? ---
**Author:** ![EricGT](https://avatars.discourse-cdn.com/v4/letter/e/f1d935/32.png) [@EricGT](https://swi-prolog.discourse.group/u/EricGT)\ **Post date:** [April 20, 2023, 2:50pm UTC](https://swi-prolog.discourse.group/t/understanding-bratko-translate-predicate/6484/4 "2023-04-20T14:50:40Z")
> [@unlocked2412](#): > > Could you please explain what these definitions mean ? Why do I need `!` cut operator ? > > give any tips to learn prolog code like this one ? How do you read it ? I am grateful you are willing to learn Prolog, are willing to accept advise and change course when suggested, and looking at the details, however this site is not for teaching Prolog. Options - Some users here will help more than others so you can ask but don’t always expect a reply. - You also have the Bratko book which is still relevant at your level and worth taking a month or two to work through. Be advised that the book does not use append/3 but instead uses conc/3. Do yourself a favor and use append/3 when the book uses conc/3. Don’t believe me, check the books index at the back. - For simple questions such as yours ChatGPT is not bad, it is also not great, but it is good at summarizing much of the information from its training data into coherent replies, however you do have to know Prolog to know when it is wrong or test out everything it tells you to know when it is wrong or hallucinating, which I too have to do for some things, e.g. [popcount](https://swi-prolog.discourse.group/t/chatgpt-prompts-swi-prolog-2/6260#check-term-is-a-bit-flag-7). - See: [Useful Prolog references](https://swi-prolog.discourse.group/t/useful-prolog-references/1089) Feel free to ask your own questions and post what you learn along the way, I am sure others will read and learn from what you post. ---
**Author:** ![unlocked2412](https://avatars.discourse-cdn.com/v4/letter/u/e47c2d/32.png) [@unlocked2412](https://swi-prolog.discourse.group/u/unlocked2412)\ **Post date:** [April 20, 2023, 7:29pm UTC](https://swi-prolog.discourse.group/t/understanding-bratko-translate-predicate/6484/5 "2023-04-20T19:29:15Z")
Thank you for all the advice. One question, if I may: do you understand why the following two lines are used: ```prolog transform( X v Y, X1 v Y) :- % Transform subexpression transform( X, X1), !. transform( X v Y, X v Y1) :- % Transform subexpression transform( Y, Y1), !. ``` instead of: ```prolog transform( X v Y, X1 v Y1) :- % Transform subexpression transform( X, X1), transform( Y, Y1). ``` From my experience in other languages, this is what I would expect, but perhaps there is a Prolog detail I do not know. I assume the latter would be true, since I am just starting my Prolog journey. In my tests, they behave differently. ---
**Author:** ![EricGT](https://avatars.discourse-cdn.com/v4/letter/e/f1d935/32.png) [@EricGT](https://swi-prolog.discourse.group/u/EricGT)\ **Post date:** [April 20, 2023, 8:29pm UTC](https://swi-prolog.discourse.group/t/understanding-bratko-translate-predicate/6484/6 "2023-04-20T20:29:53Z")
> [@unlocked2412](#): > > do you understand why the following two lines are used: > > instead of: Without checking I would speculate that as two rules either or both can be skipped, e.g. if a rule is not activated then the code will complete. With them combined into one rule if the first pattern did not need to be transformed then it would fail and if the second needed to be transformed it would not and thus an invalid result. In other words as two separate clause (rules) the code works like an `or` but together in one clause (rule) they work as an `and`, the `,` at the end of `transform( X, X1),` is the and operator ([,/2](https://www.swi-prolog.org/pldoc/doc_for?object=(%27,%27)/2)). ---
**Author:** ![unlocked2412](https://avatars.discourse-cdn.com/v4/letter/u/e47c2d/32.png) [@unlocked2412](https://swi-prolog.discourse.group/u/unlocked2412)\ **Post date:** [April 20, 2023, 8:39pm UTC](https://swi-prolog.discourse.group/t/understanding-bratko-translate-predicate/6484/7 "2023-04-20T20:39:19Z")
Thank you so much. > [@EricGT](#): > > With them combined into one rule if the first pattern did not need to be transformed then it would fail Could you clarify what do you mean by “if the first pattern did not need to be transformed” ? ---
**Author:** ![unlocked2412](https://avatars.discourse-cdn.com/v4/letter/u/e47c2d/32.png) [@unlocked2412](https://swi-prolog.discourse.group/u/unlocked2412)\ **Post date:** [April 20, 2023, 8:52pm UTC](https://swi-prolog.discourse.group/t/understanding-bratko-translate-predicate/6484/8 "2023-04-20T20:52:44Z")
> [@unlocked2412](#): > > > [@EricGT](#): > > > > With them combined into one rule if the first pattern did not need to be transformed then it would fail > > Could you clarify what do you mean by “if the first pattern did not need to be transformed” ? If I understand correctly, in this case: ```prolog transform(p v (p => (q => r)), G). ``` The left hand operand `p` cannot be further transformed, since it doesn’t match any of the facts or rules in the database, i.e, doesn’t match the patterns: ```prolog ~(~X) X => Y ~ (X v Y) ... ``` Is that right ? So, if that is the second rule: ```prolog transform( X v Y, X v Y1) :- % Transform subexpression transform( Y, Y1). ``` is required in order to transform the right hand operand. I come from a Haskell background, so this seems some form of pattern matching. ---
**Author:** ![EricGT](https://avatars.discourse-cdn.com/v4/letter/e/f1d935/32.png) [@EricGT](https://swi-prolog.discourse.group/u/EricGT)\ **Post date:** [April 20, 2023, 9:29pm UTC](https://swi-prolog.discourse.group/t/understanding-bratko-translate-predicate/6484/9 "2023-04-20T21:29:23Z")
> [@unlocked2412](#): > > Is that right ? Sounds correct. Again I am not checking this by writing code and running it. > [@unlocked2412](#): > > seems some form of pattern matching It is syntactic unification. You really haven’t read any of the Bratko book have you? Perhaps this page will help you [ChatGPT prompts - Prolog - Basics](https://swi-prolog.discourse.group/t/chatgpt-prompts-prolog-basics/6257) There are several Wiki pages in sequence of ChatGPT starting here. [LLM (Large Language Model) such as ChatGPT prompts related to Prolog](https://swi-prolog.discourse.group/t/llm-large-language-model-such-as-chatgpt-prompts-related-to-prolog/6248) ---
**Author:** ![unlocked2412](https://avatars.discourse-cdn.com/v4/letter/u/e47c2d/32.png) [@unlocked2412](https://swi-prolog.discourse.group/u/unlocked2412)\ **Post date:** [April 21, 2023, 12:40am UTC](https://swi-prolog.discourse.group/t/understanding-bratko-translate-predicate/6484/10 "2023-04-21T00:40:26Z")
> [@EricGT](#): > > It is syntactic unification. You really haven’t read any of the Bratko book have you? Thank you. I have searched the book for “pattern matching” but the results weren’t relevant. I have read some chapters of Clocksin in the past. > [@EricGT](#): > > [ChatGPT prompts - Prolog - Basics](https://swi-prolog.discourse.group/t/chatgpt-prompts-prolog-basics/6257) > > There are several Wiki pages in sequence of ChatGPT starting here. > > [LLM (Large Language Model) such as ChatGPT prompts related to Prolog](https://swi-prolog.discourse.group/t/llm-large-language-model-such-as-chatgpt-prompts-related-to-prolog/6248) Thank you for doing those wiki pages. ---
**Author:** ![EricGT](https://avatars.discourse-cdn.com/v4/letter/e/f1d935/32.png) [@EricGT](https://swi-prolog.discourse.group/u/EricGT)\ **Post date:** [April 21, 2023, 8:35am UTC](https://swi-prolog.discourse.group/t/understanding-bratko-translate-predicate/6484/11 "2023-04-21T08:35:30Z")
> [@unlocked2412](#): > > I come from a Haskell background Have you tried using ChatGPT or such to convert Prolog to Haskell or Haskell to Prolog? I have not tried this using Prolog with Haskell but I do find that using ChatGPT to translate some code from one programming language to anther will at times help in understanding the one you don’t know. ---
**Author:** ![unlocked2412](https://avatars.discourse-cdn.com/v4/letter/u/e47c2d/32.png) [@unlocked2412](https://swi-prolog.discourse.group/u/unlocked2412)\ **Post date:** [April 21, 2023, 11:50am UTC](https://swi-prolog.discourse.group/t/understanding-bratko-translate-predicate/6484/12 "2023-04-21T11:50:53Z")
> [@EricGT](#): > > Have you tried using ChatGPT or such to convert Prolog to Haskell or Haskell to Prolog? I didn’t try that. Thank you for the suggestion. I’ve started using ChatGPT and so far it’s helping. However, I tried to extend the example from Bratko book, and ChatGPT gave me a code that doesn’t work nor I don’t know how to fix. I said to ChatGPT: ```prolog Using the following Prolog code: :- op(1060, yfx, <->). :- op(1050, yfx, <-). :- op(800, yfx, xor). :- op(600, yfx, v). :- op(400, yfx, &). :- op(200, fy, ¬). % Exercise 1 unfold(F0 & G0,F & G) :- !, % Translate conjunctive formula unfold(F0,F), unfold(G0,G). unfold(Formula0,Formula) :- define(Formula0,Formula1), !, % Transformation step on Formula unfold(Formula1,Formula). unfold( X v Y, X1 v Y) :- % Transform subexpression unfold( X, X1), !. unfold( X v Y, X v Y1) :- % Transform subexpression unfold( Y, Y1), !. unfold(Formula,Formula). % Pass final formula back. define((F xor G), ¬ (F <-> G)). define((F <- G), (G -> F)). define((F -> G), ¬ F v G). define((F <-> G), (F -> G) & (F <- G)). implement a `tab/2` predicate which implements a semantic table algorithm, and returns as many open branches as it founds given any propositional formula. If that formula is inconsistent, return false ``` ChatGPT response: > **Expand disclosure triangle to see "prolog" source** > > ```prolog > :- op(1060, yfx, <->). > :- op(1050, yfx, <-). > :- op(800, yfx, xor). > :- op(600, yfx, v). > :- op(400, yfx, &). > :- op(200, fy, ¬). > > % Exercise 1 > > unfold(F0 & G0,F & G) :- !, % Translate conjunctive formula > unfold(F0,F), > unfold(G0,G). > > unfold(Formula0,Formula) :- > define(Formula0,Formula1), !, % Transformation step on Formula > unfold(Formula1,Formula). > > unfold( X v Y, X1 v Y) :- % Transform subexpression > unfold( X, X1), !. > unfold( X v Y, X v Y1) :- % Transform subexpression > unfold( Y, Y1), !. > > unfold(Formula,Formula). % Pass final formula back. > > define((F xor G), ¬ (F <-> G)). > define((F <- G), (G -> F)). > define((F -> G), ¬ F v G). > define((F <-> G), (F -> G) & (F <- G)). > > % Define the semantic table algorithm > tab(Formula, Branches) :- > unfold(Formula, Formula1), % Transform the formula > propagate(Formula1, [], Branches). % Propagate the formula through the table > > % Propagate the formula through the table > propagate([], Branches, Branches). % Base case: no more formulas to propagate > propagate([Formula | Rest], Branches, Result) :- > (inconsistent(Formula) -> % If the formula is inconsistent > Result = false % Return false > ; > split(Formula, Left, Right), % Split the formula into left and right branches > append(Rest, Left, NewFormulas), % Add the left branch to the remaining formulas > append(Rest, Right, NewFormulas1), % Add the right branch to the remaining formulas > propagate(NewFormulas, Branches, Branches1), % Propagate the remaining formulas with the current branches > propagate(NewFormulas1, Branches1, Result) % Propagate the remaining formulas with the updated branches > ). > > % Split a formula into left and right branches > split((F & G), [F, G], []). % Conjunction: add F and G as separate branches > split((F v G), [], [F, G]). % Disjunction: add F and G as separate branches > split((F <-> G), [(F -> G), (G -> F)], []). % Biconditional: add (F -> G) and (G -> F) as separate branches > split((F -> G), [(¬ F), G], []). % Implication: add (¬ F) and G as separate branches > split((F <- G), [F, (¬ G)], []). % Reverse implication: add F and (¬ G) as separate branches > split((F xor G), [(¬ (F <-> G))], []). % Exclusive disjunction: add (¬ (F <-> G)) as a separate branch > split(¬ (F & G), [¬ F], [¬ G]). % Negation of conjunction: add ¬ F and ¬ G as separate branches > split(¬ (F v G), [(¬ F) & (¬ G)], []). % Negation of disjunction: add (¬ F) & (¬ G) as a separate branch > split(¬ (F <-> G), [(F & (¬ G)) v ((¬ F) & G)], []). % Negation of biconditional: add (F & (¬ G)) v ((¬ F) & G) as a separate branch > split(¬ (F -> G), [F, ¬ G], []). % Negation of implication: add F and ¬ G as separate branches > split(¬ (F <- G), [(¬ F) v G], []). % Negation of reverse implication: add (¬ F) v G as a separate branch > split(¬ (F xor G), [(F & G) v ((¬ F) & (¬ G))], []). % Negation of exclusive disjunction: add (F & G) v ((¬ F) & (¬ G)) as a separate branch > split(Formula, [Formula], []). % Base case: single formula is a branch > > % Check if a formula is inconsistent > inconsistent(Formula) :- > member(Formula, [false, ¬ true, (F & ¬ F), (F & (¬ F)) v ((¬ F) & G), (F & (¬ F)) v ((¬ F) & ¬ G), > (F v (¬ F)), (¬ (F <-> F)), (¬ (F -> F)), (¬ (F <- F)), (¬ (F xor F))]). > > ``` **Output** : ```prolog ?- tab(a -> ¬ (b -> c), R). false ``` However, I would expect: ```prolog tab(a -> ¬ (b -> c), R). R=[¬a]; R= [b, ¬ c]; false ``` I wouldn’t know how to fix this, but nonetheless impressive what it can do so far. ---
**Author:** ![EricGT](https://avatars.discourse-cdn.com/v4/letter/e/f1d935/32.png) [@EricGT](https://swi-prolog.discourse.group/u/EricGT)\ **Post date:** [April 21, 2023, 12:09pm UTC](https://swi-prolog.discourse.group/t/understanding-bratko-translate-predicate/6484/13 "2023-04-21T12:09:45Z")
> [@unlocked2412](#): > > ChatGPT gave me a code that doesn’t work nor I don’t know how to fix. One should not expect ChatGPT or any LLM to correctly generate Prolog code. After hundreds of Prolog code prompts of ChatGPT and having seen a lot of the different Prolog code online, as many others note about training data, if it is not in the training data then don’t expect similar for the output and that garbage in will lead to garbage out. So while I do find that ChatGPT is good at giving me a compass direction to head or a general idea of what the code should look like I always test it. It often fails the `make`. However if one has seen and used lots of Prolog code it does make it easier to find the problems and quickly decide if one should abandon the ChatGPT generated code or try to fix the issues. As I don’t understand what you seek by reading the prompt I can’t judge the code. Have you tried converting the Bratko code to Haskell, then having ChatGPT extend it and get it working using your knowledge of Haskell and then convert that back to Prolog? You really need to take the time to work through the first several chapters of the book because when you get stuck you then have to learn how to [debug](https://swi-prolog.discourse.group/t/bug-hunting-toolbox/710) the Prolog and you can come back to your problem. You keep trying to skip the basics and we are not here to teach them to you. ---
**Author:** ![unlocked2412](https://avatars.discourse-cdn.com/v4/letter/u/e47c2d/32.png) [@unlocked2412](https://swi-prolog.discourse.group/u/unlocked2412)\ **Post date:** [April 21, 2023, 12:21pm UTC](https://swi-prolog.discourse.group/t/understanding-bratko-translate-predicate/6484/14 "2023-04-21T12:21:25Z")
Thank you for your valuable guidance ! I just tried to post what I’ve found in my first use of ChatGPT. I’ll take note of your suggestions and apply them. ---
**Author:** ![unlocked2412](https://avatars.discourse-cdn.com/v4/letter/u/e47c2d/32.png) [@unlocked2412](https://swi-prolog.discourse.group/u/unlocked2412)\ **Post date:** [April 21, 2023, 12:23pm UTC](https://swi-prolog.discourse.group/t/understanding-bratko-translate-predicate/6484/15 "2023-04-21T12:23:27Z")
> [@EricGT](#): > > As I don’t understand what you seek by reading the prompt I can’t judge the code. I am trying to find out whether a propositional formula built from the logical connectives defined in the code is inconsistent. If it is not, should return open branches, otherwise returns false. Is the problem, at least, clear enough for an expert Prolog programmer than yourself ? ---
**Author:** ![EricGT](https://avatars.discourse-cdn.com/v4/letter/e/f1d935/32.png) [@EricGT](https://swi-prolog.discourse.group/u/EricGT)\ **Post date:** [April 21, 2023, 12:33pm UTC](https://swi-prolog.discourse.group/t/understanding-bratko-translate-predicate/6484/16 "2023-04-21T12:33:42Z")
> [@unlocked2412](#): > > an expert Prolog programmer than yourself That was funny because I am not a Prolog expert. Jan W. and others consistently shows me my mistakes for which I am grateful. I do try to not ask the same question again and sometimes its months of revisiting the advise before the advise makes sense but learning from ones mistakes is one of the fastest ways to learn so as you I will ask when stuck but that has become less frequent. I really have no idea where you are headed in the long term. I know functional languages like Haskell eat and breath [fold](https://en.wikipedia.org/wiki/Fold_(higher-order_function)) but this is not Haskell, it is Prolog and [relationships](https://en.wikipedia.org/wiki/Finitary_relation) are king. You really need to do a context switch and dump Haskell and since the new context is empty, learn from the book to understand what a relationship means. * * * **EDIT** When it comes to naming predicates if a verb sounds correct then often the thinking is wrong. See: [Declarative reading](https://www.metalevel.at/prolog/reading) ---
**Author:** ![unlocked2412](https://avatars.discourse-cdn.com/v4/letter/u/e47c2d/32.png) [@unlocked2412](https://swi-prolog.discourse.group/u/unlocked2412)\ **Post date:** [April 21, 2023, 12:36pm UTC](https://swi-prolog.discourse.group/t/understanding-bratko-translate-predicate/6484/17 "2023-04-21T12:36:34Z")
> [@EricGT](#): > > learn from the book to understand what a relationship means. Great ! Thank you. At least I know the problem is not clear, so that means I have to study more to: - first clarify the problem, and - have a first outline of the solution. ---
**Author:** ![unlocked2412](https://avatars.discourse-cdn.com/v4/letter/u/e47c2d/32.png) [@unlocked2412](https://swi-prolog.discourse.group/u/unlocked2412)\ **Post date:** [April 21, 2023, 12:42pm UTC](https://swi-prolog.discourse.group/t/understanding-bratko-translate-predicate/6484/18 "2023-04-21T12:42:43Z")
For the sake of completeness, I leave here the complete Bratko example I used to build on top. Before, I made the mistake of using different names. Note `translate` has one extra argument compared to the book: ```prolog translate(F0 & G0,F & G) :- !, % Translate conjunctive formula translate(F0,F), translate(G0,G). translate(Formula0,Forumula) :- transform(Formula0,Formula1), !, % Transformation step on Formula translate(Formula1,Forumula). translate(Formula,Formula). ``` **Prolog source code:** ```prolog :- op( 100, fy, ~). % Negation :- op( 110, xfy, &). % Conjunction :- op( 120, xfy, v). % Disjunction :- op( 130, xfy, =>). % Implication translate(F0 & G0,F & G) :- !, % Translate conjunctive formula translate(F0,F), translate(G0,G). translate(Formula0,Formula) :- transform(Formula0,Formula1), !, % Transformation step on Formula translate(Formula1,Formula). translate(Formula,Formula). % Pass final formula back. % Transformation rules for propositional formulas transform(~(~X), X). transform( X => Y, ~X v Y). transform(~ (X & Y), ~X v ~Y). transform(~ (X v Y), ~X & ~Y). transform( X & Y v Z, (X v Z) & (Y v Z)). transform( X v Y & Z, (X v Y) & ( X v Z)). transform( X v Y, X1 v Y) :- % Transform subexpression transform( X, X1), !. transform( X v Y, X v Y1) :- % Transform subexpression transform( Y, Y1), !. transform( ~ X, ~ X1 ) :- % Transform subexpression transform( X, X1). ``` **The question asked to ChatGPT:** _Using the following Prolog code:_ > **Expand disclosure triangle to see "prolog" source** > > ```prolog > > :- op( 100, fy, ~). % Negation > :- op( 110, xfy, &). %Conjunction > :- op( 120, xfy, v). % Disjunction > :- op( 130, xfy, =>). % Implication > > translate(F0 & G0,F & G) :- !, % Translate conjunctive formula > translate(F0,F), > translate(G0,G). > > translate(Formula0,Formula) :- > transform(Formula0,Formula1), !, % Transformation step on Formula > translate(Formula1,Formula). > > translate(Formula,Formula). % Pass final formula back. > > % Transformation rules for propositional formulas > transform(~(~X), X). > transform( X => Y, ~X v Y). > transform(~ (X & Y), ~X v ~Y). > transform(~ (X v Y), ~X & ~Y). > transform( X & Y v Z, (X v Z) & (Y v Z)). > transform( X v Y & Z, (X v Y) & ( X v Z)). > transform( X v Y, X1 v Y) :- % Transform subexpression > transform( X, X1), !. > transform( X v Y, X v Y1) :- % Transform subexpression > transform( Y, Y1), !. > transform( ~ X, ~ X1 ) :- % Transform subexpression > transform( X, X1). > > ``` _Implement a `tab/2` predicate which implements a semantic table algorithm, and returns as many open branches as it founds given any propositional formula. If that formula is inconsistent, return false_ **The response from ChatGPT** > **Expand disclosure triangle to see "prolog" source** > > ```prolog > :- op( 100, fy, ~). % Negation > :- op( 110, xfy, &). %Conjunction > :- op( 120, xfy, v). % Disjunction > :- op( 130, xfy, =>). % Implication > > % Main predicate to call the semantic table algorithm > tab(Formula, Branches) :- > translate(Formula, TranslatedFormula), > create_initial_table(Table), > expand_table(Table, TranslatedFormula, NewTable), > check_consistency(NewTable, Branches). > > % Predicate to create an initial table with a single row containing the formula > create_initial_table([(1, Formula)]) :- > !, % Cut to prevent backtracking > Formula. > > % Predicate to expand the table by adding new rows based on the formula > expand_table(Table, Formulas, NewTable) :- > findall((I, F), (member((I, F), Table), expand_formula(F, Formulas)), NewFormulas), > subtract(NewFormulas, Table, FormulasToAdd), > append(Table, FormulasToAdd, NewTable). > > % Predicate to expand a formula based on its type > expand_formula(~ F, [F]). > expand_formula(F & G, [F, G]). > expand_formula(F v G, [F, G]). > expand_formula(F => G, [~ F, G]). > > % Predicate to check the consistency of the table and compute open branches > check_consistency([], []). > check_consistency([(I, F) | T], Branches) :- > (satisfiable(F) -> > check_consistency(T, Branches) > ; > Branches = [I | RestBranches], > check_consistency(T, RestBranches) > ). > > % Predicate to check if a formula is satisfiable > satisfiable(Formula) :- > \+ unsatisfiable(Formula). > > % Predicate to check if a formula is unsatisfiable > unsatisfiable(Formula) :- > \+ satisfiable(Formula). > > % Predicate to translate a formula using the given transformation rules > translate(F0 & G0, F & G) :- !, > translate(F0, F), > translate(G0, G). > translate(Formula0, Formula) :- > transform(Formula0, Formula1), !, > translate(Formula1, Formula). > translate(Formula, Formula). > > % Predicate to apply transformation rules on propositional formulas > transform(~ (~X), X). > transform(X => Y, ~X v Y). > transform(~ (X & Y), ~X v ~Y). > transform(~ (X v Y), ~X & ~Y). > transform(X & Y v Z, (X v Z) & (Y v Z)). > transform(X v Y & Z, (X v Y) & (X v Z)). > transform(X v Y, X1 v Y) :- > transform(X, X1), !. > transform(X v Y, X v Y1) :- > transform(Y, Y1), !. > transform(~ X, ~ X1) :- > transform(X, X1). > > ``` ---
**Author:** ![EricGT](https://avatars.discourse-cdn.com/v4/letter/e/f1d935/32.png) [@EricGT](https://swi-prolog.discourse.group/u/EricGT)\ **Post date:** [April 22, 2023, 8:58am UTC](https://swi-prolog.discourse.group/t/understanding-bratko-translate-predicate/6484/20 "2023-04-22T08:58:58Z")
FYI about using Discourse `Hide Details` (the triangle that expands a section) which is found in the Discourse editor toolbar ![image](https://global.discourse-cdn.com/free1/uploads/swiprolog/original/2X/f/fa815ced5ac98520f4c4d2dda408df1a585cf191.png) under the gear icon ![image](https://global.discourse-cdn.com/free1/uploads/swiprolog/original/2X/3/39a624eaba82367499475c867bd4579acd6b5d58.png) At present - Those sections are not searched by Discourse search (the magnifying glass in the upper right). ![image](https://global.discourse-cdn.com/free1/uploads/swiprolog/original/2X/f/fff380eec51a6bbe34e31e9aaf44d73808d71f09.png) - Those sections are not passed along to search engines. Since an LLM completion will often contain invalid code or hallucinations, putting those results in `Hide Details` keeps others from easily finding it, namely those learning Prolog who do not know it is wrong. Hopefully by using `Hide Details` it will also keep the code from being included in training sets for future AI. ---
**Author:** ![unlocked2412](https://avatars.discourse-cdn.com/v4/letter/u/e47c2d/32.png) [@unlocked2412](https://swi-prolog.discourse.group/u/unlocked2412)\ **Post date:** [April 22, 2023, 12:14pm UTC](https://swi-prolog.discourse.group/t/understanding-bratko-translate-predicate/6484/21 "2023-04-22T12:14:20Z")
Thank you for the information. So, I will continue using `Hide Details` if I happen to post some ChatGPT response. > [@EricGT](#): > > When it comes to naming predicates if a verb sounds correct then often the thinking is wrong. > See: [Declarative reading](https://www.metalevel.at/prolog/reading) By the way, thank you for this resource. It has helped me. I have read it from top to bottom > When it comes to naming predicates if a verb sounds correct then often the thinking is wrong. This is interesting. Could you please clarify this ? Could you give an example ? [Next page](https://swi-prolog.discourse.group/t/understanding-bratko-translate-predicate/6484.md?page=2)