Ask a query about Obama

Allikas: Kursused
Redaktsioon seisuga 3. november 2017, kell 09:46 kasutajalt Tanel (arutelu | kaastöö)
Mine navigeerimisribale Mine otsikasti

This example is a small modification of the obama1.txt problem, with actual questions added and a bit less output.

Initial setting

We repeat the obama1.txt scenario in this chapter.

Suppose we have a triplet

http://en.wikipedia.org/wiki/Barack_Obama,
id:type,
http://conceptnet5.media.mit.edu/web/c/en/person

then we encode it in the otter format like this:

  
  rdf("http://en.wikipedia.org/wiki/Barack_Obama",
      "id:type",
      "http://conceptnet5.media.mit.edu/web/c/en/person").

and then we add three handmade common-sense taxonomy rules (observe that A->B is equivalent to -A | B)

	  
  -rdf(X,"id:type",
      "http://conceptnet5.media.mit.edu/web/c/en/person") |
   rdf(X,"id:type",
      "http://conceptnet5.media.mit.edu/web/c/en/animal").
	  
   -rdf(X,"id:type",
      "http://conceptnet5.media.mit.edu/web/c/en/animal") |
   rdf(X,"id:type",
      "mortal").	  
  	  
   -rdf(X,"id:type",
      "http://conceptnet5.media.mit.edu/web/c/en/person") |
    rdf(X,"id:CapableOf",
      "id:eat").	  

Asking a Yes/No question

Now we add the question: is Obama mortal? We add the question in a negated form, like this:

-rdf("http://en.wikipedia.org/wiki/Barack_Obama",
     "id:type",
     "mortal"). 

The whole problem obama2.txt with the strategy selection and everything, ready to run:

% clear automatic strategy selection
clear(auto).

% use capital letters (A,X,T,...) as vars
set(prolog_style_variables).

% select the search strategy
set(hyper_res). % an alternative is to use set(binary_res).
set(factor).

% select sensible amount of output
clear(print_given). % remove to see search process milestones
clear(print_kept).  % do not print out generated and kept clauses.
assign(stats_level, 0).

% just make it stop after N secs
assign(max_seconds, 10).

list(sos).

% example data

 rdf("http://en.wikipedia.org/wiki/Barack_Obama",
      "id:type",
      "http://conceptnet5.media.mit.edu/web/c/en/person").
      
% some simple taxonomy rules of the form A->B i.e. -A | B in clause form.

-rdf(X,"id:type",
    "http://conceptnet5.media.mit.edu/web/c/en/person") |
 rdf(X,"id:type",
    "http://conceptnet5.media.mit.edu/web/c/en/animal").
  
 -rdf(X,"id:type",
    "http://conceptnet5.media.mit.edu/web/c/en/animal") |
 rdf(X,"id:type",
    "mortal").	  
    
 -rdf(X,"id:type",
    "http://conceptnet5.media.mit.edu/web/c/en/person") |
  rdf(X,"id:CapableOf",
    "id:eat").

% our question (negated)

-rdf("http://en.wikipedia.org/wiki/Barack_Obama",
     "id:type",
     "mortal"). 

end_of_list.

Just send it to otter via standard inpt:

otter < obama2.txt

and look at the --- PROOF --- in output.

It is a good idea to send the output to a file and then process it later, like this:

otter < obama2.txt > obama2.out


Asking a question with a factual output

Now we add the question to find an actual value: something mortal. We add the question in a negated form with a special $ans predicate added to capture the value found, like this:

-rdf(X,
     "id:type",
     "mortal") | 
$ans(X).

The whole problem obama3.txt with the strategy selection and everything, ready to run:

% clear automatic strategy selection
clear(auto).

% use capital letters (A,X,T,...) as vars
set(prolog_style_variables).

% select the search strategy
set(hyper_res). % an alternative is to use set(binary_res).
set(factor).

% select sensible amount of output
clear(print_given). % remove to see search process milestones
clear(print_kept).  % do not print out generated and kept clauses.
assign(stats_level, 0).

% just make it stop after N secs
assign(max_seconds, 10).

list(sos).

% example data

 rdf("http://en.wikipedia.org/wiki/Barack_Obama",
      "id:type",
      "http://conceptnet5.media.mit.edu/web/c/en/person").
      
% some simple taxonomy rules of the form A->B i.e. -A | B in clause form.

-rdf(X,"id:type",
    "http://conceptnet5.media.mit.edu/web/c/en/person") |
 rdf(X,"id:type",
    "http://conceptnet5.media.mit.edu/web/c/en/animal").
  
 -rdf(X,"id:type",
    "http://conceptnet5.media.mit.edu/web/c/en/animal") |
 rdf(X,"id:type",
    "mortal").	  
    
 -rdf(X,"id:type",
    "http://conceptnet5.media.mit.edu/web/c/en/person") |
  rdf(X,"id:CapableOf",
    "id:eat").

% our question (negated)

-rdf(X,
     "id:type",
     "mortal") | 
$ans(X).

end_of_list.

Just send it to otter via standard inpt:

otter < obama3.txt

and look at the --- PROOF --- in output.

It is a good idea to send the output to a file and then process it later, like this:

otter < obama3.txt > obama3.out