six degrees to kevin bacon [and other queries] using dbpedia

62kevin.pl

#!/usr/bin/perl
use strict;
use warnings;
use Compress::Raw::Bzip2;
use Graph::Undirected;
use Storable;
use Term::ReadLine; # Term::ReadLine::Gnu
$|=1;

my $root = 'Kevin_Bacon'; # dbpedia resource name of an actor
my $input = 'infobox_en.nt.bz2'; # http://downloads.dbpedia.org/3.3/en/infobox_en.nt.bz2
my $cache1 = '62kevin.g';
my $cache2 = '62kevin.sptg';
my $history = "$ENV{HOME}/.62kevin";

$root = "a:$root";
my $sptg;
if(-s $cache2) {
	printf "loading cache2 (%d bytes) ...\n", -s $cache2;
	$sptg = retrieve($cache2);
} else {
	my $g;
	if(-s $cache1) {
		printf "loading cache1 (%d bytes) ...\n", -s $cache1;
		$g = retrieve($cache1);
	} else {
		$g = Graph::Undirected->new(countvertexed=>1, countedged=>1);

		my $size = -s $input;
		printf "parse input (%d bytes compressed) ...\n", $size;
		my $bz = new Compress::Raw::Bunzip2
			or die "Cannot create bunzip2 object\n";
		open FILE, "<$input" or die "$input: $!";
		my $last = 0;
		my $buffer = '';
		my $compressed;
		my $num_parsed = 0;
		my $num_added = 0;
		while (read(FILE,$compressed,4096)) {
			my $decompressed;
			my $status = $bz->bzinflate($compressed, $decompressed);
			die "error decompressing: $status" 
				unless $status == BZ_OK or $status == BZ_STREAM_END;
			$buffer .= $decompressed;
			$buffer =~ s{[^\n]*\n}{
				$num_parsed ++;
				my $res = 'http://dbpedia\\.org/resource';
				my $prop = 'http://dbpedia\\.org/property';
				if($& =~ m{<$res/(.*?)> <$prop/starring> <$res/(.*?)>}) {
					$num_added ++;
					my $movie = "m:$1";
					my $actor = "a:$2";
					$g->add_vertex($movie);
					$g->add_vertex($actor);
					$g->add_edge($movie,$actor);
				}
				if(time-$last > 2) {
					$last = time;
					printf "\r%6.2f%% completed (buffer %d) %d triples inserted of %d parsed.", 
						(tell(FILE) / $size)*100, length $buffer, $num_added, $num_parsed;
				}
				""
			}msge;
		}
		printf "\r%6.2f%% completed (buffer %d) %d triples inserted of %d parsed.\n", 
			(tell(FILE) / $size)*100, length $buffer, $num_added, $num_parsed;
		close FILE;
		printf "saving cache level1 (%d vertices, %d edges) ...\n", 
			scalar($g->vertices), scalar($g->edges);
		store $g, $cache1;
	}

	printf "Dijkstra single-source shortest path (%d vertices, %d edges) ...\n", 
		scalar($g->vertices), scalar($g->edges);
	$sptg = $g->SPT_Dijkstra($root);

	printf "saving cache level2 ...\n";
	store $sptg, $cache2;
}

print "ready.\n";
print "type one dbpedia actor resource name per line:\n";
my $term = new Term::ReadLine '62kevin';
eval { $term->ReadHistory($history); };
while ( defined ($_ = $term->readline('actor>')) ) {
	chomp;
	next unless $_;
	s{\b\w}{uc $&}eg;
	s{\s}{_}g;
	#$term->addhistory($_); # XXX automatic on gnu version
	my $node = "a:$_";
	my $r = $root;
	$r =~ s{a:}{};
	my $w = $sptg->get_vertex_attribute($node, 'weight');
	unless(defined $w) {
		print $node eq $root ? 
				"$r is the root node, distance 0" : 
				"$_ not found, may not be connected to $r",
			"\n";
		next;
	}
	my $d = int($w/2);
	printf "%s is %s degree%s to %s through:\n", $_, $d, $d == 1 ? '' : 's', $r;
	my $p = $node;
	while($p = $sptg->get_vertex_attribute($p, 'p')) {
		local $_ = $p;
		s{m:}{movie: } or s{a:}{actor: };
		print "  $_\n";
	}
}
print "\nshutting down ...\n";
END {  
	if($term) {
		eval {
			$term->WriteHistory($history) 
				or die "$history: $!";
			print "history file $history wrote.\n";
		};
	}
}

62kevin.rq

PREFIX dbpedia: <http://dbpedia.org/resource/>
PREFIX dbpprop: <http://dbpedia.org/property/>
SELECT ?actor6 ?step
WHERE {
  { SELECT *
    WHERE { [] dbpprop:starring ?actor1, ?actor6 . }
  }
  OPTION ( transitive,
           t_distinct,
           t_no_cycles,
           t_in (?actor1),
           t_out (?actor6),
           t_step ('step_no') as ?step,
           t_min(6),
           t_max(6) )  .
  FILTER (?actor1 = dbpedia:Kevin_Bacon)
}

count62kevin.rq

PREFIX dbpedia: <http://dbpedia.org/resource/>
PREFIX dbpprop: <http://dbpedia.org/property/>
SELECT ?step ?count
WHERE
{
  {
    SELECT ?step COUNT(*) as ?count
    WHERE 
    {
      {
        SELECT ?actor1 ?actorN
        WHERE { [] dbpprop:starring ?actor1, ?actorN . }
      }
      OPTION ( transitive,
               t_distinct,
               t_no_cycles,
               t_in (?actor1),
               t_out (?actorN),
               t_step ('step_no') as ?step,
               t_min(1) ) .
      FILTER (?actor1 = dbpedia:Kevin_Bacon)
    }
    GROUP BY ?step
  }
}
ORDER BY ?step

group-concat-distinct.rq

select
(min(?n) AS ?name),
(sql:GROUP_CONCAT_DISTINCT(?p, " | ") AS ?page),
(sql:GROUP_CONCAT_DISTINCT(IF(?ln, ?ln, ?l), ", ") AS ?code)
where {
  ?s dbo:genre dbr:Game_engine ;
     rdfs:label ?n .
  FILTER (LANG(?n) = "en")
  OPTIONAL {
     ?s foaf:homepage ?p .
  }
  OPTIONAL {
     ?s dbp:programmingLanguage ?l .
     OPTIONAL {
       ?l rdfs:label ?ln .
       FILTER (LANG(?ln) = "en")
     }
  }
}
group by ?s

in-between-locations.rq

select 
min( (bif:power(bif:st_distance (?point, bif:st_point(-46.6333, -23.55)), 2) + bif:power(bif:st_distance (?point, bif:st_point(-43.1964, -22.9083)), 2) ) ) as ?rank 
min(?label) as ?label
max(?pop) as ?pop 
max(?height) as ?height
?city

where {

  ?city dbpedia-owl:country dbpedia:Brazil ;
           geo:geometry ?point ;
           dbpprop:elevationM ?height ;
           dbpedia-owl:populationTotal ?pop ;
           rdfs:label ?label .

  FILTER( langMatches( lang(?label), "pt") )

}
group by ?city
order by ?rank

inverseOf.rq

define input:inference "http://example.com/rules"
PREFIX : <http://example.com#>
SELECT ?s ?n {
  FILTER( ?o = :A )
  { SELECT ?s ?o
    WHERE {
      { ?s :is_a_subset_of ?o . }
      UNION
      { ?o :is_a_superset_of ?s . }
    }
  } OPTION ( transitive,
             t_in(?s),
             t_out(?o),
             t_distinct,
             t_no_cycles,
             t_step('step_no') as ?n,
             t_min(1) ) .
}

simpsons.rq

PREFIX skos: <http://www.w3.org/2004/02/skos/core#>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX dbpedia: <http://dbpedia.org/resource/>
PREFIX dbpprop: <http://dbpedia.org/property/>
SELECT ?chalkboard_gag, ?season_no, ?episode_no, ?episode_title
WHERE {
  ?episode rdfs:label ?episode_title ;
           dbpprop:episodeNo ?episode_no ;
           dbpprop:season ?season_no ;
           dbpprop:blackboard ?chalkboard_gag .
  
  FILTER (langMatches( lang(?episode_title), "EN" ))
  FILTER (isLITERAL(?chalkboard_gag) && !regex(?chalkboard_gag, "^None|^-$", "i"))
}
ORDER BY ?season_no ?episode_no

transitive-path.rq

sparql
prefix n: <http://lua.net/n/>
prefix p: <http://lua.net/p/>

SELECT ?s ?o ?n ?via ?path
FROM <http://lua.net>
{
  FILTER( ?o = n:990 )
  FILTER( ?s = n:2210270 )
  { SELECT ?s ?o
    WHERE {
      { ?s p:edge ?o . }
      UNION
      { ?o p:edge ?s . }
    }
  } OPTION ( transitive,
             t_in(?s),
             t_out(?o),
             t_distinct,
             t_no_cycles,
             t_step(?s) as ?via,
             t_step('step_no') as ?n,
             t_step('path_id') as ?path,
             t_min(1),
             t_max(4)
             ) .
}
;

vector-group-by.rq

sparql # clear test environment
clear graph <http://test.com/>;


sparql # load test data
prefix : <http://test.com/>
insert data into graph <http://test.com/> {
 :A a owl:Class .
 :B a owl:Class .
 :C a owl:Class .
 :D a owl:Class .
 :E a owl:Class .
 :A rdfs:subClassOf owl:Thing .
 :B rdfs:subClassOf owl:Thing .
 :C rdfs:subClassOf :A .
 :D rdfs:subClassOf :B .
 :E rdfs:subClassOf :D .
 :A :color "red" .
 :B :color "blue" .
 :C :color "green" .
 :E :color "black" .
 :a rdf:type :A .
 :b rdf:type :B .
 :c rdf:type :C .
 :d rdf:type :D .
 :e rdf:type :E .
};


sparql # show the closest colors of one individual
define input:default-graph-uri <http://test.com/>
prefix : <http://test.com/>
SELECT ?color
WHERE
{
  :e a ?c0 .
  ?c0 rdfs:subClassOf ?c1 option (transitive, t_distinct, t_step('step_no') as ?n, t_in(?c0), t_out(?c1), t_min(0)) .
  ?c1 :color ?color .
}
order by ?n
limit 1;


sparql # show all the closest colors of each class
define input:default-graph-uri <http://test.com/>
prefix : <http://test.com/>
SELECT ?c0 (bif:aref(min(bif:vector(?n, ?color)),1)) as ?v  # interesting vector aggregation
WHERE
{
  ?c0 a owl:Class .
  ?c0 rdfs:subClassOf ?c1 option (transitive, t_distinct, t_step('step_no') as ?n, t_in(?c0), t_out(?c1), t_min(0)) .
  ?c1 :color ?color .
}
group by ?c0;