|
1
|
|
|
<?php |
|
2
|
|
|
/* |
|
3
|
|
|
* Copyright (c) 2019 LinkedData.Center. All rights reserved. |
|
4
|
|
|
*/ |
|
5
|
|
|
namespace BOTK\Reasoner; |
|
6
|
|
|
|
|
7
|
|
|
/** |
|
8
|
|
|
* input sensitivity: an acceptable rate for changes (0-1) |
|
9
|
|
|
* Input stream specification: |
|
10
|
|
|
* The input stream is divided in two part, the first part contains, one for row, |
|
11
|
|
|
* a couple of object Uri, string |
|
12
|
|
|
* then there is an empty line |
|
13
|
|
|
* the second part contains, one for row, a couple of an subject uri and a token to be searched in the string |
|
14
|
|
|
* |
|
15
|
|
|
* Output stream: |
|
16
|
|
|
* rules to insert in a graphName, RDF statements that links subject and object having |
|
17
|
|
|
* the minimun Levenshtein distance evaluated on string tokens |
|
18
|
|
|
*/ |
|
19
|
|
|
class Levenshtein extends AbstractReasoner |
|
20
|
|
|
{ |
|
21
|
|
|
protected $sensitivity=0.2; |
|
22
|
|
|
|
|
23
|
|
|
public function setSensitivity( float $sensitivity ) |
|
24
|
|
|
{ |
|
25
|
|
|
assert( $sentistivity >= 0 && $sentistivity <=1 ); |
|
|
|
|
|
|
26
|
|
|
|
|
27
|
|
|
$this->sensitivity = $sensitivity; |
|
28
|
|
|
return $this ; |
|
29
|
|
|
} |
|
30
|
|
|
|
|
31
|
|
|
|
|
32
|
|
|
/** |
|
33
|
|
|
* targets must be an hash of tokenized strings |
|
34
|
|
|
*/ |
|
35
|
|
|
protected function findClosestUriToToken( $token, array $targets) |
|
36
|
|
|
{ |
|
37
|
|
|
$tokenLength=strlen($token); |
|
38
|
|
|
$absSensitivity= round($tokenLength * $this->sensitivity); |
|
39
|
|
|
$minLength=max( 1, $tokenLength - $absSensitivity); |
|
40
|
|
|
$maxLength=$tokenLength + $absSensitivity; |
|
41
|
|
|
$closestTarget = null; |
|
42
|
|
|
$shortest = -1; |
|
43
|
|
|
foreach($targets as $targetUri=>$targetTokens){ |
|
44
|
|
|
if( $shortest === 0 ) break; // a perfect match found. |
|
45
|
|
|
foreach ($targetTokens as $targetToken ){ |
|
46
|
|
|
$targetTokenLength=strlen($targetToken); |
|
47
|
|
|
if( ($targetTokenLength >= $minLength) && ($targetTokenLength <= $maxLength)) { |
|
48
|
|
|
$lev = levenshtein($token, $targetToken); |
|
49
|
|
|
if ($lev <= $shortest || $shortest < 0) { |
|
50
|
|
|
$closestTarget = $targetUri; |
|
51
|
|
|
$shortest = $lev; |
|
52
|
|
|
} |
|
53
|
|
|
if( $shortest === 0 ) break; // a perfect match found. |
|
54
|
|
|
} |
|
55
|
|
|
} |
|
56
|
|
|
} |
|
57
|
|
|
|
|
58
|
|
|
if( $closestTarget && ($shortest <= $absSensitivity) ){ |
|
59
|
|
|
return $closestTarget; |
|
60
|
|
|
} else { |
|
61
|
|
|
return false; |
|
62
|
|
|
} |
|
63
|
|
|
} |
|
64
|
|
|
|
|
65
|
|
|
|
|
66
|
|
|
public function run () |
|
67
|
|
|
{ |
|
68
|
|
|
// analyze first input stream part: read targets |
|
69
|
|
|
// only tokens with length > sensitivity are considered. |
|
70
|
|
|
$targets=[]; |
|
71
|
|
|
|
|
72
|
|
|
while (($data=fgetcsv($this->inputStream)) && isset($data[1])) { |
|
73
|
|
|
assert( !empty($data[0]) && !empty($data[1]) ) ; |
|
74
|
|
|
|
|
75
|
|
|
$subjectUri=$data[0]; |
|
76
|
|
|
$string=$data[1]; |
|
77
|
|
|
|
|
78
|
|
|
// tokenize string |
|
79
|
|
|
$tokenizedString=[]; |
|
80
|
|
|
$tok= strtok($string, ' '); |
|
81
|
|
|
while ($tok !== false) { |
|
82
|
|
|
// ignore too short tokens |
|
83
|
|
|
if( floor(strlen($tok)*$this->sensitivity)>0) { $tokenizedString[]=strtolower($tok);} |
|
84
|
|
|
$tok = strtok(' '); |
|
85
|
|
|
} |
|
86
|
|
|
|
|
87
|
|
|
if( !empty($tokenizedString) ) { $targets[$subjectUri]=$tokenizedString;} |
|
88
|
|
|
} |
|
89
|
|
|
|
|
90
|
|
|
//analyze second input stream part: token to be searched in targets |
|
91
|
|
|
fwrite($this->outputStream, "INSERT DATA { GRAPH <$this->graphName> {\n"); |
|
92
|
|
|
|
|
93
|
|
|
while (($data = fgetcsv($this->inputStream)) !== FALSE) { |
|
94
|
|
|
|
|
95
|
|
|
assert(!empty($data[0]) && !empty($data[1]) ) ; |
|
96
|
|
|
|
|
97
|
|
|
list ($uri, $token)=$data; |
|
98
|
|
|
if( $closestUri=$this->findClosestUriToToken( strtolower($token), $targets) ){ |
|
99
|
|
|
fprintf($this->outputStream, "<%s> <%s> <%s>.\n", $closestUri, $this->property, $uri); |
|
100
|
|
|
} |
|
101
|
|
|
|
|
102
|
|
|
} |
|
103
|
|
|
fwrite($this->outputStream, "}}"); |
|
104
|
|
|
} |
|
105
|
|
|
} |
linkeddatacenter /
BOTK-core
Loading history...