diff options
Diffstat (limited to 'core/src/main/scala/com/rockymadden/stringmetric/similarity/JaccardMetric.scala')
-rwxr-xr-x | core/src/main/scala/com/rockymadden/stringmetric/similarity/JaccardMetric.scala | 20 |
1 files changed, 20 insertions, 0 deletions
diff --git a/core/src/main/scala/com/rockymadden/stringmetric/similarity/JaccardMetric.scala b/core/src/main/scala/com/rockymadden/stringmetric/similarity/JaccardMetric.scala new file mode 100755 index 0000000..6ec5db4 --- /dev/null +++ b/core/src/main/scala/com/rockymadden/stringmetric/similarity/JaccardMetric.scala @@ -0,0 +1,20 @@ +package com.rockymadden.stringmetric.similarity + +import com.rockymadden.stringmetric.Metric.StringMetric + +final case class JaccardMetric(n: Int) extends StringMetric[Double] { + import com.rockymadden.stringmetric.Tokenize.NGramTokenizer + + override def compare(a: Array[Char], b: Array[Char]): Option[Double] = + if (n <= 0 || a.length < n || b.length < n) None // Because length is less than n, it is not possible to compare. + else if (a.sameElements(b)) Some(1d) + else NGramTokenizer(n).tokenize(a).flatMap { ca1bg => + NGramTokenizer(n).tokenize(b).map { ca2bg => + val i = (ca1bg.map(_.mkString) intersect ca2bg.map(_.mkString)).length + + i.toDouble / (ca1bg.length + ca2bg.length - i) + } + } + + override def compare(a: String, b: String): Option[Double] = compare(a.toCharArray, b.toCharArray) +} |