summaryrefslogtreecommitdiff
path: root/core/source/main/scala/com/rockymadden/stringmetric/similarity/JaccardMetric.scala
diff options
context:
space:
mode:
Diffstat (limited to 'core/source/main/scala/com/rockymadden/stringmetric/similarity/JaccardMetric.scala')
-rwxr-xr-xcore/source/main/scala/com/rockymadden/stringmetric/similarity/JaccardMetric.scala34
1 files changed, 10 insertions, 24 deletions
diff --git a/core/source/main/scala/com/rockymadden/stringmetric/similarity/JaccardMetric.scala b/core/source/main/scala/com/rockymadden/stringmetric/similarity/JaccardMetric.scala
index e32c926..e1fd4ed 100755
--- a/core/source/main/scala/com/rockymadden/stringmetric/similarity/JaccardMetric.scala
+++ b/core/source/main/scala/com/rockymadden/stringmetric/similarity/JaccardMetric.scala
@@ -1,20 +1,17 @@
package com.rockymadden.stringmetric.similarity
-import com.rockymadden.stringmetric.{StringMetric, StringFilter}
-import com.rockymadden.stringmetric.tokenization.NGramTokenizer
+import com.rockymadden.stringmetric.Metric.StringMetricLike
-/* An implementation of the Jaccard metric. */
-class JaccardMetric extends StringMetric[Int, Double] { this: StringFilter =>
- final override def compare(charArray1: Array[Char], charArray2: Array[Char])(implicit n: Int): Option[Double] = {
- if (n <= 0) throw new IllegalArgumentException("Expected valid n.")
+final case class JaccardMetric(private val n: Int) extends StringMetricLike[Double] {
+ import com.rockymadden.stringmetric.tokenization.NGramTokenizer
- val fca1 = filter(charArray1)
- lazy val fca2 = filter(charArray2)
+ override def compare(a: Array[Char], b: Array[Char]): Option[Double] = {
+ if (n <= 0) return None
- if (fca1.length < n || fca2.length < n) None // Because length is less than n, it is not possible to compare.
- else if (fca1.sameElements(fca2)) Some(1d)
- else NGramTokenizer.tokenize(fca1)(n).flatMap { ca1bg =>
- NGramTokenizer.tokenize(fca2)(n).map { ca2bg =>
+ if (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)
@@ -22,16 +19,5 @@ class JaccardMetric extends StringMetric[Int, Double] { this: StringFilter =>
}
}
- final override def compare(string1: String, string2: String)(implicit n: Int): Option[Double] =
- compare(string1.toCharArray, string2.toCharArray)(n: Int)
-}
-
-object JaccardMetric {
- private lazy val self = apply()
-
- def apply(): JaccardMetric = new JaccardMetric with StringFilter
-
- def compare(charArray1: Array[Char], charArray2: Array[Char])(n: Int) = self.compare(charArray1, charArray2)(n)
-
- def compare(string1: String, string2: String)(n: Int) = self.compare(string1, string2)(n)
+ override def compare(a: String, b: String): Option[Double] = compare(a.toCharArray, b.toCharArray)
}