intuit#35 fix DateMatch with NeighborhoodRange greater than 0.91 failing

README.md

@@ -114,7 +114,7 @@ Some commonly used are already available.
 * __Match Type__ : Allows 2 types of matches, which can be applied to each `Element`
     * _Equality_: Uses exact matches with token values. 
     * _Nearest Neighbor_: Finds tokens that are contained in the neighborhood range, that can be specified as a 
-    probability (0.0 - 1.0) for each element. It defaults to 0.9 
+    probability (0.0 - 1.0) for each element. It defaults to 0.99 for Dates and 0.9 otherwise. 
 
 * __Scoring__ : These are defined for `Element` and `Document` matches
     * _Element scoring_: Uses a simple average, where for each element the matching token is divided by the total tokens. 
@@ -160,7 +160,7 @@ _Note: Since each element is unique in the way it should match, if you need to m
 * __TokenizerFunction__: Override The _TokenizerFunction_ function defined by Type
 * __MatchType__: Override the MatchType defined by Type
 * __NeighborhoodRange__: Relevant only for `NEAREST_NEIGHBORS` MatchType. Defines how close should the `Value` be, to be considered a match. 
-Accepted values between 0.0 - 1.0 (defaults to 0.9) 
+Accepted values between 0.0 - 1.0 (defaults to 0.99 for Dates and 0.9 otherwise) 
 
 ### Match Service
 It supports 3 ways to match the documents

src/main/java/com/intuit/fuzzymatcher/component/TokenRepo.java

@@ -86,7 +86,7 @@ private class TokenRange {
 
         private final Object lower;
         private final Object higher;
-        private static final double DATE_SCALE_FACTOR = 1.1;
+        private static final double DATE_SCALE_FACTOR = 0.9;
 
 
         TokenRange(Token token, double pct) {
@@ -104,8 +104,8 @@ private class TokenRange {
                 this.lower = getLower((Float) value, pct).floatValue();
                 this.higher = getHigher((Float) value, pct).floatValue();
             } else if (value instanceof Date) {
-                this.lower = new Date(getLower(((Date) value).getTime(), pct * DATE_SCALE_FACTOR).longValue());
-                this.higher = new Date(getHigher(((Date) value).getTime(), pct * DATE_SCALE_FACTOR).longValue());
+                this.lower = new Date(getLower(((Date) value).getTime(), pct).longValue());
+                this.higher = new Date(getHigher(((Date) value).getTime(), pct).longValue());
             } else {
                 throw new MatchException("Data Type not supported");
             }

src/main/java/com/intuit/fuzzymatcher/domain/Element.java

@@ -21,7 +21,7 @@
  * <li>type - The ElementType for the value. This determines the functions applied at different steps of the match</li>
  * <li>weight - Used in scoring function to increase the Document score for an Element. Default is 1.0 for all elements</li>
  * <li>threshold - Value above which elements are considered a match, default 0.3</li>
- * <li>neighborhoodRange - Relevant for NEAREST_NEIGHBORS MatchType. Defines how close should the value be, to be considered a match (default 0.9) </li>
+ * <li>neighborhoodRange - Relevant for NEAREST_NEIGHBORS MatchType. Defines how close should the value be, to be considered a match (default 0.99 for dates 0.9 otherwise) </li>
  * <li>preProcessFunction - Function to pre-process the value. If this is not set, the function defined in ElementType is used </li>
  * <li>tokenizerFunction - Function to break values into tokens. If this is not set, the function defined in ElementType is used </li>
  * <li>matchType - MatchType used. If this is not set, the type defined in ElementType is used </li>
@@ -168,7 +168,7 @@ public static class Builder<T> {
         private T value;
         private double weight = 1.0;
         private double threshold = 0.3;
-        private double neighborhoodRange = 0.9;
+        private Double neighborhoodRange;
         private Function<T, T> preProcessFunction;
         private MatchType matchType;
 
@@ -222,6 +222,13 @@ public Builder setMatchType(MatchType matchType) {
 
 
         public Element createElement() {
+            if (neighborhoodRange == null){
+                if (type == ElementType.DATE) {
+                    neighborhoodRange = 0.99;
+                } else {
+                    neighborhoodRange = 0.9;
+                }
+            }
             return new Element<T>(type, variance, value, weight, threshold, neighborhoodRange, preProcessFunction, tokenizerFunction, matchType);
         }
     }

src/test/java/com/intuit/fuzzymatcher/component/MatchServiceTest.java

@@ -567,6 +567,14 @@ public void itShouldApplyMatchWithDate() {
         Assert.assertEquals(2, result.size());
     }
 
+    @Test
+    public void itShouldApplyMatchWithDateForHighNeighborhoodRange() {
+        List<Object> dates = Arrays.asList(getDate("01/01/2020"), getDate("01/02/2020"), getDate("02/01/2019"));
+        List<Document> documentList = getTestDocuments(dates, DATE, 0.999); // 0.999 neighborhood is about 18 days
+        Map<Document, List<Match<Document>>> result = matchService.applyMatch(documentList);
+        Assert.assertEquals(2, result.size());
+    }
+
 
     private List<Document> getTestDocuments(List<Object> values, ElementType elementType, Double neighborhoodRange) {
         AtomicInteger ai = new AtomicInteger(0);