introduce resemblance; compare match implementations; retro-search in progress

1 files changed, 30 insertions, 3 deletions

diff --git a/pieces/match-util.js b/pieces/match-util.js
index 1841d15..a13fddb 100644
--- a/pieces/match-util.js
+++ b/pieces/match-util.js
@@ -10,8 +10,11 @@ const createNgram = (word, n) => {          // Ngram creation
     return vector;
 };
 
-/** check similarity between 2 words
- * based on Ngram matches of N = n letters
+/** similarity
+ * rates similarity between 2 words
+ * based on Ngram matches of N = n letters;
+ * implements a 2-dim check (all a-Ngrams vs all all b-Ngrams)
+ * 
  * @param {string} a : first word
  * @param {string} b : second word
  * @param {int} n : Ngram base
@@ -37,6 +40,29 @@ const similarity = (a, b, n) => {         // Ngram match score
     return 0;
 };
 
+/** resemblance
+ * is an alternative similarity rating;
+ * implements an 1-dim Ngram similarity check
+ * and it's much faster than similarity()
+ */
+const resemblance = (a, b, n) => {
+    if (a.length > n && b.length >= a.length) {
+        const aNgram = createNgram(a, n);
+        let hits = 0;
+        for (let i = 0; i < aNgram.length; ++i) {
+            if (b.includes(aNgram[i])) {
+                hits++;
+            }
+        }
+        if (hits > 0) {
+            // rate resemblance based on hits and length-similarity
+            return (hits / aNgram.length) * (a.length / b.length);
+        }
+    }
+    return 0;
+}
+
+
 /** is_exact_match
  * 
  * check if a searching string -> query (string/latin in kb-format)
@@ -62,4 +88,5 @@ module.exports = {
     exact,
     partial,
     similarity,
-};
\ No newline at end of file
+    resemblance
+}