From e94c0349270dbf64909b0f62a212a4e481324bbc Mon Sep 17 00:00:00 2001
From: seb <sebgreenbus@gmail.com>
Date: Fri, 18 Apr 2025 09:32:44 +0200
Subject: [PATCH] Refine word similarity calculation in similarity_search.c by
 enforcing exact matches for short words and adjusting similarity thresholds.
 Increase weight of word matches in overall similarity score calculation.

---
 similarity_search.c | 27 ++++++++++++++++++++++-----
 1 file changed, 22 insertions(+), 5 deletions(-)

diff --git a/similarity_search.c b/similarity_search.c
index 87a5329..ac61580 100644
--- a/similarity_search.c
+++ b/similarity_search.c
@@ -59,13 +59,19 @@ void free_words(char *words[], int word_count) {
 float word_similarity(const char *word1, const char *word2) {
     int len1 = strlen(word1);
     int len2 = strlen(word2);
+    
+    // For very short words (3 chars or less), require exact match
+    if (len1 <= 3 || len2 <= 3) {
+        return str_case_cmp(word1, word2) == 0 ? 1.0f : 0.0f;
+    }
+    
     int max_len = len1 > len2 ? len1 : len2;
     if (max_len == 0) return 0.0f;
     
+    // Count matching characters (case insensitive)
     int matches = 0;
     int i = 0, j = 0;
     
-    // Count matching characters (case insensitive)
     while (i < len1 && j < len2) {
         if (tolower((unsigned char)word1[i]) == tolower((unsigned char)word2[j])) {
             matches++;
@@ -81,9 +87,20 @@ float word_similarity(const char *word1, const char *word2) {
     // Calculate similarity based on matching characters
     float char_similarity = (float)matches / max_len;
     
-    // If words are the same length, boost similarity
+    // Require higher similarity for shorter words
+    float min_similarity = 0.7f; // Default minimum
+    if (len1 <= 5 || len2 <= 5) {
+        min_similarity = 0.9f; // Higher requirement for short words
+    }
+    
+    // If similarity is below minimum, return 0
+    if (char_similarity < min_similarity) {
+        return 0.0f;
+    }
+    
+    // If words are the same length, give a small boost
     if (len1 == len2) {
-        char_similarity = char_similarity * 1.2f; // 20% boost for same length
+        char_similarity = char_similarity * 1.1f; // 10% boost for same length
         if (char_similarity > 1.0f) char_similarity = 1.0f;
     }
     
@@ -136,8 +153,8 @@ float calculate_similarity(const char *query, const char *target, float cutoff)
     }
     avg_word_similarity /= query_word_count;
     
-    // Combine scores: 60% weight on word matches, 40% on character similarity
-    float similarity = (word_match_score * 0.6f) + (avg_word_similarity * 0.4f);
+    // Combine scores: 70% weight on word matches, 30% on character similarity
+    float similarity = (word_match_score * 0.7f) + (avg_word_similarity * 0.3f);
     
     // If all query words were found with high similarity, ensure high overall score
     if (query_words_found == query_word_count && avg_word_similarity >= 0.8f) {