Use relative term frequency instead of absolute

src/Engines/MysqlEngine.php

@@ -46,7 +46,7 @@ public function createIndex(string $indexName)
             "CREATE TABLE IF NOT EXISTS {$this->indexName}_doclist (
                     term_id INTEGER,
                     doc_id VARCHAR(255),
-                    hit_count INTEGER);"
+                    hit_count FLOAT);"
         );
 
         $this->index->exec(
@@ -198,13 +198,15 @@ public function saveWordlist(Collection $stems)
 
     public function saveDoclist(array $terms, int $docId)
     {
+        $countTerms = count($terms);
         $insertRows = [];
         foreach ($terms as $term) {
-            $insertRows[] = '(' . $this->index->quote($term['id']) . ', ' . $this->index->quote($docId) . ', ' . $this->index->quote($term['hits']) . ')';
+            $insertRows[] = '('.$this->index->quote($term['id']).', '.$this->index->quote($docId).', '.$this->index->quote($term['hits'] / $countTerms).')';
         }
 
-        $this->index->exec('INSERT INTO ' . $this->indexName . '_doclist (term_id, doc_id, hit_count) VALUES ' . implode(',',
-                $insertRows) . '');
+        $this->index->exec(
+            'INSERT INTO '.$this->indexName.'_doclist (term_id, doc_id, hit_count) VALUES '.implode(',', $insertRows)
+        );
     }
 
     public function saveHitList(array $stems, int $docId, array $termsList)

src/Engines/RedisEngine.php

@@ -213,9 +213,10 @@ public function saveWordlist(Collection $stems)
 
     public function saveDoclist(array $terms, int $docId)
     {
+        $countTerms = count($terms);
         foreach ($terms as $term => $docsHits) {
             $redisKey = "{$this->indexName}:doclist:{$term}:{$docId}";
-            $this->redis->hset($redisKey, 'num_hits', $docsHits['num_hits']);
+            $this->redis->hset($redisKey, 'num_hits', $docsHits['num_hits'] / $countTerms);
         }
     }
 

src/Engines/SqliteEngine.php

@@ -76,7 +76,7 @@ public function createIndex(string $indexName)
         $this->index->exec("CREATE TABLE IF NOT EXISTS doclist (
                     term_id INTEGER,
                     doc_id INTEGER,
-                    hit_count INTEGER)");
+                    hit_count FLOAT)");
 
         $this->index->exec("CREATE TABLE IF NOT EXISTS fields (
                     id INTEGER PRIMARY KEY,
@@ -317,10 +317,11 @@ public function saveDoclist(array $terms, int $docId)
         $insert = 'INSERT INTO doclist (term_id, doc_id, hit_count) VALUES (:id, :doc, :hits)';
         $stmt = $this->index->prepare($insert);
 
+        $countTerms = count($terms);
         foreach ($terms as $term) {
             $stmt->bindValue(':id', $term['id']);
             $stmt->bindValue(':doc', $docId);
-            $stmt->bindValue(':hits', $term['hits']);
+            $stmt->bindValue(':hits', $term['hits'] / $countTerms);
             try {
                 $stmt->execute();
             } catch (\Exception $e) {

src/TNTSearch.php

@@ -114,8 +114,7 @@ public function search(string $phrase, int $numOfResults = 100)
         $keywords = $keywords->map(function ($keyword) {
             return $this->stemmer->stem($keyword);
         });
-        $tfWeight = 1;
-        $dlWeight = 0.5;
+
         $docScores = [];
         $count = $this->totalDocumentsInCollection();
         $noLimit = $this->engine->fuzzy_no_limit;
@@ -127,11 +126,7 @@ public function search(string $phrase, int $numOfResults = 100)
             foreach ($this->getAllDocumentsForKeyword($term, $noLimit, $isLastKeyword) as $document) {
                 $docID = $document['doc_id'];
                 $tf = $document['hit_count'];
-                $num = ($tfWeight + 1) * $tf;
-                $denom = $tfWeight
-                    * ((1 - $dlWeight) + $dlWeight)
-                    + $tf;
-                $score = $idf * ($num / $denom);
+                $score             = $idf * $tf;
                 $docScores[$docID] = isset($docScores[$docID]) ?
                     $docScores[$docID] + $score : $score;
             }