Update app.py

app.py

@@ -102,7 +102,17 @@ def extract_embeddings(text):
 def extract_embeddings_batch(texts):
     return [extract_embeddings(text) for text in texts]
 
-def compress_semantically(input_text, word_reduction_factor=0.35, num_samples=500):
+def compress_semantically(input_text, word_reduction_factor=0.35):
+
+    num_samples = 500
+    word_count = input_text.split()
+
+    thresholds = [(1500, 80), (1000, 90), (700, 110), (500, 130), (250, 160)]
+    for threshold, value in thresholds:
+        if word_count > threshold:
+            num_samples = value
+            break
+    
     semantic_embeddings = extract_embeddings(input_text)
     text_lang = detect_language_en_pt(input_text)
     stopwords = en_stop_words if text_lang == 'en' else pt_stop_words