Update backend/agents/matcher.py

backend/agents/matcher.py

@@ -1,15 +1,26 @@
+# matcher.py
+
 from sentence_transformers import SentenceTransformer, util
+import re
+from collections import Counter
 
-# Load embedding model once (fast, CPU friendly)
 model = SentenceTransformer("BAAI/bge-small-en-v1.5")
 
 def compute_match(resume_text: str, job_desc: str) -> float:
-    """
-    Compute semantic similarity between resume text and job description.
-    Returns a score between -1 and 1 (cosine similarity).
-    """
     emb_resume = model.encode(resume_text, convert_to_tensor=True, normalize_embeddings=True)
     emb_job = model.encode(job_desc, convert_to_tensor=True, normalize_embeddings=True)
-    
     score = util.cos_sim(emb_resume, emb_job).item()
     return round(float(score), 3)
+
+def extract_keywords(text, top_n=20):
+    stopwords = set([
+        "the","and","a","an","to","of","in","for","on","with",
+        "at","by","from","or","is","are","as","this","that",
+        "your","you","be","has","have","will","can","may","our"
+    ])
+    text = re.sub(r"[^a-zA-Z0-9\s]", "", text)
+    words = text.lower().split()
+    words = [w for w in words if w not in stopwords and len(w) > 2]
+    counter = Counter(words)
+    keywords = [word for word, _ in counter.most_common(top_n)]
+    return keywords