Add simple review sentiment labeler

data-science/analytics/review_sentiment.py

@@ -1,38 +1,85 @@
-"""Review sentiment analysis — placeholder.
+"""Simple review sentiment labeler.
 
-TODO: Implement basic sentiment scoring on mock reviews
-TODO: Use scikit-learn or simple lexicon-based approach
-TODO: Output sentiment labels: positive, neutral, negative
+This script reads mock agent reviews and labels each review as
+positive, neutral, or negative using a basic keyword-based approach.
+
+Formula:
+sentiment_score = positive_word_count - negative_word_count
+
+Rules:
+- score > 0  -> positive
+- score < 0  -> negative
+- score == 0 -> neutral
 """
 
 import json
 from pathlib import Path
 
 DATASETS_DIR = Path(__file__).parent.parent.parent / "datasets"
 
+POSITIVE_WORDS = {
+    "great", "good", "excellent", "helpful", "useful", "fast",
+    "accurate", "easy", "love", "best", "amazing", "reliable"
+}
+
+NEGATIVE_WORDS = {
+    "bad", "poor", "slow", "wrong", "confusing", "buggy",
+    "error", "errors", "difficult", "unhelpful", "hate", "worst"
+}
+
 
 def load_reviews() -> list[dict]:
     with open(DATASETS_DIR / "agent_reviews.json", encoding="utf-8") as f:
         return json.load(f)
 
 
-def analyze_sentiment(reviews: list[dict]) -> list[dict]:
-    """Placeholder sentiment analysis.
+def get_sentiment_score(text: str) -> int:
+    words = text.lower().split()
+
+    positive_count = sum(
+        1 for word in words if word.strip(".,!?") in POSITIVE_WORDS
+    )
+    negative_count = sum(
+        1 for word in words if word.strip(".,!?") in NEGATIVE_WORDS
+    )
+
+    return positive_count - negative_count
+
 
-    TODO: Replace with real sentiment model or lexicon
-    """
+def get_sentiment_label(score: int) -> str:
+    if score > 0:
+        return "positive"
+    if score < 0:
+        return "negative"
+    return "neutral"
+
+
+def analyze_sentiment(reviews: list[dict]) -> list[dict]:
     results = []
+
     for review in reviews:
+        score = get_sentiment_score(review["review"])
+        sentiment = get_sentiment_label(score)
+
         results.append({
             **review,
-            "sentiment": "neutral",  # TODO: compute real sentiment
-            "sentiment_score": 0.0,
+            "sentiment": sentiment,
+            "sentiment_score": score,
         })
+
     return results
 
 
 if __name__ == "__main__":
     reviews = load_reviews()
     analyzed = analyze_sentiment(reviews)
+
+    print("Review Sentiment Results:")
+    print("-" * 40)
+
     for r in analyzed:
-        print(f"[{r['sentiment']}] {r['agent_id']}: {r['review'][:60]}...")
+        print(
+            f"[{r['sentiment']}] "
+            f"score={r['sentiment_score']} "
+            f"{r['agent_id']}: {r['review'][:60]}..."
+        )