Update app.py

app.py

@@ -3,50 +3,64 @@ from transformers import pipeline
 import numpy as np
 
 # Initialize sentiment analysis pipeline
-try:
-    sentiment_pipeline = pipeline(model="EmmaL1/CustomModel_amazon", return_all_scores=True)
-    qa_pipeline = pipeline("question-answering", model="distilbert/distilbert-base-cased-distilled-squad")
-except Exception as e:
-    st.error(f"Failed to load models: {str(e)}")
-    st.stop()
-
-def get_rating(prediction):
-    """Get rating (1-5 stars) directly from model prediction"""
-    scores = [score['score'] for score in prediction[0]]  # Extract class probabilities
-    return np.argmax(scores) + 1  # Return highest probability class (+1 because ratings start at 1)
+sentiment_pipeline = pipeline(model="EmmaL1/CustomModel_amazon")
+
+# Initialize question-answering pipeline
+qa_pipeline = pipeline("question-answering", model="distilbert/distilbert-base-cased-distilled-squad")
+
+def calculate_rating(sentiment, confidence):
+    """
+    Optimized rating calculation that:
+    1. Provides more granular rating distribution (1-5 stars)
+    2. Uses confidence score to determine rating strength
+    3. Better reflects actual review distributions
+    """
+    if sentiment == "POSITIVE":
+        # Positive reviews get 3-5 stars based on confidence
+        if confidence > 0.9:
+            return 5  # Very confident positive = 5 stars
+        elif confidence > 0.7:
+            return 4  # Strong positive = 4 stars
+        else:
+            return 3  # Mild positive = 3 stars
+    else:
+        # Negative reviews get 1-2 stars based on confidence
+        if confidence > 0.7:
+            return 1  # Very confident negative = 1 star
+        else:
+            return 2  # Less confident negative = 2 stars
 
 def textclassification():
-    st.title("Amazon Review Sentiment Analysis")
-    st.write("Enter a review to analyze sentiment and predict star rating")
-
-    user_input = st.text_area("Enter review text:", height=150)
-    
-    if st.button("Analyze") and user_input:
-        with st.spinner("Analyzing..."):
-            try:
-                # Sentiment analysis and rating prediction
-                sentiment_pred = sentiment_pipeline(user_input)
-                rating = get_rating(sentiment_pred)
-                
-                # Display results
-                st.subheader("Analysis Results")
-                col1, col2 = st.columns(2)
-                with col1:
-                    st.metric("Predicted Rating", f"{rating} / 5")
-                with col2:
-                    st.metric("Sentiment", "Positive" if rating >= 3 else "Negative")
-                
-                # Reason analysis
-                st.subheader("Rating Justification")
-                qa_input = {
-                    'question': f'Why would this review receive {rating} stars?',
-                    'context': user_input  
-                }
-                qa_result = qa_pipeline(qa_input)
-                st.info(qa_result['answer'])
-                
-            except Exception as e:
-                st.error(f"Analysis error: {str(e)}")
+    st.title("Amazon Customer Sentiment Analysis, Ratings, and Reasons")
+    st.write("Enter a sentence to analyze its sentiment and reason:")
+
+    user_input = st.text_input("Input your text:")
+    if user_input:
+
+        # Sentiment Analysis
+        sentiment_result = sentiment_pipeline(user_input)
+        sentiment = sentiment_result[0]["label"]
+        confidence = sentiment_result[0]["score"]
+
+        st.write(f"Sentiment: {sentiment}")
+        st.write(f"Confidence: {confidence:.2f}")
+
+        # Determine the rating using optimized calculation
+        rating = calculate_rating(sentiment, confidence)
+
+        # Display the rating
+        st.write(f"The rating is {rating} stars")
+
+        # Question Answering
+        qa_input = {
+            'question': f'Why is the rating {rating} star?',
+            'context': user_input
+        }
+        qa_result = qa_pipeline(qa_input)
+        st.write(f"Reasons: {qa_result['answer']}")
+
+def main():
+    textclassification()
 
 if __name__ == "__main__":
-      textclassification()
+    main()