Update app.py

app.py

@@ -1,37 +1,21 @@
-import gradio as gr
-from joblib import load
+from transformers import pipeline
 
-# Define function to load SVM model
-def load_svm_model(model_url):
-    return load(model_url)
+def load_svm_model(model_name):
+    # Load the trained SVM model from Hugging Face
+    model = pipeline(task="text-classification", model=model_name)
+    return model
 
-# Load the SVM model from Hugging Face
+# Load SVM model from Hugging Face
 svm_model = load_svm_model("ankitdotpy/SVM_model_by_Group12")
 
-# Define function for Gradio interface using the loaded model
+# Define prediction function
 def predict_sentiment(text):
-    # Load the SVM model
-    clf_svm = svm_model
+    # Predict sentiment using the imported model
+    result = svm_model(text)
+    sentiment_label = result[0]['label']
+    confidence = result[0]['score'] * 100
+    return f"{sentiment_label} ({confidence:.2f}%)"
 
-    # Vectorize the input text (assuming you already have the vectorizer)
-    text_vector = vectorizer.transform([text])
-
-    # Predict sentiment
-    sentiment = clf_svm.predict(text_vector)[0]
-
-    # Get probabilities for each class
-    probabilities = clf_svm.predict_proba(text_vector)[0]
-
-    # Convert probabilities to percentages
-    percentages = [round(prob * 100, 2) for prob in probabilities]
-
-    # Choose the sentiment label based on the predicted class
-    if sentiment == "POSITIVE":
-        return f"Positive ({percentages[1]}%)"
-    elif sentiment == "NEUTRAL":
-        return f"Neutral ({percentages[2]}%)"
-    else:
-        return f"Negative ({percentages[0]}%)"
 
 # Create Gradio interface
 iface = gr.Interface(