Update app.py

app.py

@@ -1,84 +1,114 @@
+# ============================================================
+# 🌍 Multi-Lingual Sentiment Analysis (English + Persian)
+# With SHAP Interpretability
+# ============================================================
+
 import gradio as gr
 import joblib
-import shap
 import numpy as np
+import shap
 import matplotlib.pyplot as plt
-import tempfile
 import os
 
-# ---------------------------------------------------------
-# Load both models and vectorizers
-# ---------------------------------------------------------
-english_model = joblib.load("best_model.pkl")
-english_vec   = joblib.load("tfidf_vectorizer.pkl")
+# ------------------------------------------------------------
+# 1️⃣ Load Pretrained Models and Vectorizers
+# ------------------------------------------------------------
+english_model = joblib.load("english_model.pkl")
+english_vectorizer = joblib.load("english_vectorizer.pkl")
+
+persian_model = joblib.load("persian_model.pkl")
+persian_vectorizer = joblib.load("persian_vectorizer.pkl")
+
+# Define class labels
+english_labels = ["Negative", "Neutral", "Positive"]
+persian_labels = ["منفی", "خنثی", "مثبت"]
+
+# ------------------------------------------------------------
+# 2️⃣ SHAP Visualization Function
+# ------------------------------------------------------------
+def get_shap_plot(model, vectorizer, text, class_index, class_name):
+    X_input = vectorizer.transform([text])
+    explainer = shap.Explainer(model, vectorizer.transform([" ".join(text.split()[:50])]))
+    shap_values = explainer(X_input)
+    shap_for_class = shap_values.values[0][:, class_index]
+    feature_names = np.array(vectorizer.get_feature_names_out())
+
+    top_idx = np.argsort(-np.abs(shap_for_class))[:10]
+    top_words = feature_names[top_idx]
+    top_impacts = shap_for_class[top_idx]
 
-persian_model = joblib.load("logistic_regression.pkl")
-persian_vec   = joblib.load("tfidf_vectorizer_persian.pkl")
+    plt.figure(figsize=(6, 3))
+    colors = ["crimson" if v > 0 else "steelblue" for v in top_impacts]
+    plt.barh(top_words, top_impacts, color=colors)
+    plt.title(f"Top Words driving {class_name} prediction")
+    plt.xlabel("SHAP Value (Impact)")
+    plt.gca().invert_yaxis()
+    plt.tight_layout()
+    plt.savefig("shap_plot.png", bbox_inches='tight')
+    plt.close()
 
-class_names = ["Negative", "Neutral", "Positive"]
+    return top_words.tolist(), "shap_plot.png"
 
-# ---------------------------------------------------------
-# Prediction + Interpretability Function
-# ---------------------------------------------------------
+# ------------------------------------------------------------
+# 3️⃣ Prediction + Interpretability Function
+# ------------------------------------------------------------
 def predict_sentiment(text, language):
     if not text.strip():
-        return "Please enter text!", None
+        return "Please enter a comment.", None
 
     if language == "English":
-        model, vec = english_model, english_vec
+        model, vectorizer, labels = english_model, english_vectorizer, english_labels
     else:
-        model, vec = persian_model, persian_vec
+        model, vectorizer, labels = persian_model, persian_vectorizer, persian_labels
 
-    X = vec.transform([text])
-    probs = model.predict_proba(X)[0]
+    X_input = vectorizer.transform([text])
+    probs = model.predict_proba(X_input)[0]
     pred_idx = np.argmax(probs)
-    label = class_names[pred_idx]
-
-    # --- SHAP interpretability ---
-    explainer = shap.LinearExplainer(model, vec.transform([text]))
-    shap_vals = explainer(X)
-    shap_values = shap_vals.values[0][:, pred_idx]
-    feature_names = vec.get_feature_names_out()
-
-    top_idx = np.argsort(-abs(shap_values))[:10]
-    tokens = [feature_names[i] for i in top_idx]
-    impacts = [shap_values[i] for i in top_idx]
-
-    # Save temporary bar chart
-    fig, ax = plt.subplots(figsize=(6, 3))
-    colors = ["crimson" if v > 0 else "steelblue" for v in impacts]
-    ax.barh(tokens, impacts, color=colors)
-    ax.invert_yaxis()
-    ax.set_title(f"Top Words driving {label} prediction")
-    tmp_path = tempfile.mktemp(suffix=".png")
-    plt.tight_layout()
-    plt.savefig(tmp_path)
-    plt.close(fig)
+    pred_class = labels[pred_idx]
+    conf = probs[pred_idx]
 
+    # SHAP interpretation
+    top_words, shap_plot = get_shap_plot(model, vectorizer, text, pred_idx, pred_class)
+
+    # Final output
     explanation = f"""
-**Predicted Sentiment:** {label}\n
-**Confidence:** {probs[pred_idx]:.2f}\n
-**Top Influential Words:**\n
-{', '.join(tokens)}
+**Predicted Sentiment:** {pred_class}  
+**Confidence:** {conf:.2f}  
+**Top Influential Words:** {', '.join(top_words)}
 """
-    return explanation, tmp_path
+    return explanation, shap_plot
 
-# ---------------------------------------------------------
-# Gradio UI
-# ---------------------------------------------------------
-iface = gr.Interface(
+# ------------------------------------------------------------
+# 4️⃣ Gradio Interface
+# ------------------------------------------------------------
+title = "🌐 Multi-Lingual Sentiment Analysis (English + Persian)"
+description = """
+Select a language, type a comment, and see both the sentiment prediction and SHAP interpretability.
+"""
+
+examples = [
+    ["I love this product! Highly recommend.", "English"],
+    ["Worst experience ever, totally disappointed.", "English"],
+    ["The service was okay, nothing special.", "English"],
+    ["این محصول فوق‌العاده است", "Persian"],
+    ["تجربه‌ی بدی بود، ناراضی‌ام", "Persian"],
+    ["کیفیتش متوسط بود", "Persian"]
+]
+
+demo = gr.Interface(
     fn=predict_sentiment,
     inputs=[
         gr.Textbox(lines=3, label="Enter comment"),
-        gr.Radio(["English", "Persian"], label="Choose Dataset/Language")
+        gr.Radio(["English", "Persian"], label="Choose Dataset/Language", value="English")
     ],
     outputs=[
-        gr.Markdown(label="Prediction + Interpretation"),
+        gr.Markdown(label="Prediction & Explanation"),
         gr.Image(label="Top Word Contributions")
     ],
-    title="🌍 Multi-Lingual Sentiment Analysis (English + Persian)",
-    description="Select a language, type a comment, and see both the prediction and SHAP interpretability."
+    title=title,
+    description=description,
+    examples=examples,
 )
 
-iface.launch()
-
+if __name__ == "__main__":
+    demo.launch()