Update app.py

app.py

@@ -6,53 +6,59 @@ import base64
 
 # Load models
 emotion_pipeline = pipeline("text-classification", model="bhadresh-savani/distilbert-base-uncased-emotion")
-toxicity_pipeline = pipeline("text-classification", model="martin-ha/toxic-comment-model")
+EMO_THRESHOLD = 0.1  # Increased threshold for more confident detections
 
-EMO_THRESHOLD = 0.05  # Minimum confidence threshold for emotions
-
-# Comprehensive emotion mapping
+# Enhanced emotion mapping with common expressions
 EMOTION_MAPPING = {
-    "excitement": {
-        "underlying": ["Anticipation", "Eagerness", "Enthusiasm"],
-        "recommendation": "Channel this energy into planning and goal-setting!",
+    "joy": {
+        "underlying": ["Contentment", "Pleasure", "Delight", "Happiness"],
+        "recommendation": "Savor this positive feeling!",
         "polarity": 1,
-        "keywords": ["thrilled", "excited", "pumped"]
+        "keywords": ["happy", "joy", "glad", "delighted"]
+    },
+    "sadness": {
+        "underlying": ["Grief", "Loneliness", "Sorrow", "Melancholy"],
+        "recommendation": "It's okay to feel this way. Consider talking to someone.",
+        "polarity": -1,
+        "keywords": ["sad", "unhappy", "depressed", "miserable"]
     },
     "anxiety": {
         "underlying": ["Worry", "Unease", "Apprehension", "Nervousness"],
         "recommendation": "Try box breathing (4-4-4-4) and focus on preparation.",
         "polarity": -1,
-        "keywords": ["anxious", "nervous", "apprehensive"]
-    },
-    "joy": {
-        "underlying": ["Contentment", "Pleasure", "Delight"],
-        "recommendation": "Savor this moment and share your happiness!",
-        "polarity": 1,
-        "keywords": ["happy", "joyful"]
+        "keywords": ["anxious", "nervous", "worried", "stressed"]
     },
     "fear": {
-        "underlying": ["Dread", "Panic", "Terror"],
-        "recommendation": "Practice grounding techniques and focus on the present.",
+        "underlying": ["Dread", "Panic", "Terror", "Fright"],
+        "recommendation": "Practice grounding techniques to stay present.",
         "polarity": -1,
-        "keywords": ["scared", "afraid"]
-    },
-    "pride": {
-        "underlying": ["Accomplishment", "Self-worth", "Achievement"],
-        "recommendation": "Celebrate your success while staying humble.",
-        "polarity": 1,
-        "keywords": ["proud", "accomplished"]
+        "keywords": ["scared", "afraid", "fearful"]
     },
-    "disappointment": {
-        "underlying": ["Letdown", "Frustration", "Regret"],
-        "recommendation": "Acknowledge the feeling, then focus on what you can control.",
+    "anger": {
+        "underlying": ["Frustration", "Irritation", "Rage", "Resentment"],
+        "recommendation": "Take deep breaths before responding.",
         "polarity": -1,
-        "keywords": ["disappointed", "let down"]
+        "keywords": ["angry", "mad", "furious"]
+    },
+    "focus": {
+        "underlying": ["Concentration", "Attention", "Engagement"],
+        "recommendation": "Maintain this productive state with regular breaks.",
+        "polarity": 1,
+        "keywords": ["focused", "concentrating", "studying"]
     }
 }
 
 def create_visualization(emotions):
     fig, ax = plt.subplots(figsize=(8,5))
-    colors = ['#4CAF50' if EMOTION_MAPPING[e]['polarity'] > 0 else '#F44336' for e in emotions.keys()]
+    colors = []
+    for e in emotions.keys():
+        if e in EMOTION_MAPPING:
+            polarity = EMOTION_MAPPING[e]['polarity']
+            color = '#4CAF50' if polarity > 0 else '#F44336'
+        else:
+            color = '#9E9E9E'  # Neutral for unknown
+        colors.append(color)
+    
     ax.barh(list(emotions.keys()), list(emotions.values()), color=colors)
     ax.set_xlim(0,1)
     ax.set_xlabel("Confidence Score")
@@ -65,104 +71,88 @@ def create_visualization(emotions):
     plt.close()
     return f'<img src="data:image/png;base64,{base64.b64encode(buf.read()).decode("utf-8")}">'
 
-def detect_keyword_emotions(text):
+def detect_emotions(text):
+    # Keyword detection first
     emotions = {}
     text_lower = text.lower()
     
     for emo, data in EMOTION_MAPPING.items():
         for keyword in data['keywords']:
             if keyword in text_lower:
-                # Use higher confidence for exact matches
-                confidence = 0.9 if keyword in text_lower.split() else 0.7
+                confidence = 0.9 if any(kw == keyword for kw in text_lower.split()) else 0.7
                 if confidence > emotions.get(emo, 0):
                     emotions[emo] = confidence
-    return emotions
-
-def analyze_text(text):
-    # Detect keyword-based emotions first
-    keyword_emotions = detect_keyword_emotions(text)
     
-    # Get base model emotions
+    # Model detection for remaining
     emo_results = emotion_pipeline(text, top_k=None)
-    base_emotions = {e["label"].lower(): e["score"] for e in emo_results 
-                    if e["score"] > EMO_THRESHOLD}
-    
-    # Merge emotions with keyword emotions taking priority
-    final_emotions = {}
-    
-    # Add keyword emotions first
-    for emo, score in keyword_emotions.items():
-        final_emotions[emo] = score
+    for e in emo_results:
+        if e["score"] > EMO_THRESHOLD and e["label"].lower() not in emotions:
+            emotions[e["label"].lower()] = e["score"]
     
-    # Add base emotions only if not already detected by keywords
-    for emo, score in base_emotions.items():
-        # Skip if we already have a more specific version of this emotion
-        if (emo == 'joy' and 'excitement' in final_emotions) or \
-           (emo == 'fear' and 'anxiety' in final_emotions):
-            continue
-        if emo not in final_emotions:
-            final_emotions[emo] = score
+    return emotions
+
+def analyze_text(text):
+    emotions = detect_emotions(text)
     
-    # Generate detailed analysis
     details = ""
-    positive_emotions = []
-    negative_emotions = []
+    positive = []
+    negative = []
     
-    for i, (emo, score) in enumerate(final_emotions.items(), 1):
-        emotion_data = EMOTION_MAPPING.get(emo, {
-            "underlying": ["Complex feeling"],
-            "recommendation": "This emotion warrants careful reflection.",
+    for i, (emo, score) in enumerate(emotions.items(), 1):
+        data = EMOTION_MAPPING.get(emo, {
+            "underlying": ["Complex emotion"],
+            "recommendation": "This feeling deserves reflection.",
             "polarity": 0
         })
         
-        polarity = "Positive" if emotion_data["polarity"] > 0 else "Negative" if emotion_data["polarity"] < 0 else "Neutral"
+        pol = "Positive" if data["polarity"] > 0 else "Negative" if data["polarity"] < 0 else "Neutral"
         
-        details += f"Emotion {i}: {emo.capitalize()} ({polarity}, Confidence: {score:.2f})\n"
-        details += f"Underlying: {', '.join(emotion_data['underlying'])}\n"
-        details += f"Recommendation: {emotion_data['recommendation']}\n\n"
+        details += f"Emotion {i}: {emo.capitalize()} ({pol}, Confidence: {score:.2f})\n"
+        details += f"Underlying: {', '.join(data['underlying'])}\n"
+        details += f"Recommendation: {data['recommendation']}\n\n"
         
-        if emotion_data["polarity"] > 0:
-            positive_emotions.append(emo)
-        elif emotion_data["polarity"] < 0:
-            negative_emotions.append(emo)
+        if data["polarity"] > 0: positive.append(emo)
+        elif data["polarity"] < 0: negative.append(emo)
     
-    # Generate final analysis
     conclusion = "\n=== Final Analysis ===\n"
+    if positive: conclusion += f"Positive emotions: {', '.join(positive)}\n"
+    if negative: conclusion += f"Negative emotions: {', '.join(negative)}\n"
     
-    if positive_emotions:
-        conclusion += f"Positive emotions: {', '.join(positive_emotions)}\n"
-    if negative_emotions:
-        conclusion += f"Negative emotions: {', '.join(negative_emotions)}\n"
-    
-    # Context-aware final recommendation
-    if "excitement" in positive_emotions and "anxiety" in negative_emotions:
-        final_rec = "This exciting transition naturally brings some nerves. Make an action plan to build confidence."
-    elif positive_emotions and not negative_emotions:
-        final_rec = "Fully enjoy these positive feelings!"
-    elif negative_emotions and not positive_emotions:
-        final_rec = "Consider discussing these feelings with someone you trust."
+    # Enhanced recommendation logic
+    if "focus" in positive and not negative:
+        final_rec = "Great focus! Maintain this productive state."
+    elif positive and negative:
+        if "anxiety" in negative:
+            final_rec = "Channel your anxiety into constructive planning."
+        else:
+            final_rec = "Balance these mixed emotions with mindful reflection."
+    elif positive:
+        final_rec = "Enjoy these positive feelings!"
+    elif negative:
+        final_rec = "Consider discussing these feelings with someone."
     else:
-        final_rec = "Reflect on these mixed emotions to understand yourself better."
+        final_rec = "The emotional tone is neutral."
     
     conclusion += f"\nFinal Recommendation: {final_rec}"
     
-    visualization = create_visualization(final_emotions)
-    return details + conclusion, visualization
+    viz = create_visualization(emotions)
+    return details + conclusion, viz
 
 iface = gr.Interface(
     fn=analyze_text,
-    inputs=gr.Textbox(label="Your Text", placeholder="Describe your feelings..."),
+    inputs=gr.Textbox(label="How are you feeling?", placeholder="I feel..."),
     outputs=[
-        gr.Textbox(label="Emotion Analysis"),
-        gr.HTML(label="Visualization")
+        gr.Textbox(label="Emotion Breakdown"),
+        gr.HTML(label="Emotion Intensity")
     ],
-    title="Precision Emotion Analyzer",
-    description="Accurately detects specific emotions without duplicates, providing tailored recommendations.",
     examples=[
-        ["I'm thrilled about my promotion but anxious about the responsibilities"],
-        ["Feeling proud of my accomplishment yet disappointed with the recognition"],
-        ["I'm excited and nervous about the first day at my new job"]
-    ]
+        ["I am happy but at the same time anxious"],
+        ["I'm focused on studying but afraid of failure"],
+        ["I feel sad and lonely today"],
+        ["I'm angry about what happened"]
+    ],
+    title="Advanced Emotion Analyzer",
+    description="Accurately detects emotions in both simple and complex sentences."
 )
 
 iface.launch()