Update app.py

app.py

@@ -436,12 +436,11 @@ class ContentAnalyzer:
                 time.sleep(self.retry_delay * (2 ** attempt))
 
     def _create_analysis_prompt(self, transcript: str) -> str:
-        """Create the analysis prompt with smart timestamp handling"""
+        """Create the analysis prompt with smart timestamp handling and stricter evaluation criteria"""
         # First try to extract existing timestamps
         timestamps = re.findall(r'\[(\d{2}:\d{2})\]', transcript)
         
         if timestamps:
-            # Use existing timestamps
             timestamp_instruction = f"""Use the EXACT timestamps from the transcript (e.g. {', '.join(timestamps[:3])}).
 Do not create new timestamps."""
         else:
@@ -453,17 +452,17 @@ Do not create new timestamps."""
 3. Format as [MM:SS]
 Example: If a quote starts at word 300, timestamp would be [02:00] (300 words / 150 words per minute)"""
             
-            # Add word position markers to help with timestamp calculation
+            # Add word position markers
             words = transcript.split()
             marked_transcript = ""
             for i, word in enumerate(words):
-                if i % 150 == 0:  # Add marker every ~1 minute of speech
+                if i % 150 == 0:
                     minutes = i // 150
                     marked_transcript += f"\n[{minutes:02d}:00] "
                 marked_transcript += word + " "
             transcript = marked_transcript
 
-        prompt_template = """Analyze this teaching content and provide detailed assessment.
+        prompt_template = """Analyze this teaching content with strict evaluation criteria. Score 1 only if ALL requirements are fully met.
 
 Transcript:
 {transcript}
@@ -475,61 +474,107 @@ Required JSON structure:
 {{
     "Concept Assessment": {{
         "Subject Matter Accuracy": {{
-            "Score": 1,
-            "Citations": ["[MM:SS] Quote from transcript"]
+            "Score": 1,  # Score 1 ONLY if NO factual errors or misconceptions are present
+            "Citations": ["[MM:SS] Quote demonstrating accuracy or error"]
         }},
         "First Principles Approach": {{
-            "Score": 1,
-            "Citations": ["[MM:SS] Quote from transcript"]
+            "Score": 1,  # Score 1 ONLY if fundamentals are clearly explained BEFORE introducing complex concepts
+            "Citations": ["[MM:SS] Quote showing fundamental concept explanation"]
         }},
         "Examples and Business Context": {{
-            "Score": 1,
-            "Citations": ["[MM:SS] Quote from transcript"]
+            "Score": 1,  # Score 1 ONLY if MULTIPLE relevant real-world examples are provided
+            "Citations": ["[MM:SS] Quote containing practical example"]
         }},
         "Cohesive Storytelling": {{
-            "Score": 1,
-            "Citations": ["[MM:SS] Quote from transcript"]
+            "Score": 1,  # Score 1 ONLY if concepts flow logically with clear transitions
+            "Citations": ["[MM:SS] Quote showing topic transition or connection"]
         }},
         "Engagement and Interaction": {{
-            "Score": 1,
-            "Citations": ["[MM:SS] Quote from transcript"]
+            "Score": 1,  # Score 1 ONLY if MULTIPLE engagement techniques are used effectively
+            "Citations": ["[MM:SS] Quote showing audience engagement"]
         }},
         "Professional Tone": {{
-            "Score": 1,
-            "Citations": ["[MM:SS] Quote from transcript"]
+            "Score": 1,  # Score 1 ONLY if consistently professional language is maintained
+            "Citations": ["[MM:SS] Quote demonstrating tone"]
         }}
     }},
     "Code Assessment": {{
         "Depth of Explanation": {{
-            "Score": 1,
-            "Citations": ["[MM:SS] Quote from transcript"]
+            "Score": 1,  # Score 1 ONLY if code concepts are explained thoroughly with implementation details
+            "Citations": ["[MM:SS] Quote showing detailed code explanation"]
         }},
         "Output Interpretation": {{
-            "Score": 1,
-            "Citations": ["[MM:SS] Quote from transcript"]
+            "Score": 1,  # Score 1 ONLY if code outputs are explained with their significance
+            "Citations": ["[MM:SS] Quote demonstrating output explanation"]
         }},
         "Breaking down Complexity": {{
-            "Score": 1,
-            "Citations": ["[MM:SS] Quote from transcript"]
+            "Score": 1,  # Score 1 ONLY if complex concepts are broken into manageable parts
+            "Citations": ["[MM:SS] Quote showing concept breakdown"]
         }}
     }}
 }}
 
-Evaluation Criteria:
-- Subject Matter Accuracy: Check for factual errors or incorrect correlations
-- First Principles Approach: Evaluate if fundamentals are explained before technical terms
-- Examples and Business Context: Look for real-world examples
-- Cohesive Storytelling: Check for logical flow between topics
-- Engagement and Interaction: Evaluate use of questions and engagement techniques
-- Professional Tone: Assess language and delivery professionalism
-- Depth of Explanation: Evaluate technical explanations
-- Output Interpretation: Check if code outputs are explained clearly
-- Breaking down Complexity: Assess ability to simplify complex concepts
+Strict Evaluation Criteria:
+
+Concept Assessment:
+1. Subject Matter Accuracy
+   - MUST have zero factual errors or misconceptions
+   - All technical terms used correctly
+   - No oversimplification that leads to incorrect understanding
+
+2. First Principles Approach
+   - MUST explain fundamental concepts before advanced topics
+   - Clear progression from basic to complex ideas
+   - Core concepts defined before being used in explanations
+
+3. Examples and Business Context
+   - MUST include at least 2 relevant real-world examples
+   - Examples should directly relate to the concept
+   - Business context should be clearly explained
+
+4. Cohesive Storytelling
+   - MUST have clear logical flow between topics
+   - Explicit connections between related concepts
+   - Smooth transitions between subjects
+   - No abrupt topic changes
+
+5. Engagement and Interaction
+   - MUST use at least 2 different engagement techniques
+   - Rhetorical questions or direct audience involvement
+   - Clear checks for understanding
+   - Interactive elements in presentation
+
+6. Professional Tone
+   - MUST maintain consistent professional language
+   - No casual or inappropriate expressions
+   - Clear and confident delivery
+   - Appropriate technical vocabulary
+
+Code Assessment:
+1. Depth of Explanation
+   - MUST explain both what code does AND why
+   - Implementation details clearly covered
+   - Key programming concepts explained
+   - Important design decisions justified
+
+2. Output Interpretation
+   - MUST explain significance of all outputs
+   - Clear connection between code and results
+   - Error cases covered
+   - Expected vs actual results discussed
+
+3. Breaking down Complexity
+   - MUST divide complex topics into clear sub-components
+   - Step-by-step explanation of difficult concepts
+   - Clear progression in complexity
+   - Logical organization of technical content
 
 Important:
-- Each citation must include a timestamp and relevant quote
-- Citations should highlight specific examples of criteria being met or missed
-- Use only Score values of 0 or 1"""
+- Score 1 ONLY if ALL criteria for that category are met
+- Each citation must include exact timestamp and relevant quote
+- Citations should clearly demonstrate how criteria are/aren't met
+- Be strict and objective in scoring
+- Default to 0 if any doubt about meeting ALL criteria"""
 
         return prompt_template.format(
             transcript=transcript,
@@ -2618,4 +2663,5 @@ def main():
         st.error(f"Application error: {str(e)}")
 
 if __name__ == "__main__":
-    main()
+    main()
+