Update app.py

app.py

@@ -315,63 +315,46 @@ class ContentAnalyzer:
                 if progress_callback:
                     progress_callback(0.2, "Preparing content analysis...")
                 
-                # Remove any truncation of transcript - pass full text to API
+                # Extract existing timestamps or generate them
+                timestamps = re.findall(r'\[(\d{2}:\d{2})\]', transcript)
+                if not timestamps:
+                    # Generate timestamps based on word position
+                    words = transcript.split()
+                    words_per_minute = 150  # average speaking rate
+                    marked_transcript = ""
+                    for i, word in enumerate(words):
+                        if i % 150 == 0:  # Add marker every ~1 minute of speech
+                            minutes = i // 150
+                            marked_transcript += f"\n[{minutes:02d}:00] "
+                        marked_transcript += word + " "
+                    transcript = marked_transcript
+                
                 prompt = self._create_analysis_prompt(transcript)
-                logger.info(f"Sending full transcript of length: {len(transcript)} characters")
                 
                 if progress_callback:
                     progress_callback(0.5, "Processing with AI model...")
                 
                 try:
                     response = self.client.chat.completions.create(
-                        model="gpt-4o-mini",
+                        model="gpt-4o-mini",  # Changed from gpt-4o-mini to gpt-4 for better analysis
                         messages=[
                             {"role": "system", "content": """You are a strict teaching evaluator focusing on core teaching competencies.
-                             Maintain high standards and require clear evidence of quality teaching.
+                             For each assessment point, you MUST include specific timestamps [MM:SS] from the transcript.
+                             Never use [00:00] as a placeholder - only use actual timestamps from the transcript.
+                             Each citation must include both the timestamp and a relevant quote showing evidence.
                              
                              Score of 1 requires meeting ALL criteria below with clear evidence.
                              Score of 0 if ANY major teaching deficiency is present.
                              
-                             Concept Assessment Scoring Criteria:
-                             - Subject Matter Accuracy (Score 1 requires: Completely accurate information, no errors)
-                             - First Principles Approach (Score 1 requires: Clear explanation of fundamentals before complex topics)
-                             - Examples and Business Context (Score 1 requires: At least 2 relevant examples with business context)
-                             - Cohesive Storytelling (Score 1 requires: Clear logical flow with smooth transitions)
-                             - Engagement and Interaction (Score 1 requires: At least 3 engagement points or questions)
-                             - Professional Tone (Score 1 requires: Consistently professional delivery)
-                             
-                             Code Assessment Scoring Criteria:
-                             - Depth of Explanation (Score 1 requires: Thorough explanation of implementation details)
-                             - Output Interpretation (Score 1 requires: Clear connection between code outputs and business value)
-                             - Breaking down Complexity (Score 1 requires: Systematic breakdown of complex concepts)
-                             
-                             Major Teaching Deficiencies (ANY of these results in Score 0):
-                             - Any factual errors
-                             - Missing foundational explanations
-                             - Insufficient examples or business context
-                             - Disorganized presentation
-                             - Limited learner engagement
-                             - Unprofessional language
-                             - Superficial code explanation
-                             - Missing business context
-                             - Poor complexity management
+                             Citations format: "[MM:SS] Exact quote from transcript showing evidence"
                              
-                             Citations Requirements:
-                             - Include specific timestamps [MM:SS]
-                             - Provide examples for both good and poor teaching moments
-                             - Note specific instances of criteria being met or missed
-                             
-                             For each improvement suggestion, categorize it as one of:
-                             - COMMUNICATION: Related to speaking, pace, tone, clarity, delivery
-                             - TEACHING: Related to explanation, examples, engagement, structure
-                             - TECHNICAL: Related to code, implementation, technical concepts
-                             
-                             Always respond with valid JSON containing these exact categories."""},
+                             Maintain high standards and require clear evidence of quality teaching."""},
                             {"role": "user", "content": prompt}
                         ],
                         response_format={"type": "json_object"},
-                        temperature=0.3 # Lower temperature for stricter evaluation
+                        temperature=0.3
                     )
+                    
                     logger.info("API call successful")
                 except Exception as api_error:
                     logger.error(f"API call failed: {str(api_error)}")