Improve the chat response logic to reduce repetitive phrases and provide more supportive, structured replies.
Requirements:
Avoid repeating phrases like “Tell me more” or “I’m here to listen” in consecutive messages
Keep detected emotions stable for multiple turns unless there is a clear shift
Combine empathy with gentle emotional grounding or reframing
Maintain a non-clinical, student-friendly tone
Response style guideline:
Acknowledge the emotion
Normalize the experience
Offer a small supportive perspective or reflective question
Goal:
Make the conversation feel more natural, supportive, and helpful while remaining appropriate for an academic mental health project.
I'll improve the chat response logic to be more varied, supportive, and natural while reducing repetitive phrases. Let me read the current implementation first.
Now I'll improve the chat response logic to reduce repetitive phrases, maintain emotional stability across turns, and provide more supportive structured replies with the Acknowledge → Normalize → Offer perspective pattern.
I've significantly improved the chat responses to follow the Acknowledge → Normalize → Offer perspective pattern, reducing repetitive phrases by rotating through varied responses that maintain meaning, keeping emotions stable across conversation turns unless there's a clear shift, and adding emotional grounding and gentle reframing to make conversations feel more supportive and natural while maintaining the student-friendly, non-clinical tone.
I'll improve the emotion detection and response progression to make the chat feel more emotionally intelligent and less repetitive. Let me read the current chat implementation first.
Now I'll improve the emotion detection and response progression according to the requirements. I'll update the system to better detect stress/anxiety, maintain emotional stability, avoid repetitive phrases, and follow the progression pattern.
I've significantly improved the emotion detection and response progression system. The key changes include enhanced keyword detection for Stressed (deadlines, studies, workload, pressure) and Anxious (worry, expectations, fear of failure), emotion progression that maintains stability across turns and only returns to Neutral when users explicitly state they feel calm, elimination of repetitive phrases by expanding response variations, and conversation tracking that prevents staying Neutral after the first meaningful message, making conversations feel more emotionally intelligent and supportive.