Personalized programming support in higher education via emotion-guided prompt generation and GenAI
Primary research
#1171
- Topic
- unassigned (set during synthesis)
- First seen
- 2026-08-02 07:16:00
- Last seen
- 2026-08-02 07:16:00
Source raw items (1)
- Semantic Scholar2026-08-02 07:15:28Personalized programming support in higher education via emotion-guided prompt generation and GenAI
Personalized programming supports learners by incorporating their emotional states into an Artificial Intelligence (AI) system. However, existing systems face challenges, including a scarcity of emotion-adaptive tutors, limited generative AI tools for recognizing emotions, and a lack of prompt engineering techniques. This study proposes a personalized emotion-aware AI programming assistant that detects the user’s facial emotions and generates code explanations using generative AI. We develop a hybrid facial emotion recognition model using ResNet-50 to capture spatial features and LSTM to capture temporal patterns. The model classifies the learner’s emotions, which then informs emotion-guided prompt engineering to adjust programming queries based on detected emotions. The modified prompt is processed by a generative language model, displayed to the learner in a supportive and adaptive format. The model is tested in real-time with students. Experimental results show that the emotion-aware AI improved both learning performance and motivation in real-time classroom use, confirming its effectiveness in fostering engagement and better outcomes compared to a normal AI tutor.