NextTerm
Constraint-aware course scheduling with personalized, explainable AI recommendations.
Overview
Course and professor feedback is unstructured, while schedule decisions need explicit preferences and constraints. Useful recommendations must explain their reasoning and avoid time conflicts.
NextTerm extracts structured feedback tags with IBM watsonx and Granite, then applies constraint-aware decision logic to create personalized, conflict-free schedules.
How it works
- Course and professor feedback is processed into structured tags.
- Tags describe workload, teaching style, grading, and attendance.
- Student preferences, course data, and scheduling constraints inform the decision logic.
- Recommendations include explanations of the resulting choices.
Technical implementation
- LLM orchestration converts qualitative feedback into consistent attributes.
- Preference modeling connects those attributes to student priorities.
- Constraint-aware scheduling handles conflicts rather than relying solely on generated recommendations.
Feedback can be incomplete or biased, and extracted attributes can be wrong. The supplied material does not establish integration with live university enrollment or prerequisite systems.
Results
2nd Place in the General Track at the IBM-sponsored AI Hackathon @ OSU. Built explainable recommendations and personalized scheduling logic.
The resume documents implemented constraint-aware logic and conflict-free scheduling. Automated test coverage, extraction accuracy, and user evaluation results were not supplied.