RAGEAR: A Neuro-Symbolic Recommender for Academic Courses
A new neurosymbolic system called RAGEAR (Retrieval-Augmented Graph-Enhanced Academic Recommender) has been developed for recommending academic courses. This system integrates dense retrieval of complete lecture transcripts with a symbolic Knowledge Graph that represents courses, lessons, transcript segments, credits, study plans, and curricular details. The Knowledge Graph facilitates symbolic filtering and contextualization through structured criteria such as credits, academic fields, study plans, and prerequisites. Unlike traditional metadata methods, RAGEAR retrieves transcript segments that are semantically relevant to a student's inquiry. Its key innovation lies in a graph-aware aggregation function that transfers evidence from chunk-level to course-level recommendations, incorporating three elements: the proportion of retrieved similarity linked to a course, the rank-based relevance of its chunks, and additional contextual insights from the graph.
Key facts
- RAGEAR stands for Retrieval-Augmented Graph-Enhanced Academic Recommender.
- It is a neurosymbolic recommender system for academic course recommendation.
- It combines dense retrieval over full lecture transcripts with a symbolic Knowledge Graph.
- The Knowledge Graph models courses, lessons, transcript chunks, credits, study plans, and curricular information.
- The system supports symbolic filtering and contextualization based on structured constraints.
- It exploits fine-grained instructional content by retrieving transcript chunks semantically aligned with a student's query.
- The main contribution is a graph-aware aggregation function that propagates chunk-level evidence to course-level recommendations.
- The score combines three factors: share of retrieved similarity, rank-based strength of relevant chunks, and additional contextual signals.
Entities
—