A session-based recommender that uses a large language model to infer multiple user intents from a GNN-selected candidate set and aligns them with the GNN's structural representation, improving ranking metrics on Beauty and ML-1M.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.IR 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Integrating LLM-Derived Multi-Semantic Intent into Graph Model for Session-based Recommendation
A session-based recommender that uses a large language model to infer multiple user intents from a GNN-selected candidate set and aligns them with the GNN's structural representation, improving ranking metrics on Beauty and ML-1M.