Replacing raw video and audio features with MLLM-generated natural-language captions improves hit rate and nDCG for two-tower and SASRec recommenders on MicroLens-100K.
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
-
Describe What You See with Multimodal Large Language Models to Enhance Video Recommendations
Replacing raw video and audio features with MLLM-generated natural-language captions improves hit rate and nDCG for two-tower and SASRec recommenders on MicroLens-100K.