CoT-MoTE: Exploring ConTextual Masked Auto-Encoder Pre-training with Mixture-of-Textual-Experts for Passage Retrieval.
Guangyuan MaXing WuPeng WangSonglin HuPublished in: CoRR (2023)
Keyphrases
- passage retrieval
- question answering
- document retrieval
- training set
- natural language
- query expansion
- audio visual
- multimedia
- contextual information
- language model
- artificial intelligence
- bit rate
- information retrieval
- expectation maximization
- relevance model
- natural language processing
- ground truth
- feature vectors
- feature space
- retrieval model
- metadata
- quantitative evaluation