Disentangling Semantics and Syntax in Sentence Embeddings with Pre-trained Language Models.
James Y. HuangKuan-Hao HuangKai-Wei ChangPublished in: NAACL-HLT (2021)
Keyphrases
- language model
- pre trained
- language modeling
- natural language
- document level
- training data
- document retrieval
- n gram
- dependency structure
- probabilistic model
- retrieval model
- information retrieval
- language modelling
- test collection
- query expansion
- sentence retrieval
- statistical language models
- smoothing methods
- speech recognition
- language models for information retrieval
- context sensitive
- training examples
- vector space model
- vector space
- generative model
- sentence level
- data sets
- document ranking
- neural network
- control signals
- pseudo relevance feedback
- relevant documents
- information retrieval systems
- dimensionality reduction
- supervised learning
- information extraction
- bayesian networks