"Average" Approximates "First Principal Component"? An Empirical Analysis on Representations from Neural Language Models.
Zihan WangChengyu DongJingbo ShangPublished in: EMNLP (1) (2021)
Keyphrases
- language model
- principal components
- language modeling
- principal component analysis
- word clouds
- n gram
- document retrieval
- retrieval model
- language modelling
- probabilistic model
- query expansion
- dimensionality reduction
- statistical language models
- information retrieval
- test collection
- neural network
- context sensitive
- kernel space
- speech recognition
- mixture model
- vector space model
- pseudo relevance feedback
- relevance model
- ad hoc information retrieval
- language models for information retrieval
- principal component regression
- smoothing methods
- language model for information retrieval
- translation model
- query specific
- document ranking
- retrieval effectiveness
- pattern recognition