Semantic Shift Stability: Efficient Way to Detect Performance Degradation of Word Embeddings and Pre-trained Language Models.
Shotaro IshiharaHiromu TakahashiHono ShiraiPublished in: AACL/IJCNLP (1) (2022)
Keyphrases
- language model
- n gram
- language modeling
- pre trained
- probabilistic model
- language modelling
- translation model
- retrieval model
- document retrieval
- statistical language models
- test collection
- out of vocabulary
- speech recognition
- statistical language modeling
- query expansion
- information retrieval
- co occurrence
- smoothing methods
- training data
- data sets
- query terms
- semantic information
- relevance model
- keywords