Feature Selective Likelihood Ratio Estimator for Low- and Zero-frequency N-grams.
Masato KikuchiMitsuo YoshidaKyoji UmemuraTadachika OzonoPublished in: CoRR (2021)
Keyphrases
- n gram
- likelihood ratio
- hypothesis testing
- language model
- text classification
- hypothesis test
- bag of words
- language modelling
- variable length
- language independent
- likelihood ratio test
- part of speech
- confidence intervals
- inside outside algorithm
- feature set
- match scores
- feature vectors
- neural network
- document retrieval
- web documents
- information retrieval
- machine learning