Recursive n-gram hashing is pairwise independent, at best.
Daniel LemireOwen KaserPublished in: Comput. Speech Lang. (2010)
Keyphrases
- n gram
- pairwise
- language model
- language modeling
- language independent
- bag of words
- variable length
- text classification
- viterbi algorithm
- part of speech
- similarity measure
- word segmentation
- language modelling
- neural network
- hash functions
- information retrieval
- character n grams
- similarity search
- semi supervised
- web documents
- real world