A Gaussian mixture model over normalized Unicode block counts can score and filter noisy text corpora without hand-written character rules.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2019 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
uniblock: Scoring and Filtering Corpus with Unicode Block Information
A Gaussian mixture model over normalized Unicode block counts can score and filter noisy text corpora without hand-written character rules.