We present two simple finite-state transducer based strategies for tweet normalization. One relies on hand-written correction rules designed to capture commonly occurring misspellings and abbreviations, while the other tries to automatically induce an error model from a gold standard corpus of normalized tweets.
CITATION STYLE
Hulden, M., & Francom, J. (2013). Weighted and unweighted transducers for tweet normalization. In CEUR Workshop Proceedings (Vol. 1086, pp. 69–72). CEUR-WS.
Mendeley helps you to discover research relevant for your work.