A Ngram-based Statistical Machine Translation Approach for Text Normalization on Chat-speak Style Communications

Carlos A. Henr ́ iquez, Adolfo Hern · 2009

This paper reports our participation on the text normalization shared task campaign organized by the CAW 2.0 workshop. Through a Statistical Machine Translation (SMT) system we managed to produce sentences syntactically correct given sentences written with misspelled words and chatting slangs. This approach was applied on the evaluation campaign’s test set to measure its performance. Here the results over a development and test set are analyzed and commented.

Read the paper · More papers on PaperTik