Adapting Statistical Language Identification Methods for Short Queries.
Alexandru Ginsca, Emanuela Boroş, Adrian Iftene · 2011
This paper describes the participation of UAIC team at the LogCLEF 2011 initiative, language identification task. Our approach is an aggregation of known methods for recognizing languages. Short texts are a real challenge in applying a language identification tool; so, our methods had to comply with it by resisting to noisy data as only one letter, only numbers, links, different symbols. We applied n-grams extraction with distance measurement computing and a learning algorithm. The results were satisfying on specific languages, considering that our system supports only a limited number of languages.