Slavic language identification using cascade classifier approach

Automatic Language Identification (LID) of a text is an important preprocessing step in many Text Mining applications. Most successful language identification systems perform very well in discrimination between unrelated languages on standard datasets. Nevertheless, this is not a solved problem, and there are a number of scenarios in which language identification has proven to be a quite challenging task. In this paper we focus on disambiguation between closely-related Slavic languages in the conditions of short and noisy texts using cascade classifier approach. We used the datasets and the evaluation metrics from LT4VarDial'2015 Workshop (RANLP'2015) in the shared task on Discriminating between Similar Languages (DSL).