Aim: detect the languages of the words and use this information for a POS tagging that tags the different languages in their context of a foreign language
Training and testing done in a 10-fold cross validation setting. Results so far 95% of accuracy
Algorithm used: CRF features: output and confidence of Middle-English tree tagger, Latin tree tagger, character-bigram and character-trigram context: window of 5 on all features
Waiting for train data