مقاربة لتحديد اللغات تلقائيا (بدون إشراف)‎ في مدينة نصوص متعدد اللغات

Abstract EN

We present in this paper an unsupervised approach for automatic language identification.

The method sorts the texts contained in a multilingual text corpus based on their language.

This method uses n-grams of characters for text representation and the AntClass algorithm applying artificial ants, for their classification.

The proposed algorithm is evaluated on a corpus composed of texts of three different languages (Arabic, English and French) and its effect will be examined in several experiments using the cosine distance, the Euclidean distance and the Manhattan distance as similarity measures and several values of n.