Marco Lui

Accurate Language Identification of Twitter Messages

We present an evaluation of “off-theshelf” language identification systems as applied to microblog messages from Twitter. A key challenge is the lack of an adequate corpus of messages annotated for language that reflects the linguistic diversity present on Twitter. We overcome this through a “mostly-automated” approach to gathering language-labeled Twitter messages for evaluating language identification. We present the method to construct this dataset, as well as empirical results over existing datasets and off-theshelf language identifiers. We also test techniques that have been proposed in the literature to boost language identification performance over Twitter messages. We find that simple voting over three specific systems consistently outperforms any specific system, and achieves state-of-the-art accuracy on the task.

Exploring Methods and Resources for Discriminating Similar Languages

The Discriminating between Similar Languages (DSL) shared task at VarDial challenged participants to build an automatic language identification system to discriminate between 13 languages in 6 groups of highly-similar languages (or national varieties of the same language). In this paper, we describe the submissions made by team UniMelb-NLP, which took part in both the closed and open categories. We present the text representations and modeling techniques used, including cross-lingual POS tagging as well as fine-grained tags extracted from a deep grammar of English, and discuss additional data we collected for the open submissions, utilizing custombuilt web corpora based on top-level domains as well as existing corpora.