CatDetect, a framework for detecting Catalan tweets

Abstract

This work deals with language detection. It includes new proposals ranging from lexicon and morphological analysis to an increasing use of machine learning solutions. In this case, the language study is focused on Catalan, a minority language. Difficulty even increases in detecting Catalan on tweets, messages written in the Twitter social network. To achieve that, a Twitter-Catalan corpus has been generated using lexicon and morphological approaches, which then will be used to create supervised models based on Machine Learning techniques. They are also evaluated in order to see which one obtains the best prediction score and thus, the best suitability to be used. The best model is to be used in a website, where users can test the algorithm interactively in a front-end webpage and in background by means of a webservice across a RESTful API.