Reusing stanford POS tagger for tagging Urdu sentences
Adnan Naseem, Muazzama Anwar, Salman Ahmed, Avais Jan, Ahmad Kamran Malik · 2017
Several Natural Language Processing applications in a particular language consider POS tagging a necessary component. To develop a new language specific POS tagger targeting such particular language is a tedious job for unstructured data due to the variation in text, type and complexity of text. For that reason, it impacts the precision of tagging as a result of the variety of a certain language. Current research focused on the thought of reusability of a popular language specific Part of speech tagger, for example, Stanford Part of speech Tagger can be employed for tagging non-Engish phrases. For generalizeability, any translator can be used to translate the sentences, however, a well-known translator, named “Google translator” is considered for sentence translation purpose across the languages. For evaluation perspective, Urdu tweets of a hot political issue “Panama leaks” from twitter.com are extracted. To measure the accuracy, the kappa statistic along with confusion matrix is deliberated. The precision of tagging the Urdu sentences by reusing Stanford Part of speech tagger is 96.05 percent. The respected approach can be globally applied to tagging the sentences of several different languages.