An overview of Lithuanian internet media n-gram corpus

Abstract

This paper describes construction and properties of the open 70 million words Lithuanian Internet media n-gram corpus. Due to copyright limitations often contemporary media based resources availability is restricted, while n-grams corpora (e.g., Google N-gram viewer/corpus) solve the problem. Lithuanian language is under-resourced, hence n-gram corpus of Lithuanian media is designed to contribute to publicly available ready-to-use lexical resources. In this paper we report corpus construction procedure, preprocessing, corpus statistics and possible areas of application