N-gram models for code completion in Pharo

In this paper, I present applying statistical language models to improve code completion in Pharo. In particular, the goal is to use n-gram models for sorting the completion candidates and, in such a way, increase the relevancy of the suggested completions.