Classifying Linux Shell Commands using Naive Bayes Sequence Model

Using Linux shell commands is a challenging task for most of the people new to Linux. This paper presents the idea of conversion of natural language to equivalent Linux shell command. To achieve the conversion we make use of a Naive Bayes text classifier. However there could be a case of a series of flags and combination of commands. This is handled by a sequence of Naive Bayes text classifier. Owing to the small amount of data set for every command the performance of naive Bayes is equivalent to that of the other discriminative classifiers like maximum entropy models. We improve the classification accuracy by combining the naive Bayes model with linear interpolation for predication of combination of multiple commands and flags. Keywords—Natural Language Processing; Linux; ; Naïve Bayes;