Native Language Identification: a Simple n-gram Based Approach

摘要

This paper describes our approaches to Native Language Identification (NLI) for the NLI shared task 2013. NLI as a sub area of author profiling focuses on identifying the first language of an author given a text in his second language. Researchers have reported several sets of features that have achieved relatively good performance in this task. The type of features used in such works are: lexical, syntactic and stylistic features, dependency parsers, psycholinguistic features and grammatical errors. In our approaches, we selected lexical and syntactic features based on n-grams of characters, words, Penn TreeBank (PTB) and Universal Parts Of Speech (POS) tagsets, and perplexity values of character of n-grams to build four different models. We also combine all the four models using an ensemble based approach to get the final result. We evaluated our approach over a set of 11 native languages reaching 75% accuracy.