摘要: Cross language text categorization is the task of exploiting labelled documents in a source (e.g. English) to classify target Chinese). In this paper, we focus on investigating use bilingual lexicon for cross categorization. To end, propose novel refinement framework The consists two stages. first stage, model transfer proposed generate initial labels language. second expectation maximization algorithm based naive Bayes introduced yield resulting documents. Preliminary experimental results collected corpora show that effective.