我们通过对1200万字语料的统计得出,派生词约占词条总数的8.66%,构成派生词的词缀共有188个。其中,后缀“者”所构成的派生词词条数最多,构词成分最为复杂。我们采用基本词表、词例知识规则并结合词语的搭配、共现频率的混合策略对带后缀“者”的派生词进行了自动识别,封闭测试的精确率为93.06%,开放测试的精确率为82.40%。 Through the statistics of 12 million words of corpus, we derive about 8.66% of the total number of terms and 188 of affixes that constitute the derivative. Among them, the suffix “person” constitutes the largest number of derivative words, the most complicated composition of the word. We use the basic vocabulary, word case knowledge rules and word combinations, the co-occurrence frequency of hybrid strategies for the derivative with “suffix” of the derivatives were automatically identified, the accuracy of the closed test was 93.06%, open test accuracy 82.40%.