Abstract: In traditional TFIDF Chinese keywords intelligent extraction model, meet the redundant, or dynamic words, the words mining result is bad. Proposed based on the bayesian reasoning and TFIDF algorithm of Chinese keywords intelligent extraction method. Using bayesian statistical principle to text information of probability statistics. Using bayesian decision-making theory to optimize TFIDF algorithm. Overcome traditional TFIDF algorithm of defects. The experimental results show that the optimal algorithm for Chinese keywords TFIDF in intelligent extraction, the accuracy of the extraction greatly increased.