Abstract: The course-oriented Chinese FAQ auto question answering system is a subsystem of network teaching platform, which can realize automatic question answering, but the FAQ base is almost built artificially, which is time-consuming and laborious, so an approach for automatic extraction from the Baidu know platform is proposed. The similarity calculation method of question similarity semantic and syntactic dependency relations combined is also proposed. We can get the keywords vector, dependency relation pairs of the question and FAQ library’s retrieval subset from the question, the former two are used by similarity calculation algorithm to obtain the similar question from the subset, and return the answer as a solution. For the questions beyond this set, this system uses artificial service to return the answer. The experimental results show that this system model has the ability of FAQ library automatic construction and a higher response rate of accuracy and continuously strengthen the response ability, so it has broad application prospects.

Key words: words:question answering system, sentence similarity, FAQ, DOM tree, network teaching platform