With the development of educatinal technology, more and more people use PPT documents in their learning process. By constructing a concept map for PPT doc uments, learners can quickly and comprehensively understand all the knowledge points of the PPT documents, being beneficial for the learners to speed up the learning and collect his learning behaviors. In this paper, we proposed an algorithm using Microsoft Office programming technology, text mining technology and social network analysis technology to automatically extract concept terminologies and the relationships between them from the PPT documents to construct concept map. The experimental results show that the psoposed algorithm can effectively calculate the importance of each concept terminology. And all the concept terminologies extracted by the proposed algorithm have a certain accuracy, the more important the concept terminology, the higher the accuracy.