Our society is getting more and more documents that are generated and circulated through the internet, and the amount of articles to be read online is getting more and more. Our society is getting more and more documents that are generated and circulated through the internet, and the amount of articles to be read online is getting more and more. In this situation, it is impossible to read the contents of a large amount of text data in a short time and to grasp important contents. Therefore, the demand for summarizing a large amount of information as a core content is expected to grow more and more. Under this background, this study aims at deriving a method for document summarization as ‘the sentence level reduction for automatic text summarization’ which is an application of Korean information processing. The study was conducted in two directions. First, we built a seed sentence compression corpus by marked tags in the expression to be deleted. Second, we extracted possibly deleted syntactic structures to create a priori syntactic compressing rules. (Sogang University)