Chinese complex sentence hierarchy analysis is one of the challenging topics in the field of Chinese information processing. In order to solve the accuracy decline of automatic recognition caused by inadequate relationship marks, we analyze formal semantic knowledge that affects sentence association, based on which we construct a small sentence association recognition algorithm and apply it to the hierarchy decision rules of complex sentences to help analyze the hierarchical relationship. As for the analysis of the hierarchy of single and multiple marked complex sentences, we leverage the shift-reduce algorithm based on collocation rules. Finally, we propose a marked complex sentence hierarchy analysis model, which combines semantics and rules. Experimental results show that this method enhances the accuracy of automatic identification to a certain extent.