Purpose/Significance To extract adverse drug events based on the text of social network comments, and to provide references for drug research and safety regulation. Method/Process The FrameNet semantic theory is adopted and combined with the Medical Dictionary of Regulatory Activities terminology set to construct the adverse drug event classification lexicon. The lexicon and rule matching-based approach is used to identify event categories and frame elements, and semantic information is used to achieve the filling of adverse drug event frame. Result/Conclusion The selection of social network drug evaluation examples for adverse drug event information extraction is feasible and effective, which contributes to the in-depth application and value realization of the FrameNet semantic analysis method in the medical field.