Abstract: 【Purpose/significance】Aiming at the problem that the research results of current network public opinion recognition have low precision and recall rate, a semantic vocabulary-based semantic network semantic recognition method for multimedia network emergencies is proposed.【Method/process】The public opinion processing and semantic retrieval platform is built by using the module data crawling and extraction module, and the public opinion related information data is processed through filtering, filtering and de⁃ noising operations, and the case database formed by the final information data is regarded as the lyric semantics. Identify the database. Based on the lyric database, the emotional tendency of lyric vocabulary is analyzed. Based on the lexical sentiment orientation, the in⁃ tensity value of the lyric topic comment emotion is calculated. Finally, the comprehensive comment semantic mode weight and its emo⁃ tional tendency value are used to obtain the final emotional tendency of the topic comment set, and the lyric semantic recognition is completed.【Result/conclusion】The experimental results show that the proposed method has high precision and recall rate and has sig⁃ nificant reliability. The proposed network emergency crisis response strategy is: design and improve the network emergency response public response plan; establish and optimize the network emergency public opinion monitoring and early warning program and re⁃ al-time response mechanism.【Innovation/limitation】Establish and optimize the network emergency public opinion network evaluation guidance mechanisms and public opinion guide the accountability system.