This paper briefly summarizes the current development in the seismic characteristics and event type identification of earthquakes,blasting and collapses at home and abroad. Briefly with the definitions and main features of various types of events we focus on various identification methods for event type recognition. Unlike earthquakes formed by nature itself,non-natural earthquake is an event that is indirectly caused by artificial intervention or human activities. Blasting is a kind of event in which the explosive energy is rapidly released at the moment of explosion and propagates outward in the form of seismic waves,causing surface vibrations and destructive effects. The collapse is a kind of event caused by the collapse of the rock formation. Although the earthquakes recorded in the seismic network have certain common characteristics with the waveforms of blasting and collapse,there are significant differences in the P-wave first motion,the maximum amplitude ratio of P-wave and S-wave,duration time,phase,the development of short-period surface wave,seismogenic time,the distribution of spatial position and spectral characteristics due to the different types of sources,wave propagation paths,and focal depths,and etc. At present,there are two main methods to identify non-natural events such as earthquakes and blasting and collapse. A class of criteria based on waveform in terms of signal and data or clear geophysical significance is directly identified;such as waveform time-frequency analysis contrast method,wavelet transform,and etc. A class of algorithms based on the field of statistical pattern recognition;such as the minimum distance method,ICHAM,the Fisher,SAMC,SVM,and the feedforward neural network. In essence,the two types of methods first extract the effective feature criterion,that is,use the data for dimensionality reduction,and do not use all the information of the event record for event determination. Therefore,it is necessary to use an algorithm that can automatically extract various types of information from all event records and combine the underlying features to judge and identify various events.