Fecha de publicación: 02/06/2023 Fuente: Wipo "BigData" The invention relates to the technical field of big data analysis, in particular to a public opinion analysis method based on a telephone hotline, and the method comprises the steps: S1, carrying out the preprocessing of voice data, and obtaining the preprocessing data; s2, performing feature prediction on the preprocessed data to obtain location features and personal features corresponding to the preprocessed data; s3, acquiring a current business analysis demand, and assembling the preprocessed data according to the business analysis demand, the location feature and the personal feature to obtain business data; s4, performing semantic matching on the business data to obtain a statistical event group; and S5, according to the statistical event group, generating an event theme for representing public opinions and outputting the event theme. The method has the beneficial effects that by extracting the location features and the personal features from the voice data and re-assembling the preprocessed data according to the current service analysis requirement, the statistical event group corresponding to a certain event occurrence place is accurately constructed, effective association of information related to the event occurrence place is realized, and the accuracy of the statistical event group is improved. And relatively high analysis efficiency can be realized.