A user behavior prediction method, prediction device, storage medium and terminal equipment

Abstract

The present invention relates to the technical field of information processing, and in particular to a user behavior prediction method, a prediction device, a computer-readable storage medium, and a terminal device. The user behavior prediction method provided by the present invention includes: obtaining voice evaluation information of the user evaluation target service, and converting the voice evaluation information into corresponding text information; analyzing the text information, and determining the evaluation object and the evaluation object of the user evaluation target service Corresponding evaluation results; extract the voice segment corresponding to the evaluation result in the voice evaluation information; conduct phonetic analysis on the voice segment to obtain the emotion label and tone label corresponding to the evaluation result; according to the emotion label, tone label, evaluation object and evaluation result, Predict user behavior to comprehensively and accurately predict user behavior through all-round analysis of emotion, tone, evaluation object and evaluation content, avoid user behavior prediction errors, and improve the accuracy of user behavior prediction.

Description

technical field [0001] The present invention relates to the technical field of information processing, and in particular to a user behavior prediction method, a prediction device, a computer-readable storage medium, and a terminal device. Background technique [0002] In various industries, there are many cases of user loss due to poor service satisfaction and other reasons. Therefore, effective analysis of the reasons for user loss to reduce or avoid user loss has become the key to the development of various enterprises. The existing user churn analysis method basically predicts user behavior by determining the matching between user behavior data and behavior indicators, that is, only predicts whether a user is a potential churn user based on the matching with behavior indicators, and when the behavior indicators When the setting is unreasonable, this method of predicting user behavior based only on the matching between user behavior data and behavior indicators is very lik...

AI Technical Summary

Problems solved by technology

The existing user churn analysis method basically predicts user behavior by determining the matching between user behavior data and behavior indicators, that is, only predicts whether a user is a potential churn user based on the matching with behavior indicators, and when the behavior indicators When the settings are unreasonable, this method of predicting user behavior based only on the matching between user behavior data and behavior indicators is very likely to cause errors in user behavior prediction, which greatly reduces the accuracy of user behavior prediction

Images