MODEL TRAINING METHOD, DEVICE AND EQUIPMENT USED FOR ANSWERING QUESTIONS AND STORAGE MEDIUM

摘要

A model training method, device and equipment used for answering questions, and a storage medium, which are applied to the technical field of deep learning and are used for solving the problem of the rate of answering questions queried by a user being low. The present method comprises: whenever an answer to a question raised by a user is fed back by means of a target deep learning model, detecting whether the user has submitted negative evaluation information regarding the answer (101); if detected that the user submits negative evaluation information regarding the answer, obtaining an evaluation question corresponding to the negative evaluation information (102); when the number of obtained evaluation questions reaches a preset first number threshold, performing unsupervised text clustering on each evaluation question until a preset condition is met, and then obtaining each question set by clustering (103); determining a vector center of each question set (104); for each question set, respectively calculating the distance between the vector center of the question set and the vector center of each preset question group (105); if the smallest distance obtained by calculation is less than a preset distance threshold value, combining said question set with a preset question group corresponding to the smallest distance (106); if the smallest distance obtained by calculation is greater than or equal to the preset distance threshold, determining the question set to be a new preset question group (107); and re-training the target deep learning model by employing each updated preset question group, so as to obtain a trained target deep learning model (108).