Automatic evaluation method and device for manufacturing industry video content quality

The invention provides a manufacturing industry video content quality-oriented automatic evaluation method, which is used for evaluating the content quality of a manufacturing industry video and comprises the following steps of: S1, performing voice recognition on the manufacturing industry video to obtain a voice recognition text; s2, performing further pre-training by using a pre-training language model predetermined by the speech recognition text to obtain a secondary pre-training language model; s3, generating a training question based on the speech recognition text, and training the secondary pre-training language model by using the training question to obtain a trained question and answer model; and S4, inputting a predetermined test question into the question and answer model to obtain a prediction probability as an evaluation result of the content quality of the manufacturing industry video. According to the method, one or more manufacturing industry videos and reference documents related to the video