Paragraph classification model training method and device and storage medium thereof

The invention discloses a paragraph classification model training method and device and a storage medium thereof, and belongs to the technical field of natural language processing. The paragraph classification model training method comprises the following steps: S1, obtaining a text training set of privacy policies; s2, extracting semantic features in a text sequence of the text training set to obtain semantic feature vectors; s3, encoding the key information type list of the text training set to generate an embedded vector of the key information; s4, generating an embedded vector of the position information; s5, splicing the semantic feature vector, the embedding vector of the key information and the embedding vector of the position information to obtain a feature representation vector of the paragraph text; and S6, constructing a classifier, transmitting the feature representation vector of the paragraph text into a full connection layer, and calculating through a softmax function to obtain probability dist