Text classification method and apparatus, device, and computer-readable storage medium

A text classification apparatus, comprising:

• a processor and a memory, the memory storing at least one instruction, at least one program, a code set, or an instruction set, the at least one instruction, the at least one program, the code set, or the instruction set being loaded and executed by the processor to implement:
• obtaining a target text and an associated text corresponding to the target text;
• performing feature extraction on the target text and the associated text to obtain feature information, the feature information comprising one or more features;
• processing the feature information by using an attention mechanism to obtain one or more attention weights respectively corresponding to the one or more features, an attention weight indicating importance of the corresponding feature for the target text and the associated text;
• and
• obtaining a class detection result and result interpretation information corresponding to the target text based on the feature information and the one or more attention weights, the class detection result comprising a class distribution probability corresponding to the target text, and the result interpretation information describing impact of the one or more features on the class detection result,
• wherein the text classification model processes the feature information for multiple times by using different model parameters to obtain the class detection result, the class detection result comprises n groups of class detection sub-results, the result interpretation information comprises n pieces of result interpretation sub-information corresponding to the n groups of class detection sub-results, the class detection sub-results and the result interpretation sub-information are in a one-to-one correspondence, and n is a positive integer;
• and
• the processor is further configured to implement:
• generating a final class detection result corresponding to the target text, wherein the final class detection result comprises a final probability value respectively corresponding to each class;
• determining confidence of the text classification model for the target text according to the final probability value respectively corresponding to each class;
• determining the final class detection result as an accurate detection result in response to the confidence meeting a condition;
• and
• generating prompt information in response to the confidence not meeting the condition, wherein the prompt information is used for reminding to perform manual analysis on the target text.

A non-transitory computer-readable storage medium, storing at least one instruction, at least one program, a code set, or an instruction set, the at least one instruction, the at least one program, the code set, or the instruction set being loaded and executed by a processor to implement:

• obtaining a target text and an associated text corresponding to the target text;
• performing feature extraction on the target text and the associated text to obtain feature information, the feature information comprising one or more features;
• processing the feature information by using an attention mechanism to obtain one or more attention weights respectively corresponding to the one or more features, an attention weight indicating importance of the corresponding feature for the target text and the associated text;
• and
• obtaining a class detection result and result interpretation information corresponding to the target text based on the feature information and the one or more attention weights, the class detection result comprising a class distribution probability corresponding to the target text, and the result interpretation information describing impact of the one or more features on the class detection result
• wherein the text classification model processes the feature information for multiple times by using different model parameters to obtain the class detection result, the class detection result comprises n groups of class detection sub-results, the result interpretation information comprises n pieces of result interpretation sub-information corresponding to the n groups of class detection sub-results, the class detection sub-results and the result interpretation sub-information are in a one-to-one correspondence, and n is a positive integer;
• and
• the processor is further configured to implement:
• generating a final class detection result corresponding to the target text, wherein the final class detection result comprises a final probability value respectively corresponding to each class;
• determining confidence of the text classification model for the target text according to the final probability value respectively corresponding to each class;
• determining the final class detection result as an accurate detection result in response to the confidence meeting a condition;
• and
• generating prompt information in response to the confidence not meeting the condition, wherein the prompt information is used for reminding to perform manual analysis on the target text.