Method and system for extracting slot position information

Abstract

The invention provides a method and system for extracting slot position information, and the method comprises the steps: taking a business text after random fragment Mask as training data, carrying out the retraining of a pre-training model after Mask training, enabling the generated fragment Mask pre-training model to have the capability of analyzing and understanding the semantics of fragments in the business text. The business text after training the random fragment Mask is not labeled, so that the cost of manual labeling is saved. Further, a descriptive template with additional slot position information is introduced into the fragment Mask pre-training model, so that the finely-adjusted fragment Mask pre-training model can use a small amount of descriptive templates with marked slot position information, and the finely-adjusted fragment Mask pre-training model can have sufficient semantic understanding on specific tasks such as slot position extraction, therefore, the fine-tuned fragment Mask pre-training model can automatically identify the slot position information of the text to be extracted, the required annotation data is less, and the slot position information of a plurality of slot positions can be extracted at one time when the slot position information is extracted.

Description

technical field [0001] The present application relates to the technical field of natural language processing, in particular to a method and system for extracting slot information. Background technique [0002] Slot extraction is a common task type in natural language processing tasks, with applications in dialogue systems, data mining, and entity extraction. Slot extraction usually defines the slot information to be extracted, such as: consumption amount, announcement subject, etc., and then finds the corresponding slot information in the text to be extracted. [0003] The traditional slot information extraction method is generally realized by using various deep learning models (such as RNN, BiLSTM and Transformer, etc.). For each slot, a large amount of pre-labeled data is required for slot information extraction training. Pre-labeling requires manual labeling, which not only requires high labor costs, but also requires a lot of time costs. Contents of the invention [...

AI Technical Summary

Problems solved by technology

[0004] Based on this, it is necessary to provide a slot information extraction method and system for the problem that the traditional slot information extraction method requires a large amount of pre-labeled data, resulting in high labor costs and time costs.

Images