Sequence labeling method and device

Abstract

The embodiment of the invention provides a sequence labeling method and device. The method comprises the steps of obtaining a sentence to be analyzed; inputting a sentence to be analyzed into the sequence labeling model, and outputting a word segmentation result of the sentence and a labeling result of each word; wherein the sequence annotation model is obtained by performing reinforcement learning-based training based on sample sentences and segmented words and annotation data corresponding to the sample sentences. According to the sequence labeling method and device provided by the embodiment of the invention, word segmentation information of the text is implicitly learned through a reinforcement learning technology, and mismatching between segmented words in Chinese sequence labeling and a target can be relieved.

Description

technical field [0001] The embodiments of the present invention relate to the field of computer technology, and more specifically, to a sequence labeling method and device. Background technique [0002] Sequence tagging technology is an important technology in the field of natural language processing. Basic tasks such as part-of-speech tagging, chunk analysis, and named entity recognition can all be solved using sequence tagging technology. Traditional methods use Hidden Markov Models, Maximum Entropy Markov Models, and Conditional Random Fields to solve the problem. In order to enhance the effect of the model, many manual features are added to the model, such as part of speech, gazetteer and other information are integrated into the task of named entity recognition. [0003] In the Chinese sequence tagging task, it is usually necessary to segment the text first, and then the sequence tagging pipeline will easily lead to the problem that the word segmentation does not match...

AI Technical Summary

Problems solved by technology

[0004] Embodiments of the present invention provide a sequence tagging method and device to solve, or at least partly solve, the defect that the word segmentation does not match the target in the prior art

Images