Multitask learning Chinese language disease diagnosis method based on bidirectional long-short-term memory network

Abstract

The invention provides a multitask learning Chinese language disease diagnosis method based on a bidirectional long-short-term memory network. The method comprises the following steps of: a Chinese level examination composition correction data set provided by a Chinese grammar error diagnosis task (CGED) in recent years is obtained; a model input sequence is obtained through fusion word embedding,text features are fully utilized, and the diagnosis effect is improved; the obtained input sequence is input into Bi-LSTM, and a Chinese grammar error diagnosis model is obtained through learning ofa neural network; a multi-task learning method is adopted, whether sentences have morbidity or not is detected as a main task, the morbidity types and error positions of the sentences are detected asauxiliary tasks, hidden layer parameters are shared among multiple tasks, and output layers related to the tasks are reserved; according to the method, the relevance between the task of detecting whether the Chinese statement has the morbidity and the task of detecting the morbidity type and the error position is fully utilized; the problem that the detection effect is poor due to the fact that nogood feature exists in the process of detecting whether the Chinese statement has the morbidity task or not is solved, and meanwhile the generalization ability of the model is improved to a certain degree due to joint optimization of multiple tasks.

Description

technical field [0001] The invention relates to the field of computer technology, in particular to a method for diagnosing Chinese language disorders in multi-task learning based on a bidirectional long-short-term memory network. Background technique [0002] Chinese is considered to be one of the most difficult languages ​​in the world. Unlike English, Chinese does not have verb tenses and plurals, and there are usually multiple ways of expressing the same meaning in Chinese. Therefore, non-native speakers of Chinese often make various grammatical errors in their writing, and the research on the diagnosis of Chinese grammatical errors has become an urgent topic to be solved. [0003] In terms of grammar, due to the characteristics of Chinese itself and the limitations of Chinese theoretical research, the difficulty of Chinese grammar proofreading is higher than that of English. The difficulty of grammatical error checking lies in the fact that there is no morphological ch...

AI Technical Summary

Problems solved by technology

The difficulty of grammatical error checking lies in the fact that there are no morphological changes in Chinese parts of speech, there is no simple correspondence between parts of speech and syntactic components, and the flexibility of Chinese word order makes Chinese grammatical analysis very difficult.

Images