Subordinate clause detection method and device

Abstract

The embodiment of the invention provides a subordinate clause detection method and device. The method comprises the following steps that: obtaining a sentence to be detected; obtaining vector information related to the context of each minimum analysis unit in the sentence; according to a pre-constructed subordinate clause position detection model and the vector information related to the context of each minimum analysis unit, obtaining the position type of each minimum analysis unit, wherein the position type is used for indicating a position relationship between each analysis unit and the subordinate clause in the sentence; and according to the position type of each minimum analysis unit, determining the position of the subordinate clause in the sentence. In the embodiment of the invention, whether characters in a sentence to be identified appears in a training set or not is not depended, the problem caused by spare character level features in the prior art is overcome, and subordinate clause detection accuracy is greatly improved.

Description

technical field [0001] The invention relates to the technical field of natural language processing, in particular to a clause detection method and device. Background technique [0002] With the development of information technology, the amount of text information faced by people is also growing explosively, and the processing technology of text information is also constantly evolving. Taking the field of education as an example, in recent years, with the continuous development of educational information technology, various intelligent teaching systems have gradually been applied in daily teaching, such as automatic corrections, learning situation diagnosis and other automatic processing methods, which have gradually become popular among teachers and students. accepted. Among them, the automatic correction of composition is an important part of intelligent teaching, which can not only help teachers improve the efficiency of composition correction, but also improve students' ...

AI Technical Summary

Problems solved by technology

[0003] The inventor found in the process of realizing the present invention that the clause detection in the prior art is mostly based on extracting specific character-level features to train the statistical model, and the reliability and accuracy of the results depend entirely on the sentences written by the students Whether the relevant words and structures appear in the training set (that is, there is a problem of sparse character-level features), but in practical applications, students’ writing is ever-changing, and even the same meaning may have many different expressions. Therefore, the existing The clause detection method in the technique does not work well

Images