CN-122021621-A - Individualized English grammar diagnosis teaching system and method based on native language background

CN122021621ACN 122021621 ACN122021621 ACN 122021621ACN-122021621-A

Abstract

The invention relates to the field of artificial intelligence and education science and technology, and particularly discloses a personalized English grammar diagnosis teaching system and method based on a native language background, which are used for receiving input data containing English text and user background attributes, processing the text through an English grammar detection engine to generate a primary diagnosis result containing error identification information, inquiring an English grammar diagnosis rule base based on the primary diagnosis result and in combination with context information to generate fine-granularity error classification information, updating a personalized error portrait model corresponding to a user according to the error classification information, and searching and generating adaptive personalized coaching content from an English grammar teaching resource base based on the updated error portrait model and user history data, and outputting the adaptive personalized coaching content. The invention has the advantages of accurate identification of the negative migration of the native language, strong pertinence of personalized feedback, good integrity of the teaching closed loop and the like.

Inventors

LIU SHENG
LIANG JIANGUO
GONG LI

Assignees

山东石油化工学院

Dates

Publication Date: 20260512
Application Date: 20260204

Claims (8)

1. A personalized english grammar diagnostic teaching system based on native language background, comprising: the input module is used for receiving English text input by a learner and correlating native language background information of the learner; The error detection and diagnosis module is used for identifying grammar error positions in English texts, distributing error type labels from a preset grammar error classification system for each grammar error, associating error description information based on linguistic theory with the error type labels, carrying out attribution analysis on the errors based on native language background information, and outputting corrected texts; The learner model module is used for establishing and maintaining a dynamically updated user file for each learner, wherein the user file at least records time sequence data of the type of the historical errors and the occurrence frequency thereof and a grasping degree estimated value of a specific grammar knowledge point; the personalized feedback generation module is used for retrieving or generating a targeted grammar rule description from a preset multi-mode interpretation library containing cross-language comparison according to the error type label and the native language background information; and the output module is used for integrating and presenting corrected text, error diagnosis information, personalized grammar rule description and targeted practice problems to a learner.
2. The personalized English grammar diagnosis teaching system based on the native language background of claim 1, wherein the error detection and diagnosis module is realized by a fine-tuned pre-training neural network model, and the neural network model is trained on a data set containing a specific native language background learner annotation corpus in a multi-task learning mode, and simultaneously optimizes a text error correction task and an error type sequence annotation task.
3. The personalized English grammar diagnosis teaching system based on the native language background of claim 2, wherein the error detection and diagnosis module is realized by a mixed system based on rules and statistical features, and the rules in the rules and the statistical features are written according to common error modes corresponding to the native language background information.
4. The personalized English grammar diagnosis teaching system based on the native language background of claim 3, wherein the learner model module calculates the grasping degree estimation value through a knowledge tracking model, and the input of the knowledge tracking model comprises historical error type time sequence data.
5. The personalized English grammar diagnosis teaching system based on the native language background of claim 4, wherein the error detection and diagnosis module comprises a bottom universal detection layer and an upper rule diagnosis layer; the bottom layer general detection layer is configured with an English grammar detection engine and is used for executing initial grammar detection on an input text and outputting a structured preliminary correction suggestion set, wherein each correction suggestion is associated with a specific position in the text and comprises information for identifying errors and providing correction; The upper rule diagnosis layer maintains a diagnosis rule knowledge base which comprises a rule unique identifier, a trigger mode, a refined error type label and an applicable native language background field, wherein the trigger mode is used for matching an error type or a text mode returned by a general detection layer; The workflow of the error detection and diagnosis module is as follows: after receiving input data, the error detection and diagnosis module calls a bottom universal detection layer to obtain error correction suggestions, and the rule diagnosis layer queries a diagnosis rule knowledge base according to error rule identification, error context and learner native language background, maps the universal error correction suggestions into refined error type labels and outputs a structured diagnosis result.
6. The personalized English grammar diagnosis teaching system based on the native language background of claim 5, wherein the personalized feedback generation module is used for realizing feedback generation according to a preset teaching knowledge base, and the teaching knowledge base at least comprises an interpretation base and an exercise template base; The interpretation library comprises error types, a native language background, linguistic interpretation texts and cross-language comparison example sentence fields, and is used for providing targeted linguistic interpretation according to error type labels and learner native language backgrounds; The practice template library comprises error types, parameterized practice problem templates and difficulty level fields, and is used for generating personalized practice problems according to error type labels and learner historical weak points.
7. The personalized English grammar diagnosis teaching system based on the native language background of claim 6, wherein the feedback generation flow of the personalized feedback generation module comprises: Inquiring an explanation library to obtain a corresponding linguistic explanation text and a cross-language comparison sentence according to the error type label in the diagnosis result and the learner native language background; Inquiring a learner model module to obtain the first N error types with the highest accumulated occurrence times of the learner as a weak point set, selecting a target error type according to preset weight by combining the error types, selecting a corresponding template from a training template library, and instantiating through random replacement parameters to generate a targeted practice problem.
8. A personalized english grammar diagnostic teaching method based on the system of any one of claims 1-7, comprising the steps of: s1, receiving English text submitted by a learner and native language background information of the English text through an input module; s2, processing the English text through an error detection and diagnosis module, identifying grammar error positions, generating error type labels and corrected text, and carrying out error attribution based on the native language background information; S3, updating the user files of the corresponding learners through a learner model module according to the diagnosis result of the step S2; S4, generating personalized grammar rule description and targeted practice problems by a personalized feedback generation module through integrating error type labels, native language background information and history information in a user file; And S5, presenting the integrated teaching feedback to the learner through an output module, wherein the teaching feedback comprises corrected text, error diagnosis information, personalized grammar rule description and targeted practice problems.

Description

Individualized English grammar diagnosis teaching system and method based on native language background Technical Field The invention relates to the field of artificial intelligence and education science and technology, in particular to natural language processing, self-adaptive learning and personalized recommendation technology, and especially relates to a personalized English grammar diagnosis teaching system and method based on a native language background. Background In the teaching field of English as a second language (EFL), automatic and accurate correction of grammar errors is an important technical link for improving learning effect. The existing mainstream solutions are mainly divided into two categories, namely, the first category is authoring aids based on a general language model, such as Microsoft Editor and GRAMMARLY. The tool relies on a deep neural network pre-trained by massive English texts, and generates smooth and grammar-normalized texts through probability modeling. The design is initially a general authoring scene, and the specific requirements of EFL learners are not considered, so two fundamental limitations exist: (1) The diagnostic interpretation is inadequate, the nature of the error correction mechanism is probabilistic optimization of text, lacking deep linguistic analysis of error types. For example, "He even don't buy" is simply classified as "major consistent error", and this is a specific problem of "auxiliary verb third person called singular form missing" which results in weak feedback guidance. (2) And ignoring the influence of the native language migration, wherein the model training is completely based on English single language corpus and does not have cross-language analysis capability. For systematic errors caused by negative migration of the native language (such as the omission of the article by the chinese user and the confusion of the gender of the pronoun by the western user), the tool cannot detect the specific mode thereof, and cannot provide comparative explanation, so that the learner is difficult to eradicate the deep error habit. Another type is an adaptive learning system based on learning history. The system can realize content recommendation and difficulty adjustment on the level of macroscopic knowledge points by tracking answer records of learners. However, there are substantial limitations to its personalization mechanism: (1) Coarse granularity of diagnosis, that is, the system can only count the error frequency of a learner in a wide knowledge range (such as 'article use') and cannot accurately classify and attribute the language of a specific error. For example, it can find that a learner frequently makes mistakes on the "article" knowledge point, but cannot distinguish whether the mistakes are due to "miss indefinite articles before first referring to a countable noun", "misuse indefinite articles when referring to a particular noun", or "misadd indefinite articles before an countable noun". (2) Feedback and diagnosis are disjoint, namely, due to the lack of a mapping mechanism from a specific error mode to targeted teaching intervention, the system cannot provide corresponding rule interpretation and special exercises for different error types, so that the self-adaptive function is disjoint with the teaching target of deep error correction. Therefore, how to realize a technical scheme capable of accurately diagnosing English grammar errors and providing personalized teaching feedback based on a learner's native language background becomes a technical problem to be solved urgently. Disclosure of Invention The invention aims to solve the technical problem of providing a personalized English grammar diagnosis teaching system and method based on a native language background so as to realize complete technical closed loop from surface layer error correction to root diagnosis and personalized teaching intervention. In order to solve the technical problems, the technical scheme provided by the invention is that a personalized English grammar diagnosis teaching system based on a native language background comprises: the input module is used for receiving English text input by a learner and correlating native language background information of the learner; The error detection and diagnosis module is connected with the input module and used for identifying grammar error positions in English texts, distributing error type labels from a preset grammar error classification system for each grammar error, associating error description information based on linguistic theory with the error type labels, carrying out attribution analysis on the errors based on native language background information, and outputting corrected texts; the learner model module is connected with the error detection and diagnosis module and is used for establishing and maintaining a dynamically updated user file for each learner, wherein the user file at least records time sequence da