Personalized translation teaching system combining multiple modes and large language model

Abstract

The invention discloses a personalized translation teaching system combining multiple modes and a large language model, and the system comprises a task obtaining module which is used for obtaining the translation content of a student; the post-translation editing evaluation module is used for analyzing errors in translation contents according to a large language model and a deep learning algorithm and providing personalized correction suggestions; the corpus updating module is used for dynamically updating a corpus according to the student translation data and preset translation requirements; and the translation intelligent feedback module is used for providing personalized feedback according to the translation content and providing real-time improvement suggestions for voice pronunciation and picture translation of the students in combination with a multi-modal technology. According to the technical scheme, real-time, efficient and personalized translation learning experience can be provided for students, and the quality and efficiency of translation teaching are greatly improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of translation teaching, and in particular relates to a personalized translation teaching system combining multimodality and a large language model. Background Art

[0002] At present, artificial intelligence and large language models have brought about a dimensionality reduction attack on foreign language and translation teaching. Traditional translation teaching methods usually rely on manual scoring and static teaching materials, lack real-time feedback on students' translation process, and fail to effectively provide personalized learning paths. At the same time, existing translation teaching systems usually focus on text translation, ignoring multimodal factors such as voice and images, and limiting students' ability to understand and use language in a "scenario-based" manner. Therefore, existing translation teaching methods have certain shortcomings in improving students' translation ability, providing personalized teaching and cross-cultural understanding. Summary of the invention

[0003] The purpose of the present invention is to provide a personalized translation teaching system combining multimodality and a large language model to solve the problems existing in the above-mentioned prior art.

[0004] To achieve the above objectives, the present invention provides a personalized translation teaching system combining multimodality and a large language model, comprising:

[0005] Task acquisition module, used to obtain students' translation content;

[0006] The post-editing evaluation module is used to analyze errors in the translated content based on large language models and deep learning algorithms, while providing personalized correction suggestions and optimizing the translation model for the machine translation system;

[0007] The corpus update module is used to dynamically update the corpus based on student translation data and preset translation requirements;

[0008] The translation intelligent feedback module is used to provide personalized feedback based on the translation content, and combines multimodal technology to provide real-time improvement suggestions for students' voice pronunciation and picture translation.

[0009] Optionally, the post-editing evaluation module includes:

[0010] The correction suggestion submodule is used to identify errors in grammar, semantics and pragmatic dimensions in the translated content, evaluate the identified errors according to preset evaluation criteria, and provide personalized correction suggestions, which include suggestions for improving content quality and suggestions for improving machine translation accuracy.

[0011] Optionally, the post-editing evaluation module further includes:

[0012] An iterative evaluation submodule, which is used to re-evaluate the revised translation content, wherein the revised translation content is the translation content adjusted by the student according to the personalized correction suggestions;

[0013] The machine translation feedback submodule is used to process the post-editing data and provide feedback to the machine translation system to optimize its translation model and improve the quality of future translations.

[0014] Optionally, the corpus updating module includes:

[0015] The dynamic update submodule is used to analyze the students' translation content, error patterns and the changing content data of the translation tasks, and dynamically update the corpus and term base based on the changing content data.

[0016] Optionally, the corpus updating module further includes:

[0017] The automatic supplement and optimization submodule is used to supplement and optimize the translation resources of the corpus according to the preset translation requirements of different languages ​​and preset fields.

[0018] Optionally, the translation intelligent feedback module includes:

[0019] A real-time analysis submodule, used to perform real-time text analysis on the translated content and provide personalized feedback based on the text analysis results;

[0020] The multimodal feedback submodule is used to detect the student's speech pronunciation through speech recognition technology, obtain pronunciation error detection results, and provide pronunciation correction suggestions based on the pronunciation error detection results; identify the pictures and corresponding translation problems in the student's translation based on the image semantic analysis technology, and provide improvement suggestions based on the picture translation problems.

[0021] Optional, personalized translation teaching system also includes:

[0022] The learning path generation module is used to generate dynamic personalized learning paths based on students' learning progress, translation performance and error types.

[0023] Optional, personalized translation teaching system also includes:

[0024] The feedback optimization module is used to improve the system's understanding of context and translation accuracy by integrating multimodal data, and introduces multi-round dialogues, feedback mechanisms, and joint optimization strategies to adjust and optimize the teaching system.

[0025] The technical effects of the present invention are:

[0026] The present invention provides students with comprehensive translation feedback by integrating multimodal technology, thereby significantly improving the efficiency and effect of translation teaching. At the same time, it can enhance students' cross-cultural understanding ability, and achieve continuous optimization and dynamic adjustment to ensure the intelligence and personalization of translation teaching. The personalized translation teaching system of the present invention can make full use of large language models and multimodal technology to provide students with real-time, efficient and personalized translation learning experience, greatly improving the quality and efficiency of translation teaching. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0028] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0029] Figure 1 It is a structural block diagram of a personalized translation teaching system combining multimodality and a large language model in an embodiment of the present invention;

[0030] Figure 2 Schematic diagram of the workflow of the post-editing evaluation module in an embodiment of the present invention;

[0031] Figure 3 is a working principle diagram of the translation intelligent feedback module in an embodiment of the present invention;

[0032] Figure 4 FIG. 4 is a schematic diagram of the operation of the corpus updating module in an embodiment of the present invention. DETAILED DESCRIPTION

[0033] Various exemplary embodiments of the present invention will now be described in detail. This detailed description should not be considered as limiting the present invention, but should be understood as a more detailed description of certain aspects, features, and embodiments of the present invention.

[0034] It should be understood that the terms described in the present invention are only for describing special embodiments and are not intended to limit the present invention. In addition, for the numerical range in the present invention, it should be understood that each intermediate value between the upper and lower limits of the scope is also specifically disclosed. Each smaller range between the intermediate value in any stated value or stated range and any other stated value or intermediate value in the described range is also included in the present invention. The upper and lower limits of these smaller ranges can be independently included or excluded in the scope.

[0035] It will be apparent to those skilled in the art that various modifications and variations may be made to the specific embodiments of the present invention description without departing from the scope or spirit of the present invention. Other embodiments derived from the present invention description will be apparent to those skilled in the art. The present application description and examples are exemplary only.

[0036] The words “include,” “including,” “have,” “contain,” etc. used in this article are open-ended terms, meaning including but not limited to.

[0037] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0038] like Figure 1 - Figure 4 As shown, this embodiment provides a personalized translation teaching system combining multimodality and a large language model, including:

[0039] Task acquisition module, used to obtain students' translation content;

[0040] The post-editing evaluation module is used to analyze errors in the translated content based on large language models and deep learning algorithms, while providing personalized correction suggestions and optimizing the translation model for the machine translation system;

[0041] The corpus update module is used to dynamically update the corpus based on student translation data and preset translation requirements;

[0042] The translation intelligent feedback module is used to provide personalized feedback based on the translation content, and combines multimodal technology to provide real-time improvement suggestions for students' voice pronunciation and picture translation.

[0043] At present, translation teaching faces the problems of insufficient personalized feedback, difficulty in tracking learning progress, and lack of multimodal feedback. Traditional methods rely on manual scoring and static resources, and cannot efficiently adapt to students' translation errors and personalized needs. To solve these problems, this embodiment proposes an intelligent translation teaching platform that combines a large language model and multimodal technology. The platform significantly improves the translation teaching effect through three core functional innovations. Post-editing evaluation function: Based on a large language model and custom evaluation criteria, the platform automatically evaluates the post-editing results, identifies common errors in grammar, semantics, and cross-cultural contexts, provides personalized correction suggestions to translators, and feeds back post-editing to machine translation. Corpus update function: The platform dynamically optimizes the translation corpus based on teaching resources, terminology libraries, student translation content, and error patterns to ensure that the latest and most relevant translation resources are provided. Translation intelligent feedback function: Combined with a large language model, the platform analyzes student translations in real time and provides personalized feedback such as grammar correction, vocabulary optimization, and cross-cultural background analysis. At the same time, the platform uses multimodal technology to provide intelligent suggestions for speech pronunciation and picture semantic translation. The platform effectively improves the efficiency and personalization of translation teaching and enhances cross-cultural understanding. It is suitable for translation teaching, language learning software, multimodal educational tools and other fields, and has broad application prospects.

[0044] The personalized translation teaching system of this embodiment includes three core modules: post-editing evaluation module, corpus update module and translation intelligent feedback module. The system analyzes and optimizes students' translation content through a large language model, and provides multimodal feedback by combining voice and image technology. Specifically, the system analyzes and generates feedback in real time based on students' submissions, helping students understand errors in translation and how to improve them.

[0045] The post-editing evaluation module automatically evaluates post-editing results based on a large language model and customized evaluation criteria, identifies and corrects common errors in grammar, semantics, and cross-cultural contexts, and provides personalized correction suggestions. The evaluated translations are divided into two directions: one is to improve the content quality for translators and readers, and the other is to focus on the large model to help machine translation be more accurate and achieve mutual benefit between human and machine.

[0046] The corpus update module dynamically optimizes the translation corpus based on teaching resources, students' translation content, terminology database, case database, etc. to ensure that the latest and most relevant translation resources are provided;

[0047] The translation intelligent feedback module analyzes the translation content submitted by students in real time, provides personalized feedback such as grammar correction, semantic optimization, and cross-cultural background analysis, and combines multimodal technology to provide real-time improvement suggestions for students' voice pronunciation and picture translation.

[0048] The specific settings of each module include:

[0049] Post-editing evaluation module: The post-editing evaluation module is based on a large language model and uses deep learning algorithms to analyze students' post-editing content. This module first identifies common grammatical, semantic and cross-cultural errors in translation, and then provides personalized correction suggestions based on preset evaluation criteria and the output results of the large language model. Students can adjust the translation based on the feedback, and the system will re-evaluate based on the modification results to ensure the gradual improvement of translation quality. Post-editing after evaluation can be divided into two directions. The first dimension is to improve the content quality for translators and readers, and the other dimension can be oriented towards machine translation to help machine translation be more accurate and achieve mutual benefit between human and machine interaction.

[0050] The first feasible direction is to generate personalized post-editing evaluation and machine translation comparison reports, and automatically recommend optimization solutions to help students adjust their translation strategies. The second direction is to help the machine translation system continuously optimize and improve translation quality.

[0051] Corpus Update Module: The corpus update module dynamically updates the corpus and terminology database by analyzing students' translation data, error types, and changes in translation tasks. The system automatically supplements and optimizes translation resources based on translation needs in different languages ​​and specific fields. Every time a student completes a translation task, the system automatically updates the corpus to ensure the timeliness and relevance of the translation resources, thereby providing students with the latest translation references.

[0052] Translation Intelligent Feedback Module: Based on the large language model, it can analyze students' translation content in real time, adjust the model output in real time according to the translation object, translation field, and text context under different social and cultural backgrounds, provide precise guidance on error types, context differences and cultural conflicts, and provide personalized feedback, including grammar correction, vocabulary optimization, text polishing and cross-cultural background interpretation.

[0053] The translation intelligent feedback module analyzes students' translation results in real time based on a large language model and provides personalized feedback through multimodal technology. This module not only includes traditional text analysis, but also detects students' pronunciation through speech recognition technology, and provides authentic pronunciation and pronunciation correction suggestions. At the same time, the multimodal large model system combines image semantic analysis technology and uses image recognition technology to provide feedback on students' translated images and provide improvement suggestions for image translation; these intelligent feedbacks can help students improve their comprehensive language application ability during the translation process.

[0054] The system of this embodiment is not limited to text translation, but can also provide comprehensive multimodal feedback in combination with voice and pictures. Through voice recognition and image semantic analysis, the system can identify students' pronunciation errors and image translation errors during the translation process, and give targeted suggestions, thereby achieving all-round translation teaching.

[0055] This embodiment can also adjust the difficulty of teaching tasks according to students' historical learning data and real-time feedback, and dynamically optimize personalized learning paths. This embodiment can automatically generate personalized learning paths based on students' learning progress, translation performance, and error types, and adjust the difficulty of teaching tasks according to students' historical performance. The system will dynamically adjust translation tasks according to real-time feedback to help students gradually improve their translation skills.

[0056] The personalized translation teaching system described in this embodiment can also generate a visual translation progress report to help teachers and students understand the effect of translation learning and the direction of improvement in real time.

[0057] Retrieval-augmented generation optimization: The key to optimizing retrieval-augmented generation (RAG) in a personalized translation teaching system that combines multimodality with a large language model is to effectively integrate multimodal data such as text, voice, and images to improve the system's understanding of context and translation accuracy. During the optimization process, it is necessary to improve the retrieval module (such as efficient vector retrieval) and the generation module (such as personalized translation style), and perform personalized tuning based on the learner's language level and historical data. In addition, by introducing multiple rounds of dialogue, feedback mechanisms, and joint optimization strategies (such as reinforcement learning, cross-modal attention mechanisms, and transfer learning), the translation quality and user experience can be continuously improved. Ultimately, the system should ensure that the translation results meet the actual needs of users through continuous evaluation and adjustment, and realize personalized learning and intelligent teaching. The large language model continuously optimizes the system's translation capabilities and personalized feedback effects through few-sample learning and adaptive algorithms.

[0058] Through the above-mentioned embodiment, the present embodiment provides students with comprehensive translation feedback by integrating multimodal technology (including voice and image), and customizes personalized learning paths based on learners' historical data and language level, thereby significantly improving the efficiency and effect of translation teaching. At the same time, the system optimizes personalized translation style through RAG mechanism, using efficient vector retrieval technology and generation module to enhance students' cross-cultural understanding ability, and realizes continuous optimization and dynamic adjustment of translation model, ensuring the intelligence and personalization of translation teaching. The personalized translation teaching system of the present embodiment can make full use of large language model and multimodal technology, provide students with real-time, efficient and personalized translation learning experience, and greatly improve the quality and efficiency of translation teaching.

[0059] The above is only a preferred specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A personalized translation teaching system combining multimodality and large language model, characterized in that: include: Task acquisition module, used to obtain students' translation content; The post-editing evaluation module is used to analyze errors in the translated content based on large language models and deep learning algorithms, while providing personalized correction suggestions and optimizing the translation model for the machine translation system; The corpus update module is used to dynamically update the corpus based on student translation data and preset translation requirements; The translation intelligent feedback module is used to provide personalized feedback based on the translation content, and combines multimodal technology to provide real-time improvement suggestions for students' voice pronunciation and picture translation.

2. According to claim 1, a personalized translation teaching system combining multimodality and large language model is characterized in that: The post-editing evaluation module includes: The correction suggestion submodule is used to identify errors in grammar, semantics and pragmatic dimensions in the translated content, evaluate the identified errors according to preset evaluation criteria, and provide personalized correction suggestions, which include suggestions for improving content quality and suggestions for improving machine translation accuracy.

3. According to claim 2, a personalized translation teaching system combining multimodality and large language model is characterized in that: The post-editing evaluation module further includes: An iterative evaluation submodule, which is used to re-evaluate the revised translation content, wherein the revised translation content is the translation content adjusted by the student according to the personalized correction suggestions; The machine translation feedback submodule is used to process the post-editing data and provide feedback to the machine translation system to optimize its translation model and improve the quality of future translations.

4. According to claim 1, a personalized translation teaching system combining multimodality and large language model is characterized in that: The corpus updating module comprises: The dynamic update submodule is used to analyze the students' translation content, error patterns and the changing content data of the translation tasks, and dynamically update the corpus and term base based on the changing content data.

5. According to claim 4, a personalized translation teaching system combining multimodality and large language model is characterized in that: The corpus updating module further includes: The automatic supplement and optimization submodule is used to supplement and optimize the translation resources of the corpus according to the preset translation requirements of different languages ​​and preset fields.

6. The personalized translation teaching system combining multimodality and large language model according to claim 1, characterized in that: The translation intelligent feedback module includes: A real-time analysis submodule, used to perform real-time text analysis on the translated content and provide personalized feedback based on the text analysis results; The multimodal feedback submodule is used to detect the student's speech pronunciation through speech recognition technology, obtain pronunciation error detection results, and provide pronunciation correction suggestions based on the pronunciation error detection results; identify the pictures and corresponding translation problems in the student's translation based on the image semantic analysis technology, and provide improvement suggestions based on the picture translation problems.

7. The personalized translation teaching system combining multimodality and large language model according to claim 1, characterized in that: The personalized translation teaching system also includes: The learning path generation module is used to generate dynamic personalized learning paths based on students' learning progress, translation performance and error types.

8. The personalized translation teaching system combining multimodality and large language model according to claim 1, characterized in that: The personalized translation teaching system also includes: The feedback optimization module is used to improve the system's understanding of context and translation accuracy by integrating multimodal data, and introduces multi-round dialogues, feedback mechanisms, and joint optimization strategies to adjust and optimize the teaching system.

Images