Intelligent reading disorder assessment and training method and system based on large model

Abstract

The invention provides an intelligent reading disorder assessment and training method based on a large model. The method comprises the steps of obtaining user information and a test instruction of a to-be-tested user; generating a pretest text based on the user information and a test instruction; verifying the test text, adjusting a semantic problem in the test text, and generating a standard test text through iterative updating; acquiring reading voice data of the to-be-tested user through a voice recognition system; generating a reading evaluation result based on the reading voice data of the to-be-tested user; based on the method provided by the invention, the reading test text with matched capability can be automatically generated by utilizing the generation capability of a large model according to the age and reading level of the user, the reading evaluation result is fed back according to the real-time performance of the user, and the test text is updated. And the evaluation accuracy and effectiveness are improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence, and in particular to a large-model-based intelligent dyslexia assessment and training method and system. Background Art

[0002] Dyslexia is an important issue that affects children's learning and development. Especially in my country, the incidence of dyslexia among school-age children is about 5-8%. According to relevant research, the number of children with Chinese dyslexia in primary school is estimated to be between 5.35 million and 8.56 million. Dyslexia not only has a profound impact on children's academic performance, but may also damage their self-confidence and mental health. Therefore, it is crucial to identify dyslexia in a timely and accurate manner and provide appropriate support.

[0003] Current methods for detecting dyslexia mainly rely on traditional assessment tools and manual testing. These methods usually include manual testing and parent observation, but they have significant flaws: traditional assessments are highly subjective, and the results may be affected by differences in the experience and judgment of the assessors; the texts used for testing are usually manually selected, the number is limited, and the selection process is time-consuming and laborious, and there is a lack of unified standards, which affects the effectiveness and consistency of the assessment. In addition, existing methods are inefficient, consume a lot of manpower and time for assessment, and are difficult to fully adapt to children of different backgrounds and abilities, and personalized support is seriously insufficient. Summary of the invention

[0004] One of the embodiments of this specification provides a large model-based intelligent dyslexia assessment and training method. The dyslexia assessment and training method based on a multi-module agent includes:

[0005] Obtain user information and test instructions of the user to be tested;

[0006] Generate a pre-test text based on the user information and the test instruction;

[0007] Verifying the test text, adjusting semantic problems in the test text, and generating a standard test text through iterative updating;

[0008] Acquire reading voice data of the user to be tested through a voice recognition system;

[0009] generating a reading evaluation result based on the reading speech data of the user to be tested;

[0010] An enhanced test text is generated based on the reading assessment results.

[0011] In some embodiments, the test text includes a random phrase library and a reading text, wherein the random phrase library includes at least one group of random Chinese characters or phrases, and the reading text includes at least one section of text material covering grammatical structure or phonetic logic.

[0012] In some embodiments, generating the pre-test text based on the user information and the test instruction includes: matching the corresponding text recognition library based on the user information; extracting associated alternative text from the text recognition library according to the test instruction; and generating the pre-test text from the alternative text according to the subject of the test instruction.

[0013] In some embodiments, generating a reading evaluation result based on the user's reading voice data includes: extracting audio signal features of the reading voice data using an acoustic model; analyzing word sequences in the audio signal features through a large language model to convert the reading voice data into reading text data; and generating the reading evaluation result through an indicator evaluation function based on feature indicators in the reading text data.

[0014] One of the embodiments of this specification provides an intelligent dyslexia assessment and training system based on a large model, the system comprising:

[0015] The first acquisition module is used to obtain user information and test instructions of the user to be tested;

[0016] A text generation module, used to generate a pre-test text based on the user information and the test instruction;

[0017] An adjustment module, used to verify the test text, adjust semantic problems in the test text, and generate a standard test text through iterative updating;

[0018] The second acquisition module is used to acquire the reading voice data of the user to be tested through the voice recognition system;

[0019] An evaluation module, used for generating a reading evaluation result based on the reading speech data of the user to be tested;

[0020] A feedback module is used to generate a reinforced test text based on the reading assessment result.

[0021] One of the embodiments of the present specification provides an intelligent dyslexia assessment and training device based on a large model, including a processor, wherein the processor is used to execute the intelligent dyslexia assessment and training method based on a large model.

[0022] One of the embodiments of the present specification provides a computer-readable storage medium, wherein the storage medium stores computer instructions. When a computer reads the computer instructions in the storage medium, the computer executes the large model-based intelligent dyslexia assessment and training method. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] This specification will be further described in the form of exemplary embodiments, which will be described in detail by the accompanying drawings. These embodiments are not restrictive, and in these embodiments, the same number represents the same structure, wherein:

[0024] Figure 1 is an exemplary flow chart of an intelligent dyslexia assessment and training method according to some embodiments of this specification;

[0025] Figure 2 This is a display interface diagram of obtaining reading voice data of a user to be tested by a voice recognition system according to some embodiments of this specification;

[0026] Figure 3 is a display diagram of reading evaluation results according to some embodiments of this specification;

[0027] Figure 4 is a display diagram of another reading evaluation result according to some embodiments of this specification;

[0028] Figure 5 is an exemplary structural diagram of an intelligent dyslexia assessment and training system according to some embodiments of this specification;

[0029] Figure 6 It is a framework diagram of an intelligent dyslexia detection system according to some embodiments of this specification. DETAILED DESCRIPTION

[0030] In order to more clearly illustrate the technical solutions of the embodiments of this specification, the following is a brief introduction to the drawings required for the description of the embodiments. Obviously, the drawings described below are only some examples or embodiments of this specification. For ordinary technicians in this field, this specification can also be applied to other similar scenarios based on these drawings without creative work. Unless it is obvious from the language environment or otherwise explained, the same reference numerals in the figures represent the same structure or operation.

[0031] It should be understood that the "system", "device", "unit" and / or "module" used herein are a method for distinguishing different components, elements, parts, portions or assemblies at different levels. However, if other words can achieve the same purpose, the words can be replaced by other expressions.

[0032] As shown in this specification and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "comprise" and "include" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0033] Flowcharts are used in this specification to illustrate the operations performed by the system according to the embodiments of this specification. It should be understood that the preceding or following operations are not necessarily performed precisely in order. Instead, the steps may be processed in reverse order or simultaneously. At the same time, other operations may also be added to these processes, or one or more operations may be removed from these processes.

[0034] Figure 1 is an exemplary flow chart of a large model-based intelligent dyslexia assessment and training method according to some embodiments of this specification. Figure 1 As shown, the process includes the following steps.

[0035] S110, obtaining user information and test instructions of the user to be tested.

[0036] In some embodiments, user information refers to the personal identity information of the user to be tested, which may include the age, identity, etc. of the user to be tested; the test instructions may include difficulty instructions and prompt words, etc. The user can input difficulty instructions (for example: simple, medium, difficult) and prompt words (for example: friendship; sports; technology; art, etc.) to allow the system to generate a test text that matches the needs and realize personalized reading text evaluation.

[0037] S120, generating a pre-test text based on the user information and the test instruction.

[0038] In some embodiments, the test text may include a random phrase library and a reading text, wherein the random phrase library includes at least one group of random Chinese characters or phrases, and the reading text includes at least one paragraph of text material covering grammatical structure or phonetic logic.

[0039] For example, the system can generate a random phrase library of 150 words or phrases based on the information input by the user; and generate suitable reading text materials based on the user's cognitive ability. These materials cover different levels of grammatical structure and logical expression to ensure that they can adapt to the user's level.

[0040] In some embodiments, the process of generating the pre-test text based on the user information and the test instruction may include:

[0041] Matching a corresponding text recognition library based on the user information;

[0042] extracting associated alternate text from the text recognition library according to the test instruction;

[0043] The pre-test text is generated from the standby text according to the subject of the test instruction.

[0044] In some embodiments, the system can build a "text recognition library" that matches the user's cognitive ability based on text recognition standards for different age groups (such as international Chinese education standards) to help generate text that meets the user's reading ability. At the same time, a rule library covering semantic logic such as causality can also be introduced to ensure that the generated text has logic that meets the user's cognitive level.

[0045] In some embodiments, the associated backup text can be extracted from the text cognition library according to the test instruction through a retrieval augmented generation model (RAG). After receiving the user input, the RAG model first processes the user input (including user information and test instructions), extracts key information from the data, and converts the text data into a vector matrix. Furthermore, through the retrieval module, through an efficient retrieval method, the most relevant knowledge to the question is recalled, and relevant information is extracted from a large knowledge base (usually a pre-indexed document database or an external information source); then the prompt is integrated, the system refers to the current question and related knowledge (text cognition library), and the model reads from the retrieval results and performs further text generation based on these retrieved documents.

[0046] In some embodiments, this embodiment may also include reducing the processing of non-essential information in the backup text through sparse attention and sharding mechanisms, and only processing highly relevant content. Sparse attention is an optimized attention mechanism that can map a query vector and a set of key-value pairs to an output vector, and reduce the amount of calculation and memory consumption by calculating the similarity between the query vector and some key vectors. Among them, non-essential information refers to text content in the test text that is weakly related to the topic or has semantic and logical errors.

[0047] In some embodiments, the text recognition library can perform dynamic knowledge call. RAG generates more accurate results by dynamically calling the information in the knowledge base and updating the information in the knowledge base in real time. It retrieves documents related to the user input (for example, the prompt words are: friendship, technology, etc.) and provides background support for the large language model LLM.

[0048] S130, verifying the test text, adjusting semantic problems in the test text, and generating a standard test text through iterative updating.

[0049] In some embodiments, the system can evaluate the text generated by LLM based on the logic rule library and the text cognition library to check whether the text meets the user's reading level and whether there are semantic and logical errors. By matching the generated text with the word vector, if the generated result is too different from the cognition library, or there are problems in the logic rule library, it will be regenerated. The generated text is generated as a whole paragraph. If there is a problem with the text, the text is sent back and then a complete text is generated.

[0050] In some embodiments, if the text fails to meet the user's requirements, the system will feed it back to the large language model to regenerate the text until it reaches the appropriate standard.

[0051] S140, obtaining reading voice data of the user to be tested through a voice recognition system.

[0052] In some embodiments, the reading voice data of the user to be tested can be obtained through the FunASR speech recognition system, which uses the latest deep learning technology, including Transformer and Conformer architectures (these architectures have significant advantages when processing long time series data), to identify different features in the speech signal through training models and map them to the corresponding text. The FunASR speech recognition system is an open source automatic speech recognition (ASR) system developed by Alibaba DAMO Academy. It aims to provide an easy-to-use and efficient platform to help developers and researchers quickly build their own speech recognition applications. The project combines deep learning technology with large-scale data processing capabilities to achieve high-precision real-time speech-to-text functions. Figure 2 The figure shows the display interface of the speech recognition system acquiring the reading speech data of the user to be tested.

[0053] S150, generating a reading evaluation result based on the reading speech data of the user to be tested.

[0054] In some embodiments, generating a reading assessment result based on the user's reading speech data includes:

[0055] Extracting audio signal features of the reading speech data using an acoustic model;

[0056] Analyzing the word sequence in the audio signal features by a large language model, and converting the reading voice data into reading text data;

[0057] The reading evaluation result is generated by an indicator evaluation function according to the characteristic indicators in the reading text data.

[0058] In some embodiments, the FunASR speech recognition system includes three key models to process the user's speech, including: acoustic model: responsible for extracting features from audio signals; language model: predicting word sequences in language to improve recognition accuracy; end-to-end model: directly converting from audio to text, reducing intermediate steps and improving efficiency.

[0059] In some embodiments, the reading evaluation result of the user to be tested may include characteristic indicators such as the user's reading time, number of words read, total number of words in the article, number of words added or reduced, number of errors, number of missing words, number of insertions, number of sequence permutations, number of substitutions, number of serializations, etc., and obtain data statistics of the characteristic indicators.

[0060] In some embodiments, when the user is reading the test text, the system can analyze the user's pronunciation in real time, generate evaluation indicators based on the user's reading data and the results of the large language model analysis, and then automatically generate a report containing wrong characters and missing characters through internal functions based on these indicators. The report will classify the types of wrong characters in detail, such as phono-semantic errors, similar characters errors, and phonetic errors, to help analyze the user's reading problems.

[0061] like Figure 3 and Figure 4 The reading evaluation result display report shown in the figure shows that when the user reads the test text, the system will analyze the user's pronunciation in real time, generate evaluation indicators based on the user's reading data and the results of the large language model (LLM) analysis, and then automatically generate a report containing wrong characters and missing characters through the indicator evaluation function. The report will classify the wrong character types in detail, such as phono-semantic character errors, similar character errors, and phonetic errors, to help analyze the user's reading problems.

[0062] In some embodiments, the indicator evaluation function is used to calculate the evaluation indicators in the form of data through a table or text to generate an easy-to-read document. Optionally, the reading accuracy, the total number of words correctly read per minute (WCPM for short), and the number of words per minute (WPM for short) are calculated again through the obtained results as evaluation indicators of reading accuracy and reading fluency. In terms of the standards for considering reading fluency, there are many different reading measurement tables, such as: DIBELS (Dynamic Cognitive Measurement Table), DIBELS Next (Dynamic Cognitive Upgrade Table), CBM Measurement Table, ORF Measurement Table, etc. Among them, the ORF measurement table is the most widely used one. During the ORF measurement, students learning English need to read aloud a passage that has never been practiced, and the time is within 1 minute. The examiner of the test writes down the mistakes made in reading, including the number of words read incorrectly or pronounced incorrectly; which words are omitted; words read in the wrong order; or words pronounced by the examiner for students after pausing for 3 seconds. Then calculate the total number of words read correctly per minute (WCPM for short).

[0063] Specifically, the calculation of the reading accuracy is as follows:

[0064] Correct reading rate = (total number of words in the article - total number of errors) / total number of words in the article;

[0065] The calculation of WCPM is as follows:

[0066] WCPM = (total number of words in the article – total number of errors) / reading time;

[0067] The calculation of WPM is as follows:

[0068] WPM = total number of words read in the article / reading time).

[0069] In some embodiments, generating the reading evaluation result by the indicator evaluation function may include the following process:

[0070] Register fonts: If you need to support Chinese display, you can register Chinese fonts.

[0071] Title and content: Write different evaluation items into the report in a logical order according to the formatted_results passed in.

[0072] Save PDF: Save the generated assessment content as a PDF file for users to download or view. Automatically generate a report containing wrong characters and missing characters. The report will classify the types of wrong characters in detail, such as phono-semantic errors, similar characters errors, and phonetic errors, etc., to help analyze the user's reading problems. Users only need to click on the report to download the report to their user end.

[0073] Specifically, the results of the generated test report are as follows:

[0074] Total word count of reference text: 400 words;

[0075] Number of words read aloud: 389 words;

[0076] Total number of errors: 23;

[0077] Reading time: 3.08 minutes;

[0078] After getting the test results, the following evaluation indicators can be calculated:

[0079] Reading accuracy = (400–23) / 400 = 94.25%;

[0080] WCPM=(400–23) / 3.08=122.40;

[0081] WPM=389 / 3.08=126.30.

[0082] S160, generating a reinforced test text based on the reading assessment result.

[0083] In some embodiments, the system can analyze the user's weaknesses based on the reading assessment results, where the user's weaknesses can be the parts of the specific types of errors that the user to be tested makes more frequently. For example, the system can analyze the number of typos, grammatical structure errors, number of missing words and other indicators in the reading assessment results, extract relevant information from the text knowledge base based on the user's weaknesses, and further generate reinforced text to help the user conduct targeted training. For example, if the user to be tested reads "therefore" as "because" many times, the system will collect and analyze these high-frequency erroneous words or grammatical structures, generate text containing these words or grammatical structures with the help of the logic rule base and the text cognition base, and continuously iterate through feedback to finally generate targeted reinforced text.

[0084] In some embodiments, the system can automatically iterate and update the text library. With the continuous testing and feedback from users, the system will continuously update the knowledge base and logic rule base, and optimize the generated text to ensure that it can gradually improve the reading ability of users. At the same time, it will continue to verify each generated text to ensure that the content generated by the system can effectively help users overcome reading difficulties.

[0085] Through the above process, the system can dynamically adapt to the user's reading needs, evaluate and solve reading difficulties in real time. Through the mutual transmission and complementation of information among various modules in the system, it can not only improve the accuracy of user reading ability assessment, but also realize self-enhancement and personalized customization of text generation through feedback mechanism, thus effectively helping users overcome reading difficulties.

[0086] It should be noted that the above description of the relevant process is only for example and explanation, and does not limit the scope of application of this specification. For those skilled in the art, various modifications and changes can be made to the process under the guidance of this specification. However, these modifications and changes are still within the scope of this specification.

[0087] Figure 5 is an exemplary structural diagram of an intelligent dyslexia assessment and training system based on a large model according to some embodiments of this specification. Figure 5 As shown, the system 500 includes:

[0088] The first acquisition module 510 is used to acquire user information and test instructions of the user to be tested;

[0089] A text generation module 520, configured to generate a pre-test text based on the user information and the test instruction;

[0090] An adjustment module 530 is used to verify the test text, adjust semantic problems in the test text, and generate a standard test text through iterative updating;

[0091] The second acquisition module 540 is used to acquire the reading voice data of the user to be tested through the voice recognition system;

[0092] An evaluation module 550, configured to generate a reading evaluation result based on the user's reading speech data;

[0093] The feedback module 560 is used to generate a reinforced test text based on the reading assessment result.

[0094] Furthermore, the test text includes a random phrase library and a reading text, wherein the random phrase library includes at least one group of random Chinese characters or phrases, and the reading text includes at least one section of text material covering grammatical structure or phonetic logic.

[0095] The text generation module 520 can also be used to match the corresponding text recognition library based on the user information; extract associated backup text from the text recognition library according to the test instruction; filter out non-essential information in the backup text through sparse attention and segmentation mechanism; and generate the pre-test text according to the subject of the test instruction.

[0096] The evaluation module 550 can also be used to: extract the audio signal features of the reading voice data using an acoustic model; analyze the word sequence in the audio signal features through a large language model to convert the reading voice data into reading text data; and generate the reading evaluation result through an indicator evaluation function based on the feature indicators in the reading text data.

[0097] It should be understood that Figure 5 The illustrated system and its modules may be implemented in various ways.

[0098] It should be noted that the above description of the candidate item display and determination system and its modules is only for convenience of description and does not limit the present specification to the scope of the embodiments. It is understandable that, after understanding the principle of the system, those skilled in the art may arbitrarily combine the modules or form a subsystem to connect with other modules without deviating from the principle. In some embodiments, Figure 5 The modules disclosed in the specification may be different modules in a system, or one module may realize the functions of two or more modules. For example, the modules may share a storage module, or each module may have its own storage module. Such variations are within the scope of protection of this specification.

[0099] Figure 6It is a framework diagram of an intelligent dyslexia detection system according to some embodiments of this specification.

[0100] In some embodiments, the dyslexia assessment and training system can be an intelligent dyslexia detection system with a large language model (LLM) as the core, which consists of four modules, including a memory module, a planning module, a tool module, and an action module. Figure 6 As shown in the figure, the framework diagram of the intelligent dyslexia detection system with the large language model (LLM) as the core is shown. Among them, each module in the intelligent dyslexia detection system has the following functions:

[0101] 1) Memory module

[0102] This module introduces RAG (Retrieval Enhanced Generation) technology, which can accept user input and retrieve a series of relevant documents, with the source attached, to provide rich context for the large language model. At the same time, a logical rule library (covering semantic logic understood by different age groups, such as causality, etc.) and a student text recognition library (Chinese character recognition ability formulated according to international Chinese education standards) are constructed to generate language test texts that meet the user's cognitive ability in real time.

[0103] 2) Planning Module

[0104] This module mainly plays a verification function. After LLM generates the text, the verification module uses the logic rule library built by the memory module and the student text recognition library to identify whether the text meets the user's requirements and user cognitive ability. If it does not meet the requirements, it will be fed back to the large model for regeneration until it meets the requirements.

[0105] 3) Tool module

[0106] This module is further divided into real-time offline speech recognition and timely evaluation system for dyslexia (tool 1) and discovery of internal causes of user impairments and targeted training (tool 2).

[0107] For tool1, the real-time offline speech recognition and dyslexia assessment system, supported by FunASR software, generates visual reports for 150 random words and test texts, including statistics on wrong and missed characters and visualization of error types (such as phono-semantic errors, similar-shaped character errors, phonetic errors, and semantic errors). For tool2, based on the reading assessment report and the knowledge base in the Memory module, targeted training is performed on the user's weak logic or words by enhancing the generation of texts containing multiple related logic or words.

[0108] 4) Decision (Action) module

[0109] This module can coordinate the results of the above three modules and generate the final decision.

[0110] The basic concepts have been described above. Obviously, for those skilled in the art, the above detailed disclosure is only for example and does not constitute a limitation of this specification. Although not explicitly stated here, those skilled in the art may make various modifications, improvements and corrections to this specification. Such modifications, improvements and corrections are suggested in this specification, so such modifications, improvements and corrections still belong to the spirit and scope of the exemplary embodiments of this specification.

[0111] At the same time, this specification uses specific words to describe the embodiments of this specification. For example, "one embodiment", "an embodiment", and / or "some embodiments" refer to a certain feature, structure or characteristic related to at least one embodiment of this specification. Therefore, it should be emphasized and noted that "one embodiment" or "an embodiment" or "an alternative embodiment" mentioned twice or more in different positions in this specification does not necessarily refer to the same embodiment. In addition, certain features, structures or characteristics in one or more embodiments of this specification can be appropriately combined.

[0112] In addition, unless explicitly stated in the claims, the order of the processing elements and sequences described in this specification, the use of alphanumeric characters, or the use of other names are not intended to limit the order of the processes and methods of this specification. Although the above disclosure discusses some invention embodiments that are currently considered useful through various examples, it should be understood that such details are only for illustrative purposes, and the attached claims are not limited to the disclosed embodiments. On the contrary, the claims are intended to cover all modifications and equivalent combinations that are consistent with the essence and scope of the embodiments of this specification. For example, although the system components described above can be implemented by hardware devices, they can also be implemented only by software solutions, such as installing the described system on an existing server or mobile device.

[0113] Similarly, it should be noted that in order to simplify the description disclosed in this specification and thus help understand one or more embodiments of the invention, in the above description of the embodiments of this specification, multiple features are sometimes combined into one embodiment, figure or description thereof. However, this disclosure method does not mean that the features required by the subject matter of this specification are more than the features mentioned in the claims. In fact, the features of the embodiments are less than all the features of the single embodiment disclosed above.

[0114] In some embodiments, numbers describing the number of components and attributes are used. It should be understood that such numbers used in the description of the embodiments are modified by the modifiers "about", "approximately" or "substantially" in some examples. Unless otherwise specified, "about", "approximately" or "substantially" indicate that the numbers are allowed to vary by ±20%. Accordingly, in some embodiments, the numerical parameters used in the specification and claims are approximate values, which may change according to the required features of individual embodiments. In some embodiments, the numerical parameters should take into account the specified significant digits and adopt the general method of retaining digits. Although the numerical domains and parameters used to confirm the breadth of their range in some embodiments of this specification are approximate values, in specific embodiments, the setting of such numerical values ​​is as accurate as possible within the feasible range.

[0115] Each patent, patent application, patent application publication, and other materials, such as articles, books, specifications, publications, documents, etc., cited in this specification are hereby incorporated by reference in their entirety. Except for application history documents that are inconsistent with or conflicting with the contents of this specification, documents that limit the broadest scope of the claims of this specification (currently or later attached to this specification) are also excluded. It should be noted that if the descriptions, definitions, and / or use of terms in the materials attached to this specification are inconsistent or conflicting with the contents described in this specification, the descriptions, definitions, and / or use of terms in this specification shall prevail.

[0116] Finally, it should be understood that the embodiments described in this specification are only used to illustrate the principles of the embodiments of this specification. Other variations may also fall within the scope of this specification. Therefore, as an example and not a limitation, alternative configurations of the embodiments of this specification may be considered consistent with the teachings of this specification. Accordingly, the embodiments of this specification are not limited to the embodiments explicitly introduced and described in this specification.

Claims

1. A large-model-based intelligent dyslexia assessment and training method, characterized in that: include: Obtain user information and test instructions of the user to be tested; Generate a pre-test text based on the user information and the test instruction; Verifying the test text, adjusting semantic problems in the test text, and generating a standard test text through iterative updating; Acquire reading voice data of the user to be tested through a voice recognition system; generating a reading evaluation result based on the reading speech data of the user to be tested; An enhanced test text is generated based on the reading assessment results.

2. The large model-based intelligent dyslexia assessment and training method according to claim 1, characterized in that: The test text includes a random phrase library and a reading text, wherein the random phrase library includes at least one group of random Chinese characters or phrases, and the reading text includes at least one section of text material covering grammatical structure or phonetic logic.

3. The large model-based intelligent dyslexia assessment and training method according to claim 1, characterized in that: Generating the pre-test text based on the user information and the test instruction includes: Matching a corresponding text recognition library based on the user information; extracting associated alternate text from the text recognition library according to the test instruction; The pre-test text is generated from the standby text according to the subject of the test instruction.

4. The large model-based intelligent dyslexia assessment and training method according to claim 1, characterized in that: Generating a reading evaluation result based on the reading speech data of the user to be tested includes: Extracting audio signal features of the reading speech data using an acoustic model; Analyzing the word sequence in the audio signal features by a large language model, and converting the reading voice data into reading text data; The reading evaluation result is generated by an indicator evaluation function according to the characteristic indicators in the reading text data.

5. An intelligent dyslexia assessment and training system based on a large model, characterized in that: include: The first acquisition module is used to obtain user information and test instructions of the user to be tested; A text generation module, used to generate a pre-test text based on the user information and the test instruction; An adjustment module, used to verify the test text, adjust semantic problems in the test text, and generate a standard test text through iterative updating; The second acquisition module is used to acquire the reading voice data of the user to be tested through the voice recognition system; An evaluation module, used for generating a reading evaluation result based on the reading speech data of the user to be tested; A feedback module is used to generate a reinforced test text based on the reading assessment result.

6. An intelligent dyslexia assessment and training device based on a large model, comprising a processor, wherein the processor is used to execute the intelligent dyslexia assessment and training method based on a large model according to any one of claims 1 to 4.

7. A computer-readable storage medium, wherein the storage medium stores computer instructions. When a computer reads the computer instructions in the storage medium, the computer executes the large model-based intelligent dyslexia assessment and training method as described in any one of claims 1 to 4.

Images