Big data-based special education teaching resource recommendation system

Abstract

The invention relates to the technical field of intelligent education, in particular to a special education teaching resource recommendation system based on big data, and the system comprises a data collection and classification module which collects the personal data of special education students through a big data platform, obtains the disability type of each special education student from the personal data, and stores the disability type of each special education student; and the disability types are classified to generate special student classification data. According to the invention, through comprehensive analysis and collection of personal data of special education students, the disability type and learning requirements of each student are accurately identified, and the personalized requirements of special education are embodied in the education resource recommendation process. According to the scheme, classification is carried out according to the disability types of the students, the learning behaviors, cognitive competence and learning progress of the students are combined, the matching degree of teaching resources is optimized through deep learning and data analysis, and it is ensured that most suitable learning materials are provided for each student. According to the refined recommendation mechanism, special requirements of students are considered.

Description

Technical Field

[0001] The present invention relates to the field of intelligent education technology, and in particular to a special education teaching resource recommendation system based on big data. Background Art

[0002] The field of intelligent education technology includes various intelligent technologies applied to education, mainly realizing the optimal allocation of educational resources and personalized services through information technology, artificial intelligence, big data, cloud computing, etc. Its core content includes intelligent recommendation of educational content, design of personalized learning paths, intelligent evaluation and feedback, and analysis and mining of learning data.

[0003] Among them, the special education teaching resource recommendation system refers to the use of big data technology and intelligent algorithms to provide personalized teaching resource recommendation services for students in the field of special education. The system mainly targets the learning needs of special education students, uses data analysis and user behavior data, and matches students with appropriate teaching resources through precise recommendations. The system provides customized teaching content based on students' characteristic data, such as learning progress, cognitive ability, etc., to help students get the most suitable support and guidance in the learning process.

[0004] The existing technology has the problem of insufficient personalization in the process of recommending special education resources. Although the existing system can recommend resources based on the type of students' disabilities, it is often difficult to fully consider the students' real-time learning needs and progress, especially when dealing with students' individual differences and rapidly changing learning states. The adaptability of the recommendation results is poor. Many systems fail to fully incorporate interdisciplinary needs when recommending resources, resulting in limitations in the recommended content and the inability to effectively achieve integration and collaboration between disciplines. This single-subject recommendation method cannot meet the diversified learning needs of special education students, especially for students with interdisciplinary learning goals. In addition, the existing technology is relatively weak in predicting learning progress and cannot dynamically adjust the recommendation of teaching resources in real time, resulting in students encountering problems of maladaptation at certain stages of learning, affecting learning outcomes. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a special education teaching resource recommendation system based on big data.

[0006] In order to achieve the above-mentioned purpose, the present invention adopts the following technical scheme: A special education teaching resource recommendation system based on big data comprises: The data collection and classification module collects the personal data of special education students through the big data platform, obtains the disability type of each special education student from the personal data, and classifies the disability type to generate special student classification data; The teaching resource matching module collects the learning needs of each special education student in the special student classification data, calculates the matching degree between the teaching resources and the student's learning needs, matches the learning need information with the resource content in the teaching resource library according to the matching degree, gives priority to recommending teaching materials or courses that match the student's current learning needs, and generates a teaching resource recommendation list; The interdisciplinary resource recommendation module optimizes the teaching resource recommendation list in combination with the students' interdisciplinary learning goals, calculates the relevance of multiple subject resources, recommends interdisciplinary resources based on the relevance information, and generates an interdisciplinary teaching resource recommendation list; The learning progress calculation module collects the learning progress data of students according to the interdisciplinary teaching resource recommendation list, uses the long short-term memory network model to learn the students' entire learning trajectory, calculates the students' future learning status based on the learning trajectory, obtains learning progress prediction data, adjusts future learning resource recommendations based on the learning progress prediction data, and obtains the optimization result of the teaching resource recommendation list.

[0007] As a further solution of the present invention, the steps for obtaining the special student classification data are specifically as follows: Based on the big data platform, personal data of special education students are collected, including basic information and disability type data of students. Disability type of each student is obtained through school registration and assessment data. Disability types include visual impairment, hearing impairment, cognitive impairment, and student disability type data is generated; The data of each student in the student disability type data is classified according to the disability type, and stored in multiple categories respectively to generate special student classification data.

[0008] As a further solution of the present invention, the step of calculating the matching degree between teaching resources and student learning needs is specifically: Collect personal data of special education students, including basic information and learning behaviors, analyze the type of disability of each student, and determine the corresponding learning needs based on the student's behavioral data to generate student learning needs data; Based on the student learning needs data, the formula is adopted: ; Calculate the similarity between the teaching resource vector and the student demand vector ; in, and The first Features, is the characteristic vector of teaching resources, is the student's learning demand vector, is the dimension of the feature.

[0009] As a further solution of the present invention, the steps of obtaining the teaching resource recommendation list are specifically as follows: According to the similarity between the teaching resource vector and the student demand vector, the learning demand information is matched with the resource content in the teaching resource library to obtain a teaching resource matching result; Based on the teaching resource matching results, the teaching resource library screens the teaching materials or courses that match the student's needs, and sorts them from high to low according to the matching degree to generate a teaching resource recommendation list.

[0010] As a further solution of the present invention, the step of calculating the relevance of multiple subject resources is specifically: Collect the learning needs and goals of the corresponding subjects from students' learning data and performance records in multiple subjects, identify the key needs of each subject by comparing students' subject preferences, analyze students' interdisciplinary learning goals based on the needs, and obtain the demand correlation data between subjects; Based on the demand correlation data between the disciplines, the formula is adopted: ; Computing and disciplines Resource associations between , get the related information of multiple subject resources; in, and Representing two disciplines, Indicates the first Features, Represents all features, and Respectively represent subjects and disciplines Resources eigenvalues, and Indicated in the subject and disciplines In Features, is the number of feature dimensions.

[0011] As a further solution of the present invention, the steps of obtaining the recommended list of interdisciplinary teaching resources are specifically as follows: Based on the association information of the multiple subject resources, analyzing the resource relationship between each subject, comparing the similarity of the resource relationship in multiple dimensions, identifying and screening matching resources, and generating screened resource information; Based on the filtered resource information, interdisciplinary resources are recommended in order of relevance, and the recommended content of the teaching resource recommendation list is optimized according to the needs of students to generate an interdisciplinary teaching resource recommendation list.

[0012] As a further solution of the present invention, the step of obtaining the learning progress prediction data is specifically as follows: The learning progress data of the students according to the recommended list of interdisciplinary teaching resources, including learning time, task completion and correct answer rate, are analyzed to identify the students' learning trajectory, and the long short-term memory network model is used to study it to obtain the students' current learning status information; Based on the student's current learning status information, the formula is adopted: ; Calculate students' future moments Learning status , get the predicted result of students’ learning progress; in, Indicates that the student is at the current moment The learning status Indicates that the student is at the current moment The expected learning state or target learning state, Represents the learning rate coefficient.

[0013] As a further solution of the present invention, the steps for obtaining the optimization result of the teaching resource recommendation list are specifically as follows: Based on the predicted learning progress result of the student, the range of the student's learning status is analyzed, and the recommended learning resource type is determined according to the selected range, and the learning task is recommended to obtain the future learning resource recommendation result; Based on the future learning resource recommendation results, an optimization result of the teaching resource recommendation list is generated by adding tasks and materials that match the student's current learning status, such as advanced exercises and interactive learning resources.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, by comprehensively analyzing and collecting the personal data of special education students, the disability type and learning needs of each student are accurately identified, and the personalized needs of special education are reflected in the process of recommending educational resources. The program not only classifies students according to their disability type, but also combines students' learning behaviors, cognitive abilities and learning progress, optimizes the matching degree of teaching resources through deep learning and data analysis, and ensures that the most suitable learning materials are provided for each student. This refined recommendation mechanism not only takes into account the special needs of students, but also helps students develop comprehensively through interdisciplinary resource recommendations. Especially in terms of learning progress prediction, the program accurately calculates students' future learning status through dynamic analysis of students' learning trajectories, and then adjusts resource recommendation strategies to enhance the personalized service capabilities of educational resources. Through these innovative measures, it can ensure that special education students receive more customized support and guidance during the learning process, overcoming the limitations of traditional educational resource recommendations. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 is a system flow chart of the present invention; Figure 2 A flow chart for obtaining special student classification data for the present invention; Figure 3 A flowchart for calculating the matching degree between teaching resources and student learning needs in the present invention; Figure 4 A flowchart for obtaining a recommended list of teaching resources for the present invention; Figure 5 A flow chart for calculating the relevance of multiple subject resources for the present invention; Figure 6 A flowchart for obtaining a recommended list of interdisciplinary teaching resources for the present invention; Figure 7 A flowchart of obtaining learning progress prediction data for the present invention; Figure 8 A flowchart of the present invention for obtaining optimization results of the teaching resource recommendation list. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0017] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0018] See also Figure 1 ,A special education teaching resource recommendation system based on big data includes: The data collection and classification module collects the personal data of special education students through the big data platform, obtains the disability type of each special education student from the personal data, and classifies the disability type to generate special student classification data; The teaching resource matching module collects the learning needs of each special education student in the special student classification data, calculates the matching degree between the teaching resources and the student's learning needs, matches the learning needs information with the resource content in the teaching resource library according to the matching degree, gives priority to recommending textbooks or courses that match the student's current learning needs, and generates a teaching resource recommendation list; The interdisciplinary resource recommendation module optimizes the recommended list of teaching resources in combination with students' interdisciplinary learning goals. It calculates the relevance of multiple subject resources, recommends interdisciplinary resources based on the relevance information, and generates a recommended list of interdisciplinary teaching resources. The learning progress estimation module collects the learning progress data of students according to the recommended list of interdisciplinary teaching resources, uses the long short-term memory network model to learn the entire learning trajectory of the students, estimates the future learning status of the students according to the learning trajectory, obtains the learning progress prediction data, adjusts the future learning resource recommendations according to the learning progress prediction data, and obtains the optimization result of the teaching resource recommendation list; The classification data for special students include students' basic information, disability type, learning needs, and historical learning data; the recommended list of teaching resources includes audio teaching materials, video courses, interactive exercises, and graphical learning resources; the recommended list of interdisciplinary teaching resources includes linguistics teaching materials, mathematics teaching materials, science teaching materials, and cognitive development resources; the learning progress prediction data includes the completion of future learning tasks, knowledge mastery progress, identification of learning difficulties, and learning trend predictions.

[0019] See also Figure 2 ,The specific steps for obtaining special student classification data are: Based on the big data platform, personal data of special education students are collected, including basic information and disability type data of students. Disability type of each student is obtained through school registration and assessment data. Disability types include visual impairment, hearing impairment, cognitive impairment, and student disability type data is generated; First, the system obtains basic information about each student from data sources such as schools and educational institutions, including student identity information (such as name, age, gender, class, etc.). Next, based on the school's registration and assessment data, the system identifies the type of disability of each student, which mainly includes visual impairment, hearing impairment, cognitive impairment, etc. By integrating the school management platform with student registration information, the system can accurately obtain the type of disability of each student. Students with visual impairment can be marked by the type of visual impairment provided in the registration form (such as myopia, hyperopia, color blindness, etc.), students with hearing impairment are recorded according to the classification of hearing test results, and students with cognitive impairment are marked according to the type of impairment shown in the cognitive test (such as difficulty understanding, memory problems, etc.). All of this data is sorted through system automated processing and preprocessing steps (such as missing data filling, duplicate data deletion, etc.) to ensure data integrity and consistency. Finally, ensure that all data is accurately marked according to disability type and generate complete disability type data.

[0020] Classify the data of each student according to the disability type in the student disability type data, store them in multiple categories, and generate special student classification data; The data of students are classified and stored according to their disability type. For example, all students with visual impairments will be separated into a data table or classification area, all students with hearing impairments will be stored in another table or classification area, and students with cognitive impairments will be classified into a third category. For each category, the system will further subdivide based on the students' personal data, and can be further classified based on factors such as gender, age group, learning behavior, etc., so as to facilitate the subsequent personalized teaching resource recommendation. For example, students with visual impairments can be further subdivided according to the degree of their impairment (such as complete blindness or partial visual impairment), and the number of students in each subdivided category is recorded. The classification process can be implemented through SQL queries to ensure that the detailed data of each student can be efficiently and accurately stored and managed according to their disability type. Finally, the system will generate special student classification data based on the classification results.

[0021] See also Figure 3 ,The steps to calculate the matching degree between teaching resources and students’ learning needs are as follows: Collect personal data of special education students, including basic information and learning behaviors, analyze the type of disability of each student, and determine the corresponding learning needs based on the student's behavioral data to generate student learning needs data; First, the learning activities of special education students are monitored through the big data platform, including their performance in class, interaction with teaching materials, learning time, learning methods and other information, and analyzed in combination with the students' disability types. For students with visual impairments, the system will give priority to collecting their learning needs for audio teaching materials, voice courseware and other forms of resources, collecting their response to visual information, and their preference for auditory resources; for students with hearing impairments, the main focus is on their needs for teaching resources such as subtitled videos and sign language videos; and for students with cognitive impairments, the system will collect their demand data for highly interactive learning resources, such as gamified learning materials, illustrated learning tasks, etc. When collecting learning needs, the system analyzes the relationship between students' behavioral data (such as homework submission, answer accuracy, learning time, etc.) and disability types, so as to derive the intensity of each student's needs for different learning resources. For example, when the system analyzes that a visually impaired student has used audio courseware for more than 50 hours in the past three months, and has used visual materials for less than 10 hours, the system can judge that the student has a high demand for audio teaching materials, and vice versa. Similarly, for students with hearing impairments, if their study time on subtitled videos accounts for more than 60% of their total study time over a period of time, the system determines that they have a strong demand for video teaching resources. For students with cognitive impairments, the system determines the intensity of their demand by analyzing their participation in interactive tasks. For example, if a student with cognitive impairment has spent more than 30 hours on illustrated learning tasks in the past month, the system will assess the student's needs as "high demand." On the contrary, if the student's participation in such tasks is less than 10 hours, it will be assessed as "low demand." At the same time, the system will also comprehensively consider the student's accuracy in answering questions when completing tasks. If the student shows a high accuracy rate when answering questions, it means that he or she has a certain adaptability to existing learning resources, and the system can further adjust the recommendation of teaching resources based on this information.

[0022] Based on student learning needs data, the formula is used: ; Calculate the similarity between the teaching resource vector and the student demand vector , The value range is , a value close to 1 indicates that the two are very similar, a value close to 0 indicates that the two are not similar, and a value close to -1 indicates that the two are completely opposite; in, and The first features. In this formula, is the characteristic vector of teaching resources, is the student's learning needs vector. Each element of the vector and Corresponding to different features, such as text content, video format, interactivity, etc. By extracting the features of teaching resources and student needs, each feature is converted into a numerical value for representation. is the dimension of the feature, i.e., vector and The number of elements in . It represents the number of characteristics of teaching resources and student needs. For example, the characteristics may include: teaching material type, interactivity, video length, etc. The specific number of dimensions is determined by the content of teaching resources and the diversity of student needs. is the dot product of the teaching resources and student demand vectors, indicating their matching degree in each feature dimension. Specifically, each pair of feature values and The product reflects the degree of fit between resources and needs in this feature, and the sum of the products of all feature dimensions gives the total matching degree. and are the modulus (i.e., the size of the vector) of the teaching resource vector and the student demand vector, respectively. The modulus is calculated by summing the squares of each element of the vector and taking the square root. The modulus represents the total "size" of the vector. It is a normalization factor when calculating cosine similarity to prevent vectors of different lengths from affecting the similarity results.

[0023] First, features need to be extracted from teaching resources and student needs. For example, the features of teaching resources may include teaching material type (audio, video, text), content depth, interactivity, etc.; the features of student needs may include required video subtitles, duration of audio content, interactivity, etc. The acquisition of features can be completed by analyzing course content, student behavior data, and course demand surveys. Based on the extracted features, a numerical vector is constructed for each teaching resource and each student need. The element value of each vector is usually calculated based on historical data (such as the student's usage time on this type of resource, interaction frequency, etc.). For example, if a student has used audio teaching materials for 50 hours and video teaching materials for 20 hours in the past period of time, a vector containing these features can be constructed to represent the student needs. Once the element value of each vector is determined, the dot product of the teaching resource and student demand vectors and their respective moduli can be calculated. The dot product is completed by multiplying and summing the eigenvalues ​​at the same position; the moduli are completed by calculating the square sum of each eigenvalue of the vector and taking the square root.

[0024] For example, there is the following data: the vector of teaching resource A , indicating that the audio duration of the resource is 2 hours, the video duration is 3 hours, the interactivity is 0, and the text duration is 1 hour. The vector of student demand B , indicating that the student needs 1 hour of audio, 4 hours of video, 0 hours of interactive learning, and 2 hours of text learning.

[0025] Compute the dot product: ; Calculate the modulus length: , ; Calculate cosine similarity: ; The cosine similarity value This indicates that the match between teaching resource A and student demand B is high, and the system should give priority to recommending this resource to students because it has a high degree of fit with students' learning needs.

[0026] See also Figure 4 , the specific steps for obtaining the recommended list of teaching resources are as follows: According to the similarity between the teaching resource vector and the student demand vector, the learning demand information is matched with the resource content in the teaching resource library to obtain the teaching resource matching result; First, the degree of match between each teaching resource and the student's learning needs is measured by calculating the similarity value between the two. The similarity value ranges from -1 to 1, where a value close to 1 indicates that the resource and the need are highly similar, a value close to 0 indicates that the two are not similar, and a value close to -1 indicates that they are completely opposite. For example, suppose that a student's learning needs are mainly audio and video, and there is a low demand for interactive learning materials. The system constructs the student's demand vector by analyzing the student's learning history data, behavior data, and other preference information, such as , where each number represents the degree of students' demand for audio, video, interactivity, and text content. Then, the system calculates the feature vector of each teaching resource in the teaching resource library. For example, the feature vector of a resource may be , where they represent the audio duration, video duration, interactivity and text content of the resource respectively. The system calculates the similarity between the resource and the student's needs and obtains a cosine similarity value, such as 0.933, which indicates that the resource is highly consistent with the student's needs. This value indicates that the resource has a high degree of match with the student's learning needs in terms of audio, video, etc., so it will be recommended to the student first. If the similarity value calculated by the system is low, such as 0.2, this indicates that the resource is poorly matched with the student's needs, may not contain the audio or video content required by the student, or the resource is more difficult, so the resource will be sorted at the back of the recommendation list. On the contrary, if the similarity is 1, it means that the resource is completely consistent with the needs, and the system will give priority to recommending the resource. Through such a matching process, the system can dynamically adjust the recommended teaching resources according to the personalized needs of students to ensure that the recommended content best meets the students' learning goals and preferences.

[0027] Based on the teaching resource matching results, the teaching resource library selects textbooks or courses that match the student's needs, and sorts them from high to low according to the matching degree to generate a recommended list of teaching resources; Resources are sorted according to the degree of matching. For example, if the matching degree of a teaching resource is 0.933, it means that the resource is highly consistent with the needs of students, and the system will give priority to recommending this resource. At the same time, if the matching degree of some resources is low, the system will exclude them or postpone the recommendation to ensure that the resources provided to students are more in line with their actual needs. In actual operation, the system first sorts the teaching resources according to all the calculated matching values, and then selects the most suitable teaching materials and courses from the resource library according to the personalized learning needs of each student (such as preferences for audio, video, interactivity, etc.). Finally, the system will generate a recommendation list, arranged from high to low according to the matching degree, to ensure that students can have priority access to the resources that best suit their current learning status. For example, for a student who prefers audio learning, the system may recommend audio teaching materials and voice courseware, while for students who prefer video learning, it recommends teaching videos with subtitles.

[0028] See also Figure 5 , the steps for calculating the relevance of multiple subject resources are as follows: Collect the learning needs and goals of the corresponding subjects from students' learning data and performance records in multiple subjects, identify the key needs of each subject by comparing students' subject preferences, analyze students' interdisciplinary learning goals based on the needs, and obtain the demand correlation data between subjects; First, collect students' historical learning data and performance records in various subjects, and combine students' grades, interaction data, homework submission and learning time in subjects such as mathematics, language, and science to determine their preferences and weaknesses in each subject. Then, by comparing the characteristics of students' needs in different subjects (such as learning methods, difficulty and form of learning content, etc.), the system will conduct a correlation analysis of the learning goals of different subjects. For example, for a student who performs well in mathematics, the system may recommend choosing some similar logical reasoning training materials in the Chinese subject to improve his interdisciplinary ability, while for another student with poor Chinese grades, resources for improving basic language skills will be recommended first. Through this analysis method, the system not only identifies students' needs in specific subjects, but also adjusts the recommended teaching resources according to interdisciplinary learning goals, thereby optimizing the recommendation list so that each student can get appropriate teaching support in multiple subjects.

[0029] Based on the demand correlation data between disciplines, the formula is adopted: ; Computing and disciplines Resource associations between , get the related information of multiple subject resources; in, and Represent two subjects, such as mathematics and language. In the calculation process, and Will traverse all subject resources. Indicates the first For example, a subject resource may have multiple characteristics (such as difficulty, interactivity, teaching material format, etc.). From 1 to Represents all characteristics. and Respectively represent subjects and disciplines Resources The resource characteristics of each subject can include multiple dimensions, such as textbook type, interactivity, video length, difficulty, etc. Specifically, and Indicated in the subject and disciplines In The value of a characteristic (such as the interactivity or difficulty of the textbook content). It is the number of dimensions of the feature, that is, the number of features each subject resource has. For example, a textbook can contain multiple features, such as difficulty, interactivity, video length, number of pictures, etc.

[0030] and Parameters are usually obtained by extracting the characteristics of subject resources. The characteristics of each subject resource are quantified by analyzing the content of the textbook, teacher feedback, learning data, etc. For example, the difficulty of a mathematics textbook may be obtained by the teacher's rating of the textbook content and the student's performance when using the textbook. Interactivity may be evaluated by calculating the proportion of interactive links in the textbook or the degree of student participation, and the video length is calculated by calculating the length of the textbook video. Through these analytical methods, the system will assign values ​​to the characteristics of each subject resource.

[0031] For example, there are two subject resources: The resource vector for (mathematics) is , where the first element represents the difficulty of the mathematics textbook, the second element represents the interactivity, the third element represents the duration of the video content, and the fourth element represents the duration of the text content. The resource vector for (language subject) is , where each element is the same as the mathematics subject resources and represents the individual characteristics of the language subject resources.

[0032] The calculation process is as follows: Calculate the absolute value of the difference for each feature: , calculate the sum of the differences: .

[0033] Calculate the correlation: ; What the results mean: Indicates subject and disciplines The resource correlation between them is 0.5.

[0034] See also Figure 6 , the specific steps for obtaining the recommended list of interdisciplinary teaching resources are: Based on the association information of multiple subject resources, analyze the resource relationship between each subject, compare the similarity of resource relationships in multiple dimensions, identify matching resources and screen them, and generate screened resource information; First, by comparing the correlation between multiple subject resources, we analyze the similarities and differences of each subject resource in different feature dimensions. The calculated result of the correlation usually falls between -1 and 1, indicating the strength of the relationship between two subject resources. In order to better understand and judge the significance of these calculation results, we divide the correlation into the following ranges and judge the size of the correlation according to the range: 0.8-1.0: The correlation is extremely high, indicating that the two subject resources are highly similar in most dimensions. The system should give priority to recommending such resources. There is no need to make too many screening adjustments when recommending them, because they can well meet the interdisciplinary needs of students. 0.5-0.8: The correlation is moderately high, indicating that the resources have strong similarities in some feature dimensions, but may also have differences in other dimensions. Such resources can be considered as recommended options, but when recommending, they may need to be appropriately adjusted according to the specific needs of students, such as by combining other subject resources or further analyzing student preferences to optimize the recommendation effect. 0.2-0.5: The correlation is moderately low, indicating that the similarity between the two subject resources is weak, and caution is needed when recommending. In this case, the system may need to further analyze other related subjects or make more content combinations to meet students' interdisciplinary needs and avoid recommending inappropriate resources. 0-0.2: Low correlation, indicating that the subject resources are quite different in multiple dimensions. The system needs to be very careful when recommending such resources. If such resources are directly recommended, they may not effectively meet students' interdisciplinary learning needs. They need to be combined with resources from other subjects or conduct more in-depth analysis. -1.0-0: Negative correlation, indicating that the two subject resources are completely opposite. Such resources are usually not recommended because the content and teaching methods between them are too different. Recommending such resources may not meet students' interdisciplinary needs. For example, the correlation between mathematics and language subjects is 0.5, and the system will regard it as a medium-to-high correlation. This means that in some characteristics (such as teaching difficulty or form), the resources of the two subjects have certain similarities, so they can be recommended as options for interdisciplinary resources, but other factors need to be combined for optimization when recommending. The system will further screen out the most suitable resources by analyzing students' specific needs and learning goals in different subjects, thereby ensuring that students can get the best interdisciplinary learning support.

[0035] Based on the filtered resource information, interdisciplinary resources are recommended in the order of relevance, and the recommended content of the teaching resource recommendation list is optimized according to the needs of students to generate an interdisciplinary teaching resource recommendation list; Based on the filtered resource information, the recommended content is further optimized. The system will prioritize resources that are highly relevant in multiple disciplines and can meet students' interdisciplinary learning needs, and generate a final recommendation list. This list will be sorted according to students' actual needs, interdisciplinary learning goals, and the relevance of resources to ensure that the resources recommended to students are efficient and targeted. Ultimately, the system will form a personalized interdisciplinary teaching resource recommendation list so that students can choose and learn according to their interdisciplinary goals.

[0036] See also Figure 7 , the specific steps for obtaining learning progress prediction data are: According to the recommended list of interdisciplinary teaching resources, the learning progress data, including learning time, task completion and answer accuracy, is analyzed to identify the student's learning trajectory and use the long short-term memory network model to study it and obtain the student's current learning status information; The system first collects and organizes these data. For each student, the system records his performance in different learning tasks, such as learning time, task completion, and correct answer rate. These data will be input into the long short-term memory network (LSTM) model, which can capture students' learning patterns by learning these time series data. For example, suppose a student's learning progress data in a mathematics course is as follows: in the first stage, the learning time is 2 hours, 80% of the tasks are completed, and 70% of the questions are answered correctly; in the second stage, the learning time is 1.5 hours, 75% of the tasks are completed, and 85% of the questions are answered correctly. The system arranges these data in chronological order and inputs them into the LSTM model. Based on these historical learning data, the LSTM model will analyze the students' learning performance in these two stages, identify the students' learning rules, and infer the students' possible learning behaviors and task completion in the next stage based on the past learning trajectory. Through the processing of the long short-term memory network, the model can not only remember the students' learning behaviors in the past stage, but also make predictions based on the time series characteristics and infer the students' possible learning progress in the future. In this way, the system can understand students' learning progress more accurately, provide a basis for subsequent learning resource recommendations, and provide data support for the calculation of future learning status.

[0037] Based on the student's current learning status information, the formula is used: ; Calculate students' future moments Learning status , ranging from 0 to 100. For example, 50 means that the student has mastered 50% of the course content, and the student's learning progress prediction result is obtained; in, Indicates that the student is at the current moment The learning status includes the student's current learning progress, knowledge mastered, task completion status, etc. This value is calculated based on the student's historical learning data, learning time, percentage of completed tasks, correct answer rate, etc. Indicates that the student is at the current moment The expected learning state or target learning state. The expected learning state is calculated based on the student's historical learning data and the difficulty of the current learning task, and is usually the best state that the student can achieve. For example, if the student is currently performing well in a certain subject, the expected learning state may be high; if the performance is poor, the expected learning state may be low. It is estimated based on the student's historical learning performance, answer accuracy, learning time and other data. Represents the learning rate coefficient, which controls the speed of updating the learning state. The value of is usually between 0 and 1. The larger the value, the faster the learning status is updated, and the faster the system adapts to the student's learning progress. This value is dynamically adjusted by the model according to the student's personalized learning progress, and is usually set based on the student's learning behavior (such as learning efficiency and learning frequency).

[0038] For example, a student is in the process of learning a course. is 60, the expected learning state is 75, the learning rate coefficient Set to 0.5.

[0039] Calculate the difference: ; Update learning status: ; Indicates the student's learning status in the next stage. Through this calculation, the student's learning status in the future can be predicted. This predicted value can help the system adjust the recommended learning resources according to the student's learning progress. If the student shows better learning progress during this period, the system will recommend more advanced resources, otherwise it will provide basic resources to help them consolidate their learning content.

[0040] See also Figure 8 , the specific steps for obtaining the optimization results of the teaching resource recommendation list are: Based on the predicted results of students' learning progress, the range of students' learning status is analyzed, and the recommended learning resource types and learning tasks are determined according to the selected range to obtain the recommended results of future learning resources; Adjust future learning resource recommendations based on learning progress prediction data and calculate results Indicates the student's learning status at a future moment. According to the preset status range, the learning status It is usually divided into the following levels: 0-50: Indicates that the student's current learning status is poor, the learning progress is slow, and there may be many difficulties. The system needs to recommend basic review materials and resources to help students understand basic concepts. -50-70: Indicates that the student's learning status is medium, the learning progress is normal, but still needs certain support. At this time, the system can recommend moderate challenging tasks to help students further consolidate their foundation and gradually improve their learning ability. -70-90: Indicates that the student's learning status is good, the learning progress is fast, and they can master more complex knowledge content. At this time, the system can recommend more advanced tasks, challenging content, and extended learning resources. -90-100: Indicates that the student's learning status is very good, most of the knowledge content has been mastered, and the progress is very fast. The system can recommend more advanced courses and independent projects to help students further broaden their knowledge. The student's predicted learning status is 67.5, which belongs to the range of 50-70, indicating that the student's learning progress is at a normal level, but there is still room for improvement. Therefore, the system needs to recommend moderate tasks and resources that can not only consolidate basic knowledge, but also slightly increase the difficulty of learning to help students maintain a good learning momentum. For example, assuming that the student is studying mathematics, the system will recommend some math questions of moderate difficulty based on his current learning progress, including application questions of basic concepts as well as some slightly more complex exercises, to ensure that the student can further improve through challenges without falling into learning tasks that are too simple or too difficult.

[0041] Based on the future learning resource recommendation results, the optimization results of the teaching resource recommendation list are generated by adding tasks and materials that match the students' current learning status, such as advanced exercises and interactive learning resources; Based on the adjustment results, the optimization results of the recommended list of teaching resources are obtained, and the system will adjust the recommended content according to the student's learning status. In this case, since the student's learning progress is at a normal level, the resources in the recommended list will include some moderate learning tasks and materials, such as advanced exercises and concept review materials that are suitable for their learning ability. In addition, the system may add some highly interactive learning resources, such as online simulation tests and interactive video courses, to enhance students' learning experience and interest. Ultimately, the optimized recommendation list will be adjusted according to the specific needs of students to ensure that the recommended resources can help students steadily improve their learning progress and master more knowledge points.

[0042] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A special education teaching resource recommendation system based on big data, characterized in that: The system comprises: The data collection and classification module collects the personal data of special education students through the big data platform, obtains the disability type of each special education student from the personal data, and classifies the disability type to generate special student classification data; The teaching resource matching module collects the learning needs of each special education student in the special student classification data, calculates the matching degree between the teaching resources and the student's learning needs, matches the learning need information with the resource content in the teaching resource library according to the matching degree, gives priority to recommending teaching materials or courses that match the student's current learning needs, and generates a teaching resource recommendation list; The interdisciplinary resource recommendation module optimizes the teaching resource recommendation list in combination with the students' interdisciplinary learning goals, calculates the relevance of multiple subject resources, recommends interdisciplinary resources based on the relevance information, and generates an interdisciplinary teaching resource recommendation list; The learning progress calculation module collects the learning progress data of students according to the interdisciplinary teaching resource recommendation list, uses the long short-term memory network model to learn the students' entire learning trajectory, calculates the students' future learning status based on the learning trajectory, obtains learning progress prediction data, adjusts future learning resource recommendations based on the learning progress prediction data, and obtains the optimization result of the teaching resource recommendation list.

2. The special education teaching resource recommendation system based on big data according to claim 1 is characterized in that: The steps for obtaining the special student classification data are specifically as follows: Based on the big data platform, personal data of special education students are collected, including basic information and disability type data of students. Disability type of each student is obtained through school registration and assessment data. Disability types include visual impairment, hearing impairment, cognitive impairment, and student disability type data is generated; The data of each student in the student disability type data is classified according to the disability type, and stored in multiple categories respectively to generate special student classification data.

3. The special education teaching resource recommendation system based on big data according to claim 2 is characterized in that: The steps of calculating the matching degree between teaching resources and students' learning needs are specifically as follows: Collect personal data of special education students, including basic information and learning behaviors, analyze the type of disability of each student, and determine the corresponding learning needs based on the student's behavioral data to generate student learning needs data; Based on the student learning needs data, the formula is adopted: ； Calculate the similarity between the teaching resource vector and the student demand vector ; in, and The first Features, is the characteristic vector of teaching resources, is the student's learning demand vector, is the dimension of the feature.

4. The special education teaching resource recommendation system based on big data according to claim 3 is characterized in that: The steps for obtaining the teaching resource recommendation list are specifically as follows: According to the similarity between the teaching resource vector and the student demand vector, the learning demand information is matched with the resource content in the teaching resource library to obtain a teaching resource matching result; Based on the teaching resource matching results, the teaching resource library screens the teaching materials or courses that match the student's needs, and sorts them from high to low according to the matching degree to generate a teaching resource recommendation list.

5. The special education teaching resource recommendation system based on big data according to claim 4 is characterized in that: The steps of calculating the relevance of multiple subject resources are specifically as follows: Collect the learning needs and goals of the corresponding subjects from students' learning data and performance records in multiple subjects, identify the key needs of each subject by comparing students' subject preferences, analyze students' interdisciplinary learning goals based on the needs, and obtain the demand correlation data between subjects; Based on the demand correlation data between the disciplines, the formula is adopted: ； Computing and disciplines Resource associations between , get the related information of multiple subject resources; in, and Representing two disciplines, Indicates the first Features, Represents all features, and Respectively represent subjects and disciplines Resources eigenvalues, and Indicated in the subject and disciplines In Features, is the number of feature dimensions.

6. The special education teaching resource recommendation system based on big data according to claim 5 is characterized in that: The steps for obtaining the recommended list of interdisciplinary teaching resources are specifically as follows: Based on the association information of the multiple subject resources, analyzing the resource relationship between each subject, comparing the similarity of the resource relationship in multiple dimensions, identifying and screening matching resources, and generating screened resource information; Based on the filtered resource information, interdisciplinary resources are recommended in order of relevance, and the recommended content of the teaching resource recommendation list is optimized according to the needs of students to generate an interdisciplinary teaching resource recommendation list.

7. The special education teaching resource recommendation system based on big data according to claim 6 is characterized in that: The steps for obtaining the learning progress prediction data are specifically as follows: The learning progress data of the students according to the recommended list of interdisciplinary teaching resources, including learning time, task completion and correct answer rate, are analyzed to identify the students' learning trajectory, and the long short-term memory network model is used to study it to obtain the students' current learning status information; Based on the student's current learning status information, the formula is adopted: ； Calculate students' future moments Learning status , get the predicted result of students’ learning progress; in, Indicates that the student is at the current moment The learning status Indicates that the student is at the current moment The expected learning state or target learning state, Represents the learning rate coefficient.

8. The special education teaching resource recommendation system based on big data according to claim 7 is characterized in that: The steps for obtaining the optimization results of the teaching resource recommendation list are specifically as follows: Based on the predicted learning progress result of the student, the range of the student's learning status is analyzed, and the recommended learning resource type is determined according to the selected range, and the learning task is recommended to obtain the future learning resource recommendation result; Based on the future learning resource recommendation results, an optimization result of the teaching resource recommendation list is generated by adding tasks and materials that match the student's current learning status, such as advanced exercises and interactive learning resources.

Images