Education service matching method and system based on multi-source big data Ai system

Abstract

The invention discloses an education service matching method and system based on a multi-source big data Ai system, and belongs to the field of education service matching, and the method comprises a data collection module which is used for collecting user data and education service data, and the user data comprises user basic information, learning historical data, social data and online behavior data; the education service data comprises course information, mechanism information and service dynamic data. The data preprocessing module is used for cleaning, standardizing and normalizing the collected multi-source data. And the user portrait construction module is used for mining potential demands and learning preferences of the user by applying a deep learning algorithm based on the preprocessed user data, and constructing a multi-dimensional user portrait. By collecting multi-source data, user demands and education service characteristics are comprehensively and accurately reflected, and compared with a traditional matching mode of a single data source, the user and service characteristics can be more accurately grasped, and the matching accuracy is improved.

Description

Technical Field

[0001] The present invention relates to the field of educational service matching technology, and in particular to an educational service matching method and system based on a multi-source big data Ai system. Background Art

[0002] With the rapid development of the education industry, the variety of educational services is becoming increasingly diverse, including online courses, offline training, one-on-one tutoring, and educational consulting. Students, parents, and other education demanders hope to efficiently obtain education services that match their needs, while education service providers also hope to accurately connect with potential customers. However, the current education service matching mainly faces the following problems:

[0003] The data source is single and mainly relies on simple information filled in by users, which cannot fully reflect the real needs and characteristics of users, resulting in inaccurate matching results.

[0004] The matching algorithm is simple, usually based only on keyword matching or information comparison in a small number of dimensions. It is difficult to take into account the diversity of educational services and the complexity of user needs, and the matching efficiency and quality are low.

[0005] There is a lack of a dynamic adjustment mechanism, and matching results cannot be optimized in a timely manner according to real-time changes in user behavior and educational services, making it difficult to meet the ever-changing needs of users.

[0006] Therefore, there is an urgent need for an education service matching method and system based on multi-source big data and advanced artificial intelligence technology to improve the accuracy, efficiency and quality of education service matching. Summary of the Invention

[0007] In response to the shortcomings of the existing technology, the present invention provides an education service matching method and system based on a multi-source big data Ai system. By integrating multi-source data and applying artificial intelligence algorithms, it can achieve accurate and efficient matching of education services with user needs, improve the utilization efficiency of education service resources, and meet the diverse education needs of users.

[0008] Technical Solution: To solve the above technical problems, according to one aspect of the present invention, more specifically, an education service matching system based on a multi-source big data AI system, comprising:

[0009] Data collection module: used to collect user data and education service data. The user data includes user basic information, learning history data, social data, and online behavior data. The education service data includes course information, institution information, and service dynamic data.

[0010] Data preprocessing module: used to clean, standardize and normalize the collected multi-source data.

[0011] User portrait construction module: Based on pre-processed user data, deep learning algorithms are used to explore users' potential needs and learning preferences to build multi-dimensional user portraits.

[0012] Education service portrait construction module: uses natural language processing technology and machine learning algorithms to analyze and evaluate education service data and construct education service portraits.

[0013] Matching model construction and training module: Build a deep learning-based matching model and use historical matching data and user feedback data for training and optimization.

[0014] Matching and recommendation module: Input user profiles and education service profiles into the matching model, calculate the matching score, sort and recommend education services based on the score, and provide matching basis instructions.

[0015] Dynamic optimization and feedback module: monitor user behavior and collect feedback information in real time, and dynamically update and optimize user portraits, educational service portraits and matching models.

[0016] According to another aspect of the present invention, more specifically, an education service matching method based on a multi-source big data Ai system, based on the above-mentioned education service matching system, includes the following steps:

[0017] S1. Multi-source data collection and preprocessing

[0018] User data is collected through the data collection module, covering basic user information (age, gender, education level, location, etc.), learning history data (course learning records, test scores, learning time, etc.), social data (learning-related content on social platforms, social learning exchange information, etc.), and online behavior data (browsing, searching, and staying data on educational websites or apps, etc.).

[0019] The data collection module is used to collect educational service data, including course information (name, content, teacher, duration, difficulty, etc.), institution information (qualifications, faculty, reputation, etc.), and service dynamic data (course updates, remaining places, promotional activities, etc.).

[0020] The collected data is processed with the help of the data preprocessing module, and duplicate, erroneous and invalid data are removed through data cleaning; data standardization is used to convert data of different formats and ranges into a unified format; data normalization is used to make the data within a specific numerical range.

[0021] S2. User portrait construction

[0022] The user profile building module uses pre-processed user data and deep learning algorithms, such as convolutional neural networks and recurrent neural networks, to uncover users' potential needs and learning preferences. It analyzes learning history data to identify subject strengths and learning styles, and combines social and online behavior data to clarify learning interests and goals.

[0023] The user portrait construction module constructs multi-dimensional user portraits in a labeled form based on the mining results, such as "mathematics-weak students", "English learning enthusiasts", "online course preference", etc.

[0024] S3. Construction of educational service portrait

[0025] The education service portrait construction module uses natural language processing technology to perform semantic analysis and keyword extraction on texts such as course content and institution introductions based on education service data, and to extract core characteristics and advantages.

[0026] Combined with service dynamic data, the education service portrait construction module uses machine learning algorithms, such as decision trees and random forests, to classify and evaluate education services, and determine attributes such as applicable population and teaching quality level.

[0027] The education service portrait construction module constructs education service portraits in a labeled form, such as "elementary programming courses", "lectures by senior teachers", "limited-time discount courses", etc.

[0028] S4. Matching model construction and training

[0029] The matching model construction and training module constructs a matching model based on deep learning, such as a deep neural network, which takes the user portrait and the education service portrait as the input layer, extracts and fuses features through multiple layers of neurons in the hidden layer, and obtains the matching score in the output layer.

[0030] The matching model construction and training module collects historical matching data and user feedback data, uses the back-propagation algorithm to adjust the model parameters, minimizes the error between the predicted matching degree and the actual feedback matching degree, and improves the model accuracy and generalization ability.

[0031] S5. Educational service matching and recommendation

[0032] The matching and recommendation module inputs the constructed user portrait and education service portrait into the trained matching model to calculate the matching score between each education service and the user.

[0033] The matching and recommendation module ranks educational services according to the matching scores, selects several high-scoring educational services and recommends them to users, and provides detailed matching basis descriptions, such as matching tags, advantage comparisons, etc.

[0034] S6. Dynamic Optimization and Feedback

[0035] The dynamic optimization and feedback module monitors users' usage behaviors of recommended educational services in real time, such as clicks, registrations, learning progress, etc., and collects feedback information such as satisfaction evaluations and improvement suggestions.

[0036] The dynamic optimization and feedback module dynamically updates user portraits and educational service portraits based on user behavior and feedback data, adjusts matching model parameters, and optimizes matching strategies to adapt to demand and market changes.

[0037] The beneficial effects of the educational service matching method and system based on the multi-source big data Ai system of the present invention are:

[0038] (1) The present invention collects multi-source data to comprehensively and accurately reflect user needs and educational service characteristics. Compared with the traditional matching method of a single data source, it can more accurately grasp the characteristics of users and services and improve the accuracy of matching.

[0039] By using advanced artificial intelligence technologies such as deep learning, natural language processing and machine learning, we build user portraits, educational service portraits and matching models, fully considering the diversity of educational services and the complexity of user needs. Compared with simple matching algorithms, it greatly improves the efficiency and quality of matching.

[0040] It has a dynamic optimization mechanism that can adjust matching strategies in real time based on user behavior and feedback, adapt to the ever-changing education service market and user needs, and continuously provide users with more expected education service recommendations, thereby improving user satisfaction and the utilization efficiency of education service resources.

[0041] Through transfer learning and adaptive architecture design, the matching model can be quickly migrated to fields such as e-commerce recommendations, talent recruitment, and medical resource matching. Simply replacing the input data source and adjusting the labeling system can reduce the development costs of multi-scenario applications.

[0042] The system achieves millisecond-level response through model lightweighting and computational optimization, supporting tens of thousands of concurrent requests per second. It also adopts online learning and integration strategies to ensure that the accuracy of recommendation results remains above 90%, meeting users' dual needs for real-time and accuracy in high-concurrency scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The present invention will be further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0044] Figure 1 Schematic diagram of the process of the educational service matching method based on the multi-source big data Ai system of the present invention;

[0045] Figure 2 This is a schematic diagram of the framework of the education service matching system based on the multi-source big data Ai system of the present invention. DETAILED DESCRIPTION

[0046] The present invention will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other.

[0047] In order to make the technical solution of the present invention clearer, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0048] Example 1

[0049] 1. Education service matching system based on multi-source big data AI system, including:

[0050] The data collection module includes a user data collection unit: collecting basic information through user registration and login forms; cooperating with schools and educational institutions to obtain learning history data; accessing the open interface of the social platform to collect social data with user authorization; and embedding data collection code in educational websites or apps to record online behavior data.

[0051] Education service data collection unit: establishes a data interface with education service providers and regularly synchronizes course and institution information; uses web crawler technology to capture service dynamic data in real time, such as course updates, promotional activities, etc.

[0052] Data preprocessing module, including data cleaning unit: using regular expressions and data validation rules to identify and delete duplicate records, repair erroneous data; using outlier detection algorithm to remove unreasonable data

[0053] Data standardization unit: Z-score standardization is used for numerical data, converting it into a standard normal distribution with a mean of 0 and a standard deviation of 1; one-hot encoding conversion is used for categorical data.

[0054] Data normalization unit: Use the Min-Max normalization method to map the data to the [0,1] interval. The formula is

[0055]

[0056] where X norm is the normalized data, X is the original data, and X max and X min are the maximum and minimum values ​​in the original data, respectively.

[0057] The user portrait construction module includes a potential demand mining unit: using convolutional neural networks to extract features from images, videos, etc. in the user's learning history and analyze learning interests; using recurrent neural networks to model learning record sequences and predict learning trends and needs.

[0058] Learning preference analysis unit: Combined with social data text, natural language processing technology is used to perform sentiment analysis and topic extraction to clarify users' attitudes and interests in different learning areas.

[0059] User profile generation unit: This unit categorizes and summarizes the mined user features and assigns labels. For example, a "math-weak student" label is assigned based on math grades and time spent studying, while an "English learning enthusiast" label is assigned based on English learning behavior. These labels are then combined to form a complete user profile.

[0060] The education service portrait construction module includes a text information analysis unit: it uses natural language processing to perform lexical, syntactic and semantic analysis on texts such as course content and institution introductions to extract keywords and key phrases, such as core knowledge points of the course and the characteristics and advantages of the institution.

[0061] Educational service evaluation unit: Use the decision tree algorithm to classify courses based on course difficulty, applicable population, etc.; use the random forest algorithm, combined with user evaluation, teaching staff, etc. to evaluate the teaching quality of educational services and determine the level.

[0062] Educational service profile generation unit: This unit converts the extracted keywords, classification results, and evaluation levels into labels to construct an educational service profile. For example, a Python programming course for beginners can be given labels such as "Python Programming Course," "Beginner Course," and "Suitable for beginners."

[0063] The matching model construction and training module includes a model construction unit: building a deep neural network, the input layer receives the user portrait and education service portrait feature vectors; the hidden layer has multiple neuron layers, and extracts and nonlinearly transforms features through activation functions such as ReLU; the output layer uses the Softmax function to calculate the matching probability and obtain the matching score.

[0064] Model training unit: Collect historical matching data and user feedback, dividing it into training, validation, and test sets. Using mean squared error as the loss function, the model is trained on the training set using the backpropagation algorithm. Parameters are adjusted to minimize the loss on the validation set, and good generalization is evaluated on the test set.

[0065] The matching and recommendation module includes a matching degree calculation unit: the user portrait and education service portrait feature vectors are input into the trained matching model to calculate the matching degree score between each education service and the user.

[0066] Sorting and recommendation unit: Sort educational services in descending order according to the matching scores, select the top services and recommend them to users, generate matching basis descriptions, and display matching tags and key information.

[0067] The dynamic optimization and feedback module includes a user behavior monitoring unit: it uses tracking technology to record user clicks, registrations, learning progress and other behavioral data on the education service platform.

[0068] Feedback collection unit: Set up a user feedback portal to collect satisfaction evaluations and improvement suggestions for recommended educational services.

[0069] Dynamic update unit: Based on user behavior data, such as frequent clicks on certain courses without signing up, update user profiles and adjust learning preferences; based on user feedback, revise educational service profiles and improve service descriptions and evaluations.

[0070] Model optimization unit: Use updated data to retrain the matching model, adjust parameters to optimize the matching strategy, and improve the accuracy of subsequent matching.

[0071] The matching model construction and training module further integrates the transfer learning framework, and enables the model to have cross-domain generalization capabilities through the pre-training-fine-tuning paradigm; at the same time, it uses knowledge distillation technology to generate a lightweight inference model to ensure the real-time performance of matching calculations.

[0072] The dynamic optimization module supports cross-domain feedback data access and continuously improves the versatility of the model.

[0073] Example 2

[0074] The specific implementation steps of the education service matching method based on the multi-source big data AI system are as follows:

[0075] S1. Multi-source data collection and preprocessing

[0076] Data collection: According to the implementation methods of the user data collection unit and the education service data collection unit of the data collection module, user data and education service data are collected respectively.

[0077] Data cleaning: Clean the collected data according to the implementation of the data cleaning unit of the data preprocessing module.

[0078] Data standardization: The cleaned data is standardized according to the implementation method of the data standardization unit of the data preprocessing module.

[0079] Data normalization: According to the implementation of the data normalization unit of the data preprocessing module, the standardized data is normalized.

[0080] S2. User portrait construction

[0081] Potential demand mining: Follow the implementation method of the potential demand mining unit of the user portrait construction module to mine users' potential needs.

[0082] Learning preference analysis: Analyze user learning preferences according to the implementation of the learning preference analysis unit of the user portrait construction module.

[0083] User portrait generation: Generate a user portrait according to the implementation method of the user portrait generation unit of the user portrait construction module.

[0084] S3. Construction of educational service portrait

[0085] Text information analysis: Analyze the education service text information according to the implementation method of the text information analysis unit of the education service portrait construction module.

[0086] Educational service evaluation: Evaluate educational services based on the implementation of the educational service evaluation unit of the educational service portrait construction module.

[0087] Educational service portrait generation: Generate an educational service portrait according to the implementation method of the educational service portrait generation unit of the educational service portrait construction module.

[0088] S4. Matching model construction and training

[0089] Model construction: Build a matching model by referring to the implementation method of the model construction unit of the matching model construction and training module.

[0090] Model training: Based on the implementation of the model training unit of the matching model building and training module, the model is trained using historical data and user feedback.

[0091] S5. Educational service matching and recommendation

[0092] Matching degree calculation: According to the implementation method of the matching degree calculation unit of the matching and recommendation module, the matching degree score between the educational service and the user is calculated.

[0093] Sorting and recommendation: Based on the implementation of the sorting and recommendation unit of the matching and recommendation module, educational services are sorted and recommended, and matching basis descriptions are provided.

[0094] S6. Dynamic Optimization and Feedback

[0095] User behavior monitoring: Monitor user behavior according to the implementation method of the user behavior monitoring unit of the dynamic optimization and feedback module.

[0096] Feedback collection: Collect user feedback according to the implementation of the feedback collection unit of the dynamic optimization and feedback module.

[0097] Dynamic update: Update the user profile and education service profile according to the implementation method of the dynamic update unit of the dynamic optimization and feedback module.

[0098] Model optimization: Optimize the matching model according to the implementation method of the model optimization unit of the dynamic optimization and feedback module.

[0099] The above-described embodiments merely illustrate several implementations of the present invention, and while their descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

Claims

1. Education service matching system based on multi-source big data AI system, characterized by: include: A data collection module is used to collect user data and educational service data. The user data includes basic user information, learning history data, social data, and online behavior data. The educational service data includes course information, institution information, and service dynamic data. Data preprocessing module, used to clean, standardize and normalize the collected multi-source data; The user portrait construction module uses deep learning algorithms to mine users' potential needs and learning preferences based on pre-processed user data, and constructs a multi-dimensional user portrait; The education service profile building module uses natural language processing technology and machine learning algorithms to analyze and evaluate education service data and build an education service profile; Matching model construction and training module: builds a matching model based on deep learning and uses historical matching data and user feedback data for training and optimization; The matching and recommendation module inputs the user profile and the education service profile into the matching model, calculates the matching score, sorts and recommends education services based on the score, and provides an explanation of the matching basis; The dynamic optimization and feedback module monitors user behavior and collects feedback information in real time, and dynamically updates and optimizes user portraits, educational service portraits, and matching models.

2. The education service matching system based on multi-source big data Ai system according to claim 1 is characterized in that: The deep learning algorithm includes convolutional neural network and recurrent neural network, the natural language processing technology includes semantic analysis and keyword extraction, and the machine learning algorithm includes decision tree and random forest.

3. The education service matching system based on multi-source big data Ai system according to claim 2 is characterized in that: The user portrait and educational service portrait are both presented in a labeled form.

4. The education service matching system based on multi-source big data Ai system according to claim 3 is characterized in that: The matching model is a deep neural network, and the matching score is calculated by a Softmax function.

5. The educational service matching method based on the multi-source big data Ai system uses the educational service matching system based on the multi-source big data Ai system as claimed in claim 4, characterized in that: The steps include: S1. Using the data collection module to collect user data and educational service data, the user data includes user basic information, learning history data, social data, and online behavior data; the educational service data includes course information, institution information, and service dynamic data, and the data preprocessing module performs cleaning, standardization, and normalization on the collected data; S2. Using the user profile building module, based on the pre-processed user data, a deep learning algorithm is used to mine the user's potential needs and learning preferences, and to build a multi-dimensional user profile; S3. Analyze and evaluate education service data using the education service profile building module and natural language processing technology and machine learning algorithms to build an education service profile. S4. Using the matching model construction and training module to build a matching model based on deep learning, and using historical matching data and user feedback data to train and optimize the matching model; S5. Input the user profile and the education service profile into the matching model through the matching and recommendation module, calculate the matching score, sort and recommend education services according to the score, and provide a matching basis description; S6. Relying on the dynamic optimization and feedback module, monitor user behavior and collect feedback information in real time, and dynamically update and optimize user portraits, educational service portraits and matching models.

6. The educational service matching method based on the multi-source big data Ai system according to claim 5 is characterized in that: The deep learning algorithm includes convolutional neural network and recurrent neural network, the natural language processing technology includes semantic analysis and keyword extraction, and the machine learning algorithm includes decision tree and random forest.

7. The educational service matching method based on the multi-source big data Ai system according to claim 5 is characterized in that: The user portrait and educational service portrait are both presented in a labeled form.

8. The educational service matching method based on the multi-source big data Ai system according to claim 5 is characterized in that: The matching model is a deep neural network, and the matching score is calculated by a Softmax function.

9. The education service matching system based on multi-source big data Ai system according to claim 1 is characterized in that: The system has cross-domain versatility: the matching model construction and training module adopts a transfer learning framework to extract cross-domain universal features through pre-trained deep neural networks; The label system of user portraits and educational service portraits supports custom extensions, and adapts to resource matching scenarios in e-commerce, recruitment, and medical fields by replacing the input data source; the dynamic optimization and feedback module accesses cross-domain user behavior data to continuously optimize the model's generalization capabilities.

10. The education service matching system based on multi-source big data Ai system according to claim 1 is characterized in that: The system is efficient and convenient: the matching and recommendation module integrates knowledge distillation technology to compress the size of the matching model, and uses the approximate nearest neighbor (ANN) algorithm to establish a portrait vector index to achieve millisecond-level matching response; the dynamic optimization and feedback module deploys an online learning mechanism to avoid full retraining delays by incrementally updating model parameters; the matching model integrates a collaborative filtering lightweight model and uses a weighted voting strategy to ensure a matching accuracy rate of >90% and a response time of ≤200ms.

Images