College AI intelligent one-network communication system based on agent platform

Abstract

The invention relates to the field of intelligent service platforms, and discloses a college AI intelligent one-network communication system based on an agent platform, and the system comprises an SQL processing module which is used for converting a natural language inputted by a user into an SQL statement of a target database; the vector library management module is butted and adapted with various mainstream opening and closing source vector libraries, and interacts with different vector libraries through a uniform API (Application Program Interface); the extensible architecture module supports a multi-model and multi-reasoning framework and processes data from a database, a webpage or a PDF (Portable Document Format) file; and the vertical field professional knowledge module is used for carrying out data preprocessing and model pre-training on vertical field data of colleges and universities, outputting material completeness scores, missing item detection and emergency degree classification information, adjusting parameter setting according to industry characteristics, and customizing and optimizing interaction experience through role prompt words. According to the invention, intelligent processing and analysis of multi-source data can be realized, and the business handling efficiency and the user experience of the university online office hall are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent service platforms, and specifically to a university AI intelligent one-stop service system based on an agent platform. Background Art

[0002] With the in-depth development of university informatization construction, the online service hall has become an important channel for teachers and students to handle campus business. However, in the actual use process, teachers and students face many challenges. First of all, there are a wide variety of services in the online service hall, and it is difficult to find the entrances. Especially in complex business fields such as finance, the existence of multiple entrances makes it difficult for teachers and students to quickly locate the correct entrance when handling business such as reimbursement. Secondly, teachers and students often do not understand the business rules well, resulting in frequent obstacles due to missing materials during the handling process, which not only wastes time but also affects the smooth handling of business. In addition, the lack of process guidance is also a major problem. Teachers and students need to continuously query the process guidance when handling business, resulting in slow progress of business handling and poor user experience.

[0003] To solve the above problems, universities have begun to explore the introduction of AI technology into the online service hall, and through API docking between the agent platform and the port of each business system in the online service hall, intelligent business handling is realized. However, the existing agent platforms still have deficiencies in many aspects: the existing platforms lack flexible support for multiple vector databases, and it is difficult to handle the storage and management of large-scale data. Especially in high-concurrency scenarios, the efficiency of data access and processing is low. The natural language processing models used by mainstream GPT platforms are relatively weak in the professionalism of database operations, and it is difficult to accurately understand and generate complex database query statements; it is difficult to support users to flexibly expand and customize the model according to specific needs and different database environments; when dealing with problems in specific industries such as universities, it may not be able to provide in-depth and accurate answers like professionals. Summary of the Invention

[0004] The purpose of the present invention is to solve the problems existing in the prior art, and to propose an AI large model agent platform system applied to the education and government industries, aiming to solve the problems such as difficult entrance finding, lack of understanding of business rules, lack of process guidance, etc. existing in the process of handling business in the university online service hall through intelligent means, as well as the technical problems such as low efficiency of data access and processing, difficulty in accurately understanding and generating complex database query statements existing in the existing agent platforms; it is difficult to support users to flexibly expand and customize the model according to specific needs and different database environments; and it is impossible to provide in-depth and accurate answers.

[0005] To achieve the above purpose, the present invention adopts the following technical solutions: A university AI intelligent one-stop service system based on an agent platform, comprising: The SQL processing module is used to organize the SQL of existing mainstream databases, and convert the natural language input by users into SQL statements of the target database through pre-trained machine learning or deep learning models; The vector library management module is docked and adapted with various mainstream open-source and closed-source vector libraries, interacts with different vector libraries through a unified API interface, and supports dynamic data synchronization and multi-modal joint retrieval; The extensible architecture module supports multiple models and multiple inference frameworks, including submitting queries, the retriever component selecting relevant information, and the re-ranker refining the selection, is used to process data from databases, web pages or PDF files, and continuously optimizes the data-driven engine through proxy and plugin mechanisms, and ensures the scalability of the platform through the knowledge base and intelligent agent workflow component construction; user-defined data types, index strategies and query optimization plugins; The vertical domain expertise module performs data preprocessing and pre-training models on data in the vertical domain of universities, outputs material completeness scores, missing item detection and urgency classification information, and adjusts parameter settings according to industry characteristics, and customizes and optimizes the interaction experience through role-based prompt words.

[0006] Furthermore, the SQL processing module includes: The requirement analysis unit is used to determine the target database type and the type of SQL statements to be supported, as well as the conversion requirements from natural language to SQL; The data collection unit is used to collect database SQL statement samples and natural language query samples and their corresponding SQL statements; The data preprocessing unit is used to standardize the collected SQL statements and perform natural language processing operations on natural language queries, including word segmentation and part-of-speech tagging; The model selection and training unit is used to select machine learning or deep learning models and train the models using the preprocessed data so that they can understand natural language and generate corresponding SQL statements; The model optimization unit is used to evaluate the models, optimize the models according to the evaluation results, and integrate the trained models into the intelligent agent platform.

[0007] Furthermore, the vector library management module includes: The document acquisition unit is used to obtain the API interface documents of each vector library from the preselected list of vector libraries; The platform API interface unit is used to provide a unified API interface, enabling the platform to interact with different vector libraries through these interfaces, and the abstraction layer of the API interface can handle the differences between different vector libraries.

[0008] Furthermore, the extensible architecture module includes: Plugin architecture unit, which is used to integrate custom plugins developed by third-party developers and provide necessary tools and libraries; Data type and index specification definition unit, which is used to define the specifications of new data types and customized indexes; among them, the specifications of the new data types include data structures and operation interfaces, and the specifications of the customized indexes include index structures and index methods; Query optimization unit, which is used to implement a query optimizer and support user-defined query optimization strategies; Plugin management tool, which is used to facilitate users to install, configure and uninstall plugins; Testing and verification unit, which is used to test the plugin framework and new functions, and verify whether the new data types and index types meet the data storage requirements of specific fields.

[0009] Furthermore, the vertical domain expertise module includes: Data collection and preprocessing unit, which is used to collect the proprietary data sets of the university industry, including text and voice data, and perform data preprocessing, including data cleaning, annotation and structuring; Model selection and training unit, which is used to process materials through a multi-task learning framework, select appropriate machine learning or deep learning algorithms, train the model using the proprietary data sets after data preprocessing, and adjust parameters according to the model performance to achieve multi-modal material pre-review; Role division unit, which is used to customize different prompt words and interaction logics according to the role division of the user organization structure, and simulate the tone, intonation and speech rate of university industry users; User feedback unit, which is used to collect user feedback and perform iterative optimization according to the feedback.

[0010] Furthermore, the SQL processing module further includes an agent interaction layer, which uses a Bi-LSTM+CRF hybrid model to implement business intent recognition, constructs a dynamic business topology map, dynamically adjusts node weights according to the real-time access volume, provides recommended entrances and confidence scores for users, and supports multi-round dialogue management, and can perform context understanding according to user feedback.

[0011] Furthermore, among them, the SQL processing module further includes an NLP engine, a rule engine and a speech engine, which are respectively used for intent recognition, rule reasoning and speech processing; Among them, the NLP engine integrates intent recognition and multi-round dialogue management, and the model selection and training unit trains SQL statements through the machine learning or deep learning algorithm to realize the conversion from natural language to SQL statements, and supports university users to perform data query operations; The rule engine dynamically loads SWRL rules provided by each business department, supporting the automated management of business processes; The speech engine supports mixed Chinese and English speech processing, uses the Conformer model for speech recognition, and is fine-tuned on the corpus of college scenarios, while providing multiple speech synthesis modes.

[0012] Furthermore, the Bi-LSTM+CRF hybrid model is constructed by combining bidirectional long short-term memory networks and conditional random fields, including: An input layer for receiving text data and converting each character or word into a word vector representation; A word embedding layer for mapping word vectors to a high-dimensional space to capture semantic relationships between words; A Bi-LSTM layer composed of a forward LSTM and a backward LSTM, which can capture both forward and backward information in the sequence and the output at each time step contains the context information of the current position; A fully connected layer for mapping the output of the Bi-LSTM layer to the label space and outputting the scores of each position corresponding to each label; A CRF layer for globally optimizing the output of the fully connected layer according to the dependencies between labels, and outputting the optimal label sequence by calculating the transition matrix and the emission matrix.

[0013] Furthermore, the SQL processing module also includes a data service layer that is connected to multiple business databases in real time, constructs a cross-system data federation query interface, and integrates a multi-modal feature extractor; the data service layer is connected to multiple business databases of academic affairs, finance, and student affairs in real time, constructs a cross-system data federation query interface, and supports the storage, processing, and analysis of large-scale data. It integrates a multi-modal feature extractor, which can perform feature extraction and analysis on various types of data such as text, images, and tables, providing rich data support for upper-layer applications.

[0014] Furthermore, the vertical domain expertise module also includes a natural language processing model and a speech recognition model, which are respectively used to process text and speech data; The deep learning algorithm of the speech recognition model includes a Transformer model for processing long sequence data and capturing global dependencies, and provides a rich industry knowledge base and a question answering system, supporting users to obtain industry knowledge and answer questions through natural language queries.

[0015] Compared with the prior art, the present invention has the following beneficial effects: (1) Through the enhancement of the intelligent agent interaction layer, the present invention realizes more accurate business intention recognition, provides more precise business entrance recommendations for teachers and students, and greatly shortens the time to find business entrances. By integrating the NLP engine and the rule engine, the conversion from natural language to SQL statements is realized, supporting teachers and students to directly query the database through voice or text without manually writing SQL statements, improving the query efficiency.

[0016] (2) Through the construction and real-time adjustment of the dynamic business topology map, the system of the present invention can dynamically adjust the node weights according to the real-time access volume, provide more reasonable business entrance sorting and confidence scores for users, and improve the user experience. The multi-round dialogue management function supports complex dialogue processes, can understand the context according to user feedback, and realizes a more natural and smooth interaction experience.

[0017] (3) The vector library management module of the present invention supports a variety of mainstream open-source and closed-source vector libraries, realizes dynamic data synchronization and multi-modal joint retrieval, and improves the flexibility and efficiency of data processing. The data service layer is connected to multiple business databases in real time, constructs a cross-system data federation query interface, supports the storage, processing and analysis of large-scale data, and provides rich data support for upper-layer applications.

[0018] (4) The extensible architecture module of the present invention supports multiple models and multiple inference frameworks. Users can flexibly expand and customize the models according to specific requirements and different database environments, improving the scalability of the system. The vertical domain professional knowledge module conducts a large amount of data preprocessing and pre-training for specific industries such as universities, improving the system's ability to answer questions in specific industries. Through role-based prompt word customization and a rich industry knowledge base, the system can simulate the tone, intonation and speech rate of university industry users, providing a more user-demand-oriented interaction experience.

[0019] In summary, through the integration of SQL processing capabilities, vector library management, extensible architecture and vertical domain professional knowledge applications, the present invention realizes the intelligent processing and analysis of multi-source data, significantly improves the business handling efficiency and user experience of the university online service hall, and contributes to the digital transformation and intelligent upgrading of the education industry. Brief Description of the Drawings

[0020] Figure 1 It is a schematic diagram of the modules of a university AI intelligent one-stop service system based on an intelligent agent platform proposed by an embodiment of the present invention.

[0021] Figure 2 It is the overall platform architecture of a university AI intelligent one-stop service system based on an intelligent agent platform proposed by an embodiment of the present invention; Figure 3Schematic diagram of the construction process of the SQL processing module of an AI large model intelligent agent platform system applied to the education and government industries proposed in an embodiment of the present invention.

[0022] Figure 4 Schematic diagram of the construction process of the vector library management module of a university AI intelligent one-stop service system based on an intelligent agent platform proposed in an embodiment of the present invention.

[0023] Figure 5 Schematic diagram of the construction process of the scalable architecture module of a university AI intelligent one-stop service system based on an intelligent agent platform proposed in an embodiment of the present invention.

[0024] Figure 6 Schematic diagram of the construction process of the vertical domain professional knowledge module of a university AI intelligent one-stop service system based on an intelligent agent platform proposed in an embodiment of the present invention. Detailed implementation manners

[0025] Next, the technical solutions of the present invention will be clearly and completely described in conjunction with the accompanying drawings. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0026] In the description of the present invention, it should be noted that the orientation or positional relationship indicated by the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. is based on the orientation or positional relationship shown in the accompanying drawings, and is 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 a limitation of the present invention. In addition, the terms "first", "second", and "third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0027] In addition, the technical features involved in different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0028] Embodiment 1 Refer to Figures 1-6 As shown, a university AI intelligent one-stop service system 100 based on an intelligent agent platform includes: An SQL processing module 110, which is used to sort out the SQL of existing mainstream databases and convert the natural language input by the user into SQL statements of the target database through a pre-trained machine learning or deep learning model; Among them, the target database type (such as MySQL, PostgreSQL, etc.) stores data such as student information, course information, and library collections in the school. The supported SQL statement types (such as SELECT, WHERE, etc.) can meet the data query needs of teachers and students. Analyze the conversion requirements from natural language to SQL, including the complexity of queries (such as multi-table joins) and flexibility (such as support for fuzzy queries), etc.

[0029] The vector library management module 120 is docked and adapted with various mainstream open-source and closed-source vector libraries, interacts with different vector libraries through a unified API interface, and supports dynamic data synchronization and multi-modal joint retrieval. Among them, the mainstream open-source and closed-source vector libraries include Milvus, Pinecone, Vespa, etc. According to the characteristics, performance, and API interfaces of each vector library, the most suitable vector library for the school's needs is selected. Through the unified API interface, the platform can interact with different vector libraries through these interfaces, ensuring that the abstraction layer of the API interface can handle the differences between different vector libraries and realizing the transparent access of the vector library. For example, in scientific research project management, the system can automatically extract key information (such as project name, person in charge, research content, etc.) from project documents and store it in the vector library. When teachers and students need to query relevant information about a certain project, the system can quickly find relevant text, images, and other data through multi-modal joint retrieval and display it to the user.

[0030] The extensible architecture module 130 supports multiple models and multiple inference frameworks, including submitting queries, the retriever component selecting relevant information, and the re-ranker refining the selection, for processing data from databases, web pages, or PDF files, and continuously optimizing the data-driven engine through proxy and plugin mechanisms, and ensuring the scalability of the platform through the knowledge base and intelligent agent workflow component construction; user-defined data types, indexing strategies, and query optimization plugins. Among them, through the proxy mechanism, transparent access and scheduling of the data-driven engine are realized, and through the plugin mechanism, the performance and functions of the data-driven engine are continuously optimized to adapt to new business requirements and data sources. For example, in the library management system, the system supports the storage and retrieval of multiple data types, such as book information, borrowing records, etc., allowing users to customize data types (such as text, image, audio, etc.), indexing types (such as full-text index, vector index, etc.), and query optimization strategies (such as caching strategies, query rewriting, etc.) through the graphical interface or API interface, automatically creating corresponding data structures according to user definitions, optimizing the storage and retrieval efficiency of book information, and the system integrates query optimization plugins, such as caching plugins, to improve the query speed of borrowing records.

[0031] The vertical domain expertise module 140 performs a large amount of data preprocessing and pre-trained models (such as multi-task learning frameworks) on data in the vertical domain of universities, outputs material completeness scores, missing item detection, and urgency classification information, adjusts parameter settings according to industry characteristics, and customizes and optimizes the interaction experience through role-based prompt words.

[0032] Among them, the system outputs material completeness scores, missing item detection, and urgency classification information. For example, the system can check the integrity, missingness, and urgency of the materials submitted by the user (such as reimbursement vouchers, application forms, etc.), and give the material completeness score respectively. This score can be a specific value (such as 0-100 points) or a level (such as "complete", "basically complete", "missing some materials", etc.). According to the score, it is decided whether to continue processing or prompt the user to supplement materials; identify the missing materials or information items, inform the user of what materials or information are missing, help the user quickly complete the materials, and improve the efficiency of business handling; classify the urgency of the business according to factors such as the business type submitted by the user and the current time (such as "urgent", "general", "non-urgent", etc.), help the system give priority to processing urgent business, and improve the efficiency of business processing. At the same time, it also provides reference information for business handling for users, and users can reasonably arrange the handling time according to the urgency.

[0033] Among them, the vertical domain expertise module 140 performs a large amount of data preprocessing on data in the vertical domain of universities. For example, it collects university industry-specific data sets, including relevant text and voice data for business such as scholarship applications and reimbursements, and performs cleaning, annotation, and structuring on the data for subsequent model training. Role-based prompt word customization, for example, is divided according to the user's organizational structure role (such as student, teacher, administrative staff, etc.), and different prompt words and interaction logics are customized. For example, for student users, the system can use more friendly and easy-to-understand prompt words and interaction methods.

[0034] Specifically, such as Figure 2As shown in the figure, the overall architecture of the AI ​​intelligent one-stop service system 100 for colleges and universities based on the intelligent agent platform includes seven levels: (1) Basic environment layer: The platform supports industry-specific cloud environments such as campus network and teaching cloud, and also supports private deployment and mainstream shared cloud environments such as Alibaba Cloud. (2) Model layer: The platform currently builds a text-to-SQL dedicated large model for SQL statements and a dedicated model for various industries such as colleges and governments. At the same time, the platform supports various current mainstream large models and various third-party open and closed source large models such as Tongyi Qianwen, Doubao, and KIMI. (3) Protocol layer: The platform supports various current mainstream models, interfaces, and database interfaces, including DSL, AgentFrame, Operators, Parallelization, etc. (4) Module layer: The platform includes a management framework module that supports various microservice applications, a RAGS knowledge base retrieval module, and an intelligent agent development and deployment module for building various intelligent agent applications. (5) Service layer: The platform provides various automated business process services based on intelligent question and answer, knowledge base retrieval services, API interface services, database retrieval services, and other large model training services. (6) Application layer: The platform currently implements application scenarios including intelligent question and answer, policy document summary, statistical report generation, AI automatic business process, intelligent analysis, etc. (7) Visual layer: The platform supports the function of converting analytical data into analytical reports, and can build various analytical report visualizations for users with different roles and permissions.

[0035] Preferably, in some embodiments, the SQL processing module includes: The requirements analysis unit is used to determine the target database type and the SQL statement types that need to be supported, as well as the natural language to SQL conversion requirements; A data collection unit, used to collect database SQL statement samples and natural language query samples and their corresponding SQL statements; The data preprocessing unit is used to standardize the collected SQL statements and perform natural language processing operations on natural language queries, including word segmentation and part-of-speech tagging; The model selection and training unit is used to select a suitable machine learning or deep learning model and use the preprocessed data to train the model so that it can understand natural language and generate corresponding SQL statements; The model optimization unit is used to evaluate the model, optimize the model according to the evaluation results, and integrate the trained model into the intelligent agent platform.

[0036] like Figure 3 As shown, the specific construction process of the SQL processing module 110 includes the following steps: S111: Demand Analysis: Determine the target database type (such as MySQL, PostgreSQL, Oracle, etc.); determine the types of SQL statements to be supported (such as SELECT, INSERT, UPDATE, DELETE, etc.); determine the conversion requirements from natural language to SQL, including complexity, flexibility, etc.

[0037] S112: Data collection Collect SQL statement samples of each mainstream database; collect natural language query samples and their corresponding SQL statements.

[0038] S113: Data preprocessing Standardize the collected SQL statements to ensure uniform format; perform natural language processing operations such as word segmentation and part-of-speech tagging on natural language queries.

[0039] S114: Model selection and training Select a suitable machine learning or deep learning model, such as Seq2Seq, Transformer, etc.; use the preprocessed data to train the model so that it can understand natural language and generate corresponding SQL statements.

[0040] S115: Model evaluation and optimization Evaluate the model and check its conversion effect on different types of SQL statements; optimize the model according to the evaluation results to improve the accuracy and efficiency of conversion.

[0041] S116: Integration and testing Integrate the trained model into the platform; test the platform to ensure that the natural language to SQL conversion function works properly.

[0042] S117: User interface design Design a user-friendly interface that allows users to input natural language queries; design a result display interface to show the converted SQL statements.

[0043] S118: Deployment and maintenance Deploy the platform to the production environment; regularly maintain and update the model to adapt to new SQL statements and natural language queries.

[0044] The platform constructed through the above steps needs to organize the existing SQL of each mainstream database. Through continuous training of SQL statements, the platform can understand the meanings of various SQL statements, which helps the platform better convert various natural languages into SQL statements, thereby facilitating college students to query their grades, course schedules, library collections, etc., and improving the query efficiency.

[0045] Furthermore, in some embodiments, the SQL processing module further includes an agent interaction layer. The agent interaction layer uses a Bi-LSTM+CRF hybrid model to implement business intent recognition, constructs a dynamic business topology graph, dynamically adjusts node weights according to real-time traffic, provides recommended entry points and confidence scores for users, and supports multi-turn dialogue management, capable of understanding the context based on user feedback.

[0046] Among them, constructing a dynamic business topology graph includes: Graph construction: According to the business process and the dependency relationship between nodes, construct an initial business topology graph, where nodes represent business steps or services, and edges represent the flow relationship between steps. Dynamic adjustment: Real-time monitor the traffic of each node, and dynamically adjust the node weights according to the traffic. The weights reflect the importance or popularity of the nodes and are used for sorting the recommended entry points. Confidence score: Calculate a confidence score for each recommended entry point, based on historical data, user behavior, and model prediction. The confidence score is used to guide users to select the most suitable business entry point.

[0047] In this embodiment, by dynamically adjusting the business topology graph to adapt to real-time demand changes and introducing confidence scores, the recommendation accuracy can be improved. For example, during peak periods, the system automatically increases the weight of the "Query account balance" node and displays it first in the recommended entry points.

[0048] Among them, multi-turn dialogue management includes: Context understanding: By maintaining the dialogue state, track the information and intent of the user in the previous turn of the dialogue; use the output of the Bi-LSTM+CRF model and the dialogue history for context understanding. Dialogue flow management: According to user feedback and the current dialogue state, determine the next dialogue flow or business step, support dialogue interruption, backtracking, and jumping, to improve the flexibility and robustness of the dialogue.

[0049] In this embodiment, by combining context information and user feedback, the coherence and accuracy of multi-turn dialogue are achieved, supporting complex dialogue flow management, and improving user satisfaction. For example, User: "What should I do if I forget my password?" System: "You can reset your password with a mobile verification code. What is your mobile phone number?" After the user provides the mobile phone number, the system continues to guide the user to complete the password reset process.

[0050] Among them, the Bi-LSTM+CRF hybrid model is constructed by combining a bidirectional long short-term memory network (Bi-LSTM) and a conditional random field (CRF). Bi-LSTM (bidirectional long short-term memory network): It is used to capture the temporal dependencies in the user input statement, processes sequence data simultaneously from both the forward and backward directions, and can better understand the context. CRF (conditional random field): Based on the output of Bi-LSTM, it further considers the dependencies between labels, globally optimizes the sequence labeling problem, and improves the accuracy of intent recognition. The specific components of this hybrid model include: An input layer, which is used to receive text data and convert each character or word into a word vector representation; A word embedding layer, which is used to map the word vectors into a high-dimensional space to better capture the semantic relationships between words; The Bi-LSTM layer, which consists of a forward LSTM and a backward LSTM, can capture both the forward and backward information in the sequence simultaneously, and the output at each time step contains the context information of the current position. Its output is the hidden state of each word; A fully connected layer, which is used to map the output of the Bi-LSTM layer into the label space and output the scores of each position corresponding to each label; The CRF layer, which is used to globally optimize the output of the fully connected layer according to the dependencies between labels. The output of Bi-LSTM is used as the input of the CRF. By calculating the transition matrix and the emission matrix, the optimal label sequence (i.e., the business intent) is output.

[0051] The process of using the above Bi-LSTM+CRF hybrid model to achieve business intent recognition mainly includes the following steps: (1) Forward propagation: Convert the input sequence into a word vector sequence through the word embedding layer; Input the word vector sequence into the Bi-LSTM layer to obtain the hidden state at each time step; Input the hidden state into the fully connected layer to obtain the scores of each position corresponding to each label; Use the CRF layer to consider the dependencies between labels and calculate the optimal label sequence.

[0052] (2) Define the loss function: Use the negative log-likelihood loss function to calculate the difference between the predicted label sequence and the true label sequence. The negative log-likelihood loss function used is:

[0053] Among them, y is the true label sequence, and x is the input sequence (the hidden state sequence after being processed by the word embedding layer and the Bi-LSTM layer). P(y|x) is the probability that the output sequence is y given the input sequence x.

[0054] The loss function consists of two parts, the emission score and the transition score, corresponding to the output of the fully connected layer and the transition matrix of the CRF layer respectively. Specifically, this probability can be calculated through the transition matrix and the emission matrix of the CRF layer:

[0055] where y′ is one of all possible label sequences. Given the input sequence x, the model may output multiple different label sequences, and y′ is one of these possible label sequences. When calculating the loss function, it is necessary to traverse all possible label sequences y′ to calculate the probability that the output sequence is y′ given the input sequence x, and normalize these probabilities. represents the set of all possible label sequences given the input sequence x; represents the exponential sum of calculating the scores (scores obtained through the transition matrix and the emission matrix of the CRF layer) for all possible label sequences y′ (belonging to the set ).

[0056] (3) Backpropagation: By calculating the gradient of the loss function, update the model parameters, including the parameters of the word embedding layer, the Bi-LSTM layer, the fully connected layer, and the CRF layer.

[0057] (4) Decoding: In the test or inference stage, use the Viterbi algorithm to decode the optimal label sequence on the CRF layer.

[0058] Let be the label sequence to be decoded, and be the corresponding input sequence (the hidden state sequence after being processed by the word embedding layer and the Bi-LSTM layer).

[0059] The traditional CRF decoding formula is expressed as:

[0060] where A is the transition matrix and P is the emission matrix.

[0061] The present invention improves the traditional CRF decoding formula, introduces dynamic weight adjustment based on context information, and defines a weight adjustment factor , and this factor dynamically adjusts the label according to the context information at the current decoding position i The weight. Therefore, the improved decoding formula is expressed as:

[0062] Among them, the weight adjustment factor is calculated based on the Bi-LSTM hidden state at the current position and the output of the fully connected layer For example, in some optional embodiments of the present invention, a simple feedforward neural network (FFNN) is used to calculate and to calculate :

[0063] Among them, is an activation function (such as the sigmoid function), which is used to map the output to the interval (0, 1) as the weight adjustment factor.

[0064] During the training process, the present invention needs to optimize the parameters of the Bi-LSTM layer, the fully connected layer, and the weight adjustment factor calculation network (such as FFNN) simultaneously. This can be achieved through the backpropagation algorithm, where the loss function is still the negative log-likelihood loss function, but the influence of the weight adjustment factor needs to be considered when calculating the gradient. Specifically, the training process of the Bi-LSTM+CRF hybrid model includes: Data preparation: Collect and label the training data (including user queries, business intent labels, etc.), and convert each character or word into a word vector representation. Model initialization: Initialize the model parameters, including the parameters of the word embedding layer, Bi-LSTM layer, fully connected layer, and CRF layer. Forward propagation: Input the training data into the model for forward propagation and calculate the loss function. Backward propagation: Calculate the gradient of the loss function and update the model parameters. Iterative training: Repeat the forward propagation and backward propagation processes until a predetermined number of training rounds is reached or the model performance no longer improves significantly. Model evaluation: Evaluate the model performance on the validation set or test set, such as accuracy, recall rate, F1 value, etc. Model tuning: Adjust the model parameters or network structure according to the evaluation results to improve the model performance. Through the above steps, a Bi-LSTM+CRF hybrid model with excellent performance can be trained to solve the sequence labeling problem. In practical applications, this model can be widely used in the field of natural language processing, such as named entity recognition, part-of-speech tagging, text classification, etc.

[0065] By introducing a dynamic weight adjustment mechanism based on context information, the CRF decoding process can better adapt to a specific context environment, thereby improving the accuracy and robustness of label sequence decoding. Especially when dealing with complex or ambiguous input sequences, this mechanism can significantly improve the performance of the model.

[0066] In this embodiment, by combining the time series modeling ability of Bi-LSTM and the global optimization ability of CRF, the accuracy of intent recognition is improved, and the model is dynamically updated to adapt to the changes in new services or user behaviors. For example, when the user asks, "I want to query my account balance." The system uses the Bi-LSTM+CRF model to recognize that the business intent is "query account balance".

[0067] Furthermore, the SQL processing module further includes an NLP engine, a rule engine, and a speech engine, which are respectively used for intent recognition, rule inference, and speech processing; Among them, the NLP engine integrates intent recognition and multi-turn dialogue management, and adopts a sequence-to-sequence (Seq2Seq) model, such as Transformer or LSTM+Attention, to convert the natural language input by the user into an SQL query statement. The model selection and training unit uses the machine learning or deep learning algorithm and uses a large number of labeled natural language-SQL pairs as training data to achieve the conversion from natural language to SQL statement, supporting college users to perform data query operations; The rule engine dynamically loads the SWRL rules provided by each business department to support the automated management of business processes; uses SWRL (Semantic Web Rule Language) to define business rules. The rule engine dynamically loads the SWRL rules provided by each business department, makes inferences and decisions according to the rules, dynamically loads the rules, and adapts to business changes. Supports complex business logics and rule inferences. Application example: The business department defines a rule: "If the user's account balance is less than 100 yuan, then send a reminder notice." The rule engine loads and executes this rule. When the user's account balance is less than 100 yuan, a reminder notice is automatically sent.

[0068] The speech engine supports mixed Chinese and English speech processing, uses the Conformer model for speech recognition, and is fine-tuned on the college scenario corpus, while providing multiple speech synthesis modes. Among them, Conformer combines the advantages of Transformer and LSTM and has better recognition performance. Fine-tuning the model on the college scenario corpus improves the recognition accuracy. Providing multiple speech synthesis modes, such as standard speech, emotional speech, etc., meets the requirements in different scenarios. Supports mixed Chinese and English speech processing and adapts to multilingual scenarios. Adopting the advanced Conformer model improves the recognition accuracy. Providing multiple speech synthesis modes enhances the user experience. For example: The user inputs by voice: "I want to query my English score." The system recognizes and converts it into text: "I want to query my English score." The system replies in standard speech: "Your English score is 85 points." Furthermore, the SQL processing module further includes a data service layer, which is connected to multiple business databases in real time, constructs a cross-system data federation query interface, and integrates a multi-modal feature extractor; the data service layer is connected to multiple business databases such as teaching affairs, finance, and student affairs in real time, constructs a cross-system data federation query interface, and supports the storage, processing, and analysis of large-scale data. By integrating a multi-modal feature extractor, it can perform feature extraction and analysis on various types of data such as text, images, and tables, providing rich data support for upper-layer applications.

[0069] Furthermore, the vector library management module 120 includes: A document acquisition unit, which is used to obtain the API interface documents of each vector library from a preselected list of vector libraries; A platform API interface unit, which is used to provide unified API interfaces, enabling the platform to interact with different vector libraries through these interfaces, and the abstraction layer of the API interfaces can handle the differences between different vector libraries.

[0070] As Figure 4 shown, the specific construction process of the vector library management module 120 includes the following steps: S121: Research and select vector libraries Determine the list of vector libraries to be supported, including Milvus, Pinecone, Vespa, Weaviate, Vald, GSI, and Qdrant, etc. Research the characteristics, performance, and API interfaces of each vector library.

[0071] S122: Obtain API interface documents: Obtain the API interface documents of each vector library to understand how to use the API for data operations; for example, for Pinecone, the usage methods of API interfaces such as initialization, creating an index, retrieving the index list, and inserting data can be obtained through the official documentation.

[0072] S123: SDK development According to the API interface documents, develop or integrate an SDK (Software Development Kit) to facilitate the use of these APIs in the platform. For example, Vald provides two types of interfaces, gRPC and REST API, and it is recommended to use gRPC for better performance.

[0073] S124: Design of platform API interfaces Design unified API interfaces, enabling the platform to interact with different vector libraries through these interfaces. Ensure that the abstraction layer of the API interfaces can handle the differences between different vector libraries.

[0074] S125: Platform SDK development The SDK of the development platform encapsulates the API interfaces of different vector libraries and provides a unified interface for developers to use. For example, the usage tutorial of Weaviate shows how to import data into Weaviate and how to perform similarity searches.

[0075] S126: Testing and Validation Test the API interfaces of each vector library to ensure that they can work properly on the platform.

[0076] Verify whether the SDK of the platform can interact correctly with each vector library.

[0077] S127: Documentation and Sample Code Provide detailed API interface documentation and sample code to help developers understand and use the API interfaces of the platform. For example, Milvus provides the Python SDK PyMilvus, allowing developers to interact with Milvus in a Python environment.

[0078] S128: User Feedback and Iteration Collect user feedback and optimize the API interfaces and SDK according to the feedback; continuously iterate and update to support new versions and new features of vector libraries.

[0079] The vector library management module 120 constructed through the above steps in this embodiment, when a certain university is conducting scientific research project management and academic achievement display, needs to process a large amount of multi-modal data such as text, images, and audio. The vector library management module 120 can be docked and adapted with various mainstream open-source and closed-source vector libraries, support dynamic data synchronization and multi-modal joint retrieval, and can improve data processing efficiency.

[0080] Furthermore, the extensible architecture module 130 includes: a plugin architecture unit for integrating third-party developer custom plugins and providing necessary tools and libraries; a data type and index specification definition unit for defining the specifications of new data types and customized indexes; wherein, the specifications of the new data types include data structures and operation interfaces, and the specifications of the customized indexes include index structures and index methods; a query optimization unit for implementing a query optimizer to support user-defined query optimization strategies; a plugin management tool for facilitating users to install, configure, and uninstall plugins; a testing and validation unit for testing the plugin framework and new functions and verifying whether the new data types and index types meet the data storage requirements of specific domains.

[0081] As Figure 5 shown, the specific construction process of the extensible architecture module 130 includes the following steps: S131: Requirement Analysis and Planning Analyze user requirements to determine the goals and scope of the plug-in mechanism; plan the supported data types, index types, and query optimization strategies.

[0082] S132: Design the plug-in architecture: Design a flexible plug-in architecture to ensure that third-party developers can easily integrate custom plug-ins; define the plug-in interfaces and APIs to ensure compatibility with the platform.

[0083] S133: Develop the plug-in framework: Implement the plug-in framework and provide the necessary tools and libraries so that developers can build plug-ins on this basis; include the loading, unloading, execution, and management mechanisms of the plug-ins.

[0084] S134: Define the data types and index specifications: Define the specifications for new data types, including data structures and operation interfaces; define the specifications for customized indexes, including index structures and index methods.

[0085] S135: Implement the query optimization strategy: Implement a query optimizer to support user-defined query optimization strategies; allow users to customize query plans according to application scenarios.

[0086] S136: Develop the plug-in management tool: Develop a plug-in management tool to facilitate users to install, configure, and uninstall plug-ins; provide a user interface or command-line tool to simplify the management process.

[0087] S137: Testing and verification: Test the plug-in framework and new functions to ensure stability and performance.

[0088] Verify whether the new data types and index types meet the data storage requirements of specific domains.

[0089] S138: Documentation and examples: Provide detailed development documentation and example code to help third-party developers understand how to develop plug-ins.

[0090] Include guidelines on how to define new data types, indexes, and query optimization strategies.

[0091] S139: Release and feedback collection: Release the plug-in mechanism and related tools; collect user feedback and perform iterative optimization based on the feedback.

[0092] S1310: Continuous support and updates: Provide continuous technical support and updates to adapt to new technological developments and user needs.

[0093] The scalable architecture module 130 constructed through the above steps in this embodiment supports multiple models and multiple inference frameworks, and allows users to customize data types, indexing strategies, and query optimization plugins, achieving the scalability of the system and facilitating the future integration of new models, algorithms, and data sources.

[0094] Furthermore, the vertical domain expertise module 140 includes: A data collection and preprocessing unit, which is used to collect industry-specific datasets of higher education institutions, including text and voice data, and perform data preprocessing, including data cleaning, annotation, and structuring; A model selection and training unit, which is used to process materials through a multi-task learning framework, select appropriate machine learning or deep learning algorithms, train the model using the preprocessed industry-specific datasets, and adjust parameters according to the model performance to achieve multi-modal material pre-review; A role division unit, which is used to customize different prompt words and interaction logics according to the user organization structure role division, and simulate the tone, intonation, and speech rate of users in the higher education industry; A user feedback unit, which is used to collect user feedback and perform iterative optimization based on the feedback.

[0095] Furthermore, the vertical domain expertise module 140 also includes a natural language processing model and a speech recognition model, which are respectively used to process text and voice data; The deep learning algorithm of the speech recognition model includes a Transformer model for processing long sequence data and capturing global dependencies, and provides a rich industry knowledge base and a question-answering system to support users in obtaining industry knowledge and answering questions through natural language queries.

[0096] As Figure 6 shown, the specific construction process of the vertical domain expertise module 140 includes the following steps: S141: Requirement research and analysis: Cooperate with higher education institutions and government departments to understand their specific needs and pain points; collect industry-specific datasets, including text, voice, and other data.

[0097] S142: Data preprocessing: Design a data preprocessing process according to the actual database structure of the user; clean, annotate, and structure the data to meet the needs of model training.

[0098] S143: Model design and selection: Design a model architecture suitable for industry characteristics.

[0099] Select appropriate machine learning or deep learning algorithms, such as natural language processing models, speech recognition models, etc. Specifically: In view of the characteristics of the vertical domain of universities, the present invention designs a hybrid model architecture that combines the advantages of deep learning and knowledge graphs. This model architecture mainly includes the following parts: Deep learning basic model: (1) Text processing sub-model: Adopt a self-attention mechanism model based on Transformer (such as BERT or RoBERTa) to process and analyze text data. The Transformer model can capture long-distance dependencies through the self-attention mechanism and is very suitable for processing long text data in universities, such as policy documents, academic papers, etc. (2) Speech processing sub-model: Adopt the Conformer model, which combines the advantages of Transformer and convolutional neural network (CNN) and can efficiently process speech signals, especially performing well in processing speech data in the university environment with complex background noise.

[0100] Knowledge graph fusion module: Construct a knowledge graph for the university domain, including entities such as students, teachers, courses, scientific research projects, etc. and their relationships. Combine the knowledge graph with the deep learning model to enhance the model's understanding and reasoning ability of domain knowledge. For example, when processing a student's course selection request, the model can use the course dependency relationship (such as prerequisite courses) in the knowledge graph to provide more accurate suggestions.

[0101] After selecting the above model architecture, the present invention further selects appropriate machine learning or deep learning algorithms to train these models. The following are the specific algorithms for text and speech processing: Specifically, for the text processing sub-model, the following text processing algorithms are used: Fine-tune on the BERT or RoBERTa basic model to adapt to the data in the university domain. Taking the BERT basic model as an example: Parameter definition: Word embedding dimension ( ): The dimension of the word vector in the model, usually set to 768 or 1024. Number of heads (h): The number of heads in the multi-head self-attention mechanism, usually set to 8 or 12. Feed-forward neural network dimension (dff): The hidden layer dimension in the feed-forward neural network, usually set to 3072 or 4096. Number of layers (L): The number of layers of the Transformer encoder, usually set to 12 or 24.

[0102] Input representation: Convert the input text into a sequence of word vectors and add position encoding and segment encoding. The formula is as follows: Input = WordEmbeddings + PositionEmbeddings + SegmentEmbeddings Among them, WordEmbeddings is the word vector sequence, PositionEmbeddings is the position encoding, and SegmentEmbeddings is the segment encoding.

[0103] Transformer Encoder: For each word vector, it is processed through the multi-head self-attention mechanism and the feed-forward neural network. The calculation formula is as follows:

[0104] Among them, Q, K, and V are the query, key, and value matrices respectively, is the dimension of the key vector.

[0105] Feed-forward Neural Network: ) Among them, and are the weight matrices, and are the bias terms.

[0106] For the speech processing sub-model, the Conformer model is used with the speech processing algorithm. Taking the self-attention mechanism in Conformer as an example: Parameter Definition: Convolution Kernel Size: Used for convolution operations in the convolutional layer, usually set to 3 or 5. Number of Convolution Layers: The number of convolutional layers in Conformer, adjusted according to the model complexity. Number of Attention Heads: The same as Transformer, usually set to 8 or 12.

[0107] Position Encoding: Similar to Transformer, Conformer also uses position encoding to capture the position information in the sequence, and uses sine and cosine functions for position encoding.

[0108] Self-attention Mechanism: The same as Transformer, but Conformer adds a convolutional layer in the self-attention mechanism to capture local features. The calculation formula of the self-attention mechanism is the same as that in the above Transformer.

[0109] Convolutional Layer: After the self-attention mechanism, Conformer applies a series of depthwise separable convolutional layers to further extract features. The calculation formula of the convolutional layer is:

[0110] Among them, DepthwiseConv is the depthwise separable convolution, PointwiseConv is the pointwise convolution, and BN is the batch normalization.

[0111] The knowledge graph fusion module uses a graph neural network (GNN) to fuse knowledge graph information. Specifically, taking the graph convolutional network GCN as an example: Parameter definition: Learning rate (η): The step size for model training, adjusted according to the training situation. Weight decay (λ): The regularization term coefficient used to prevent overfitting. Number of layers (LGCN): The number of layers of GCN, usually set to 1 or 2.

[0112] Node feature update: For each node in the knowledge graph, its feature representation is updated by aggregating the features of its neighbor nodes. The calculation formula is as follows: where A is the adjacency matrix, D is the degree matrix, is the node feature matrix of the l-th layer, is the weight matrix of the l-th layer, and σ is the activation function (such as ReLU).

[0113] S144: Targeted training and parameter adjustment: Train the model using the industry user's proprietary dataset; adjust the parameters according to the model's performance to optimize the model's performance.

[0114] S145: Role division and prompt word customization: Customize different prompt words and interaction logics according to the role division of the user's organizational structure; simulate the tone, intonation, and speech rate of industry users to improve the user experience.

[0115] S146: Model training and verification: Conduct a large number of model trainings on the preprocessed data; verify the accuracy and adaptability of the model to ensure that it meets the needs of industry users.

[0116] S147: System integration and testing: Integrate the trained model into the platform; conduct comprehensive system testing to ensure the stability and reliability of the system.

[0117] S148: User feedback and iterative optimization: Collect user feedback and evaluate the actual application effect of the model; perform iterative optimization based on the feedback to improve the accuracy of the model and the user experience.

[0118] S149: Deployment and monitoring: Deploy the optimized model to the production environment.

[0119] Implement monitoring to ensure the smooth operation of the system and respond promptly to possible problems.

[0120] S1410: Documentation and training: Provide detailed system documentation and user manuals.

[0121] Train users to ensure that they can make full use of the platform functions.

[0122] In the embodiment of the present invention, the vertical domain professional knowledge module 140 is constructed through the above steps. A large amount of preprocessing and pre-training are carried out on the data in the vertical domain of colleges and universities, improving the system's understanding and processing capabilities for specific business scenarios. Thereby, the material completeness score, missing item detection, and urgency classification information are output, and the parameter settings are adjusted according to industry characteristics, and the interaction experience is customized and optimized through role-based prompt words.

[0123] In summary, the present invention realizes the intelligent processing and analysis of multi-source data by integrating SQL processing capabilities, vector library management, scalable architecture, and vertical domain professional knowledge applications, significantly improving the business handling efficiency and user experience of the online service hall of colleges and universities, and contributing to the digital transformation and intelligent upgrade of the education industry.

[0124] The above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the embodiments, those of ordinary skill in the art should understand that: the specific implementation manners of the present invention can still be modified or equivalently replaced, and any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered by the scope of the claims of the present invention.

Claims

1. An AI intelligent one-stop service system for colleges and universities based on an intelligent agent platform, characterized in that: include: SQL processing module, which is used to organize SQL statements of existing mainstream databases and convert the natural language input by users into SQL statements of the target database through pre-trained machine learning or deep learning models; The vector library management module is compatible with various mainstream open and closed source vector libraries, interacts with different vector libraries through a unified API interface, and supports dynamic data synchronization and multi-modal joint retrieval; Extensible architecture module, supporting multiple models and multiple reasoning frameworks, including submitting queries, selecting relevant information from retriever components, and re-selecting refinements, for processing data from databases, web pages, or PDF files, and continuously optimizing the data-driven engine through agents and plug-in mechanisms, and ensuring platform scalability through the construction of knowledge base and intelligent agent workflow components; user-defined data types, indexing strategies, and query optimization plug-ins; The vertical field professional knowledge module performs data preprocessing and pre-training models for vertical field data of universities, outputs material completeness scores, missing item detection and urgency classification information, adjusts parameter settings according to industry characteristics, and optimizes the interactive experience through customized role-based prompts.

2. The system according to claim 1, characterized in that The SQL processing module includes: The requirements analysis unit is used to determine the target database type and the SQL statement types that need to be supported, as well as the natural language to SQL conversion requirements; A data collection unit, used to collect database SQL statement samples and natural language query samples and their corresponding SQL statements; The data preprocessing unit is used to standardize the collected SQL statements and perform natural language processing operations on natural language queries, including word segmentation and part-of-speech tagging; The model selection and training unit is used to select a machine learning or deep learning model and use the preprocessed data to train the model so that it can understand natural language and generate corresponding SQL statements; The model optimization unit is used to evaluate the model, optimize the model according to the evaluation results, and integrate the trained model into the intelligent agent platform.

3. The system according to claim 1, characterized in that The vector library management module includes: A document acquisition unit, used for acquiring the API interface document of each vector library from the pre-selected vector library list; The platform API interface unit is used to provide a unified API interface so that the platform can interact with different vector libraries through these interfaces, and the abstract layer of the API interface can handle the differences between different vector libraries.

4. The system according to claim 1, characterized in that The extensible architecture module includes: The plug-in architecture unit is used to integrate third-party developers’ custom plug-ins and provide necessary tools and libraries; A data type and index specification definition unit, used to define specifications for new data types and customized index specifications; wherein the specifications for new data types include data structures and operation interfaces, and the specifications for customized indexes include index structures and index methods; The query optimization unit is used to implement the query optimizer and support user-defined query optimization strategies; Plugin management tool, used to facilitate users to install, configure and uninstall plugins; The testing and verification unit is used to test the plug-in framework and new functions, and to verify whether the new data types and index types meet the data storage requirements of specific fields.

5. The system according to claim 1, characterized in that The vertical domain expertise modules include: Data collection and preprocessing unit, used to collect university industry-specific data sets including text and voice data, and perform data preprocessing, including data cleaning, labeling and structuring; Model selection and training unit, which is used to process materials through a multi-task learning framework, select appropriate machine learning or deep learning algorithms, train models using proprietary data sets after data preprocessing, and adjust parameters based on model performance to achieve multimodal material pre-review; The role division unit is used to customize different prompt words and interaction logic according to the user's organizational structure role division, and simulate the tone, intonation and speed of university industry users; The user feedback unit is used to collect user feedback and perform iterative optimization based on the feedback.

6. The system according to claim 2, characterized in that The SQL processing module also includes an agent interaction layer, which uses a Bi-LSTM+CRF hybrid model to realize business intent recognition, build a dynamic business topology map, dynamically adjust node weights according to real-time access volume, provide users with recommended entrances and confidence scores, and support multi-round dialogue management, and can perform context understanding based on user feedback.

7. The system according to claim 6, characterized in that in, The Bi-LSTM+CRF hybrid model is constructed by combining a bidirectional long short-term memory network and a conditional random field, including: The input layer receives text data and converts each character or word into a word vector representation. The word embedding layer is used to map word vectors to a high-dimensional space to capture the semantic relationship between words; The Bi-LSTM layer, which consists of a forward LSTM and a backward LSTM, can capture both the forward and backward information in the sequence and the output of each time step contains the contextual information of the current position; The fully connected layer is used to map the output of the Bi-LSTM layer to the label space and output the score of each label corresponding to each position; The CRF layer is used to globally optimize the output of the fully connected layer according to the dependencies between labels, and output the optimal label sequence by calculating the transfer matrix and the emission matrix.

8. The system according to claim 6, characterized in that in, The SQL processing module also includes an NLP engine, a rule engine and a speech engine, which are used for intent recognition, rule reasoning and speech processing respectively; The NLP engine integrates intent recognition and multi-round dialogue management. The model selection and training unit trains SQL statements through the machine learning or deep learning algorithm to achieve the conversion from natural language to SQL statements, supporting university users to perform data query operations. The rule engine dynamically loads SWRL rules provided by each business department to support automated management of business processes; The speech engine supports mixed speech processing in Chinese and English, uses the Conformer model for speech recognition, and is fine-tuned on university scenario corpus, while providing multiple speech synthesis modes.

9. The system according to claim 7, characterized in that The SQL processing module also includes a data service layer, which connects to multiple business databases in real time, builds a cross-system data federation query interface, and integrates a multimodal feature extractor; the data service layer connects to multiple business databases of academic affairs, finance, and student affairs in real time, builds a cross-system data federation query interface, and supports the storage, processing and analysis of large-scale data, integrates a multimodal feature extractor, and can extract and analyze features of various types of data such as text, images, and tables, providing rich data support for upper-level applications.

10. The system according to claim 5, characterized in that The vertical domain expertise module also includes a natural language processing model and a speech recognition model, which are used to process text and speech data respectively; The deep learning algorithm of the speech recognition model includes a Transformer model for processing long sequence data and capturing global dependencies, and provides a rich industry knowledge base and question-answering system to support users in acquiring industry knowledge and answering questions through natural language queries.

Images