An integrated archival information storage management system and method based on artificial intelligence

By leveraging artificial intelligence technology, the automated management and precise evaluation of user profile information have been achieved, solving the problems of time-consuming and labor-intensive management and untimely data synchronization in traditional management methods, thereby improving management efficiency and analytical accuracy.

CN118733859BActive Publication Date: 2024-11-22SHENZHEN SHIXINDA TECH CO LTD
View PDF 6 Cites 0 Cited by

Patent Information

Application Number
CN202411206817.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-30
Publication Date
2024-11-22
Estimated Expiration
2044-08-30

AI Technical Summary

Technical Problem

Traditional user profile management methods are time-consuming and labor-intensive, data synchronization is not timely, approval processes are prone to errors, task monitoring and feedback are lacking and inefficient, and personalized performance analysis reports cannot be generated.

Method used

An integrated archival information storage management system based on artificial intelligence is adopted. Data is entered and preprocessed through convolutional neural networks, tasks are assigned and monitored through an automated workflow module, information is confirmed through an intelligent feedback and communication module, and performance analysis reports are generated using decision tree and random forest algorithms.

Benefits of technology

It enables timely updates and consistency of file information, reduces manual data entry and synchronization workload, improves the efficiency and accuracy of the approval process, and provides accurate user attention assessment and personalized performance analysis.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118733859B_ABST
    Figure CN118733859B_ABST
Patent Text Reader

Abstract

The present invention discloses an integrated management system and method for archival information storage based on artificial intelligence, and belongs to the field of artificial intelligence technology. The present invention synchronizes data in real time through an interface, pre-processes scanned paper archives, and extracts key information and metadata; the administrator pre-configures the workflow, automatically allocates tasks for archival information approval, monitors task execution, generates a process analysis report, and integrates with other information system tasks; whenever a user needs to confirm archival information, the user archival information is sent to the corresponding user for confirmation, and feedback information is sent to the administrator when the user confirms an error; after each user confirms the score in an exam, an interaction is established, and a user score analysis report is generated using a decision tree algorithm; the user score analysis report is combined with the key information and metadata in the user archival information for analysis, and the comprehensive attention needs of different users and the attention needs of the user in different aspects are generated.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an integrated archive information storage management system and method based on artificial intelligence. Background Art

[0002] In modern education management, the management and analysis of user profile information is of great significance to improving the quality and efficiency of education. The traditional user profile management method mainly relies on paper documents and simple electronic spreadsheets, which has the following main problems:

[0003] Traditional paper archives need to be manually entered into electronic systems, which is time-consuming, labor-intensive and prone to errors. In addition, data synchronization between different systems is not timely, resulting in delayed information updates. The approval process for archival information usually involves multiple departments and personnel, and manual management of these processes is prone to omissions and delays. Without an effective task monitoring and reminder mechanism, important tasks are easily overdue or uncompleted. When users are required to confirm archival information, traditional methods usually rely on telephone, email and other means, which have low feedback efficiency and are difficult to track, and cannot quickly generate and share personalized performance analysis reports. The existing system lacks in-depth analysis of user archival information and performance data, and cannot provide accurate attention assessments based on the user's comprehensive performance. Summary of the invention

[0004] The purpose of the present invention is to provide an integrated archival information storage management system and method based on artificial intelligence to solve the problems raised in the above-mentioned background technology.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0006] An integrated archival information storage management system based on artificial intelligence, comprising:

[0007] Data entry module: synchronizes data with the academic affairs management system, HR system and email in real time through interfaces, pre-processes scanned paper archives, uses convolutional neural networks to convert text content in images into editable text information, performs syntax parsing and semantic analysis on the entered text information, and extracts key information and metadata;

[0008] Automated workflow module: Administrators pre-configure workflows, automatically assign tasks for file information approval, monitor task execution, and automatically remind relevant personnel of tasks that are about to expire or have expired; collect statistics on the execution data of various workflows, generate process analysis reports, and provide API interfaces to achieve task integration with other information systems;

[0009] Intelligent feedback communication module: Whenever the user needs to confirm the file information, the user file information will be sent to the corresponding user for confirmation. If the user's confirmation is incorrect, the feedback information will be sent to the administrator, and the automated workflow module will be re-executed for the information to confirm the correct file information and update it to other systems. After each user confirms the test score, the interaction between parents, teachers and users is established, and the decision tree algorithm is used to generate a user score analysis report;

[0010] User attention need calculation module: Use the random forest algorithm to combine and analyze the user's performance analysis report with the key information and metadata in the user's profile information, generate the comprehensive attention needs of different users and the user's attention needs in different aspects, and send them to the corresponding class teacher.

[0011] Preferably, the data entry module uses Gaussian filtering to remove noise from the image, convert the color image into a grayscale image, simplify the information representation of the image, binarize the grayscale image, convert the image into a black and white image, and enhance the image to increase contrast and clarity;

[0012] Annotate the preprocessed image, mark the position of each user's profile information to be entered in the image and the corresponding characters; use VGG as the model structure of the convolutional neural network, the input layer accepts the preprocessed user profile image as input, set the convolution kernel size, number of convolution kernels, activation function, padding and step size of different convolution layers, set the pooling window size and step size of different maximum pooling layers; stack several convolution layers, keep the convolution kernel size, activation function and step size settings unchanged, and increase the number of convolution kernels; set the number of neurons and activation function of the fully connected layer; the number of neurons in the output layer is the same as the character set size of the input content, use Adam optimizer as the loss function for compilation and training, use the test data set to evaluate the trained model, and perform text recognition after evaluation.

[0013] Preferably, the data entry module identifies the sentence structure in the text through syntactic analysis, identifies the sentences containing the user's key information in the user profile information, wherein the user's key information is preset by the administrator; performs part-of-speech tagging on each word in the text, analyzes the dependency relationship between the key information words in the text, constructs a dependency tree structure of the user's key information, and extracts the key information;

[0014] The main purpose of sentence analysis is to identify the sentence structure in the user profile information and extract sentences containing key information, using syntactic analysis and semantic analysis methods in natural language processing technology.

[0015] In text preprocessing, the input text is decomposed into individual words, common words that are not helpful for analysis, such as "de" (of), "shi" (is), etc., are removed, and part-of-speech tags, such as nouns, verbs, adjectives, etc., are assigned to each word; syntactic analysis is performed to construct the dependency tree of the input sentence, identify the dependency relationships between words, construct the constituent tree of the input sentence, and identify the constituent structure of the sentence, such as the subject, predicate, object, etc. Semantic analysis is carried out to identify the named entities in the text, such as user names, student numbers, course names, dates, etc., and extract the relationships between entities from the text, such as the relationship between users and grades.

[0016] Extract the attribute information of the user profile from the text, extract the numerical information related to the user profile, and extract the relationships between the user profile information in the text as metadata.

[0017] The administrator defines the specific process of file information approval in the automated workflow module. The specific process content of the approval includes approval nodes, approval personnel, approval logic, approval time limit, and approval order.

[0018] When a new file information approval application is submitted, the system automatically triggers the corresponding approval process instance; according to the defined approval process, the system automatically assigns the approval task to the corresponding approval personnel according to the preset rules; the system monitors the execution status of the approval task in real time. The execution status includes the current processor of the task, approval progress, remaining time, task submission time, task type, approval history, and approval result; the system monitors the remaining time of the approval task according to the set approval time limit and triggers a reminder notice when approaching the time limit or when it has expired; if the task has expired and is not processed, the system will automatically send an emergency reminder notice to the relevant personnel; the system records the processing process and result of the approval task.

[0019] Preferably, the automated workflow module analyzes the collected execution data, and the obtained analysis results include the execution efficiency of the workflow, the processing duration of the approval task, and the approval passing rate. A process analysis report is generated according to the analysis results. The process analysis report includes a text description and a chart display.

[0020] Design an API interface for providing query and integration functions of execution data to other information systems. The interface includes functions for querying task status, obtaining approval history, and triggering approval processes.

[0021] Preferably, in the intelligent feedback communication module, when the user needs to confirm the profile information, the system triggers the profile information confirmation event according to preset conditions, and the preset conditions include the start of a new semester, change of student status, graduation or leaving school of the user, and change of school policy; the system sends the profile information that needs to be confirmed to the corresponding user according to the triggered profile information confirmation event; the user receives the confirmation task notification, logs in to the system, browses the profile information and confirms it;

[0022] If the user confirms that the file information is incorrect, the system provides a feedback mechanism, allowing the user to mark the errors and provide explanations or remarks; after receiving the user's feedback, the system automatically sends feedback to the administrator to remind the administrator to deal with the incorrect file information; after the administrator receives the feedback, the system automatically triggers the automated workflow module to re-process the information; in the automated workflow module, the administrator re-checks the file information, confirms the correct information and updates it; after confirming the correct file information, the system automatically synchronizes the updated information to other systems.

[0023] Preferably, the intelligent feedback communication module generates a performance analysis report using a decision tree algorithm based on the confirmed performance, wherein the performance analysis report includes a comparative analysis of the current and history performances in each subject, the current and history class rankings, the current and history grade rankings, and the current and history learning situations;

[0024] Information gain is used to select the split point in the decision tree and measure the information increment that a feature can bring when splitting on the data set. The information gain formula is as follows:

[0025] ;

[0026] Among them, D is the collected data set of the current user, A is the candidate split feature, including feature vectors extracted from the current and history subject scores, current and history class rankings, and current and history grade rankings. are all possible values ​​of feature A, is a subset of feature A whose value is v, is the entropy of the data set D;

[0027] in, is defined as follows:

[0028] ;

[0029] in, It is a collection of categories that classify the user's learning status. The categories include: excellent, good, medium, pass and fail. is the probability of category c in data set D;

[0030] The optimal splitting feature is selected based on information gain, and a subtree is recursively constructed. Information gain is recursively performed on the child nodes until the node purity is higher than the threshold or the maximum depth is reached. The decision tree model is trained using labeled data. Based on the model's prediction results, a comparative analysis of the current and historical learning situations is obtained, and a performance analysis report is generated together with the grades and rankings.

[0031] Preferably, the user attention need calculation module extracts the user's performance characteristics, behavior characteristics, background characteristics and health characteristics from the user performance analysis report and the user profile information to obtain a feature set; uses the known user's characteristics and attention need labels as training data to train a random forest model for predicting the user's comprehensive attention need and various aspects of attention need;

[0032] In the random forest model, each tree is constructed using information gain to select the best split feature. When constructing a single decision tree, for a given training data set, a subset is randomly sampled with replacement, and a feature subset is randomly selected from the feature set for splitting. The information gain is used to select the optimal split point and construct a decision tree. The final prediction result is obtained by voting among all decision trees.

[0033] When calculating the comprehensive attention, the user's performance characteristics, behavior characteristics, background characteristics and health characteristics are integrated to form a comprehensive feature vector; the comprehensive feature vector is input into the trained random forest model to predict each user's comprehensive attention needs and various aspects of attention needs.

[0034] The final prediction result is obtained by voting of all decision trees. Assume that there are N decision trees in the random forest. , , ..., ; For an input sample x, each tree Give a prediction ; The final prediction result is the category with the most votes among all trees’ predictions:

[0035]

[0036] Among them, mode represents the mode, and the predicted category with the highest frequency is output;

[0037] When integrating the user's performance characteristics, behavior characteristics, background characteristics, and health characteristics, each characteristic needs to be standardized. Z-score standardization is used for processing. , calculate its mean and standard deviation , the formula is as follows:

[0038]

[0039] in, It is a feature The standardized value of

[0040] Assign a weight to each feature, and set the weights of performance features, behavior features, background features, and health features to be , , and , the score feature vector is , the behavior feature vector is , the background feature vector is , the health feature vector is , sum up all kinds of features according to their weights to form a comprehensive feature vector X:

[0041] ;

[0042] in, , , and They are the standardized values ​​of the performance feature vector, behavior feature vector, background feature vector and health feature vector respectively.

[0043] An integrated management method for archival information storage based on artificial intelligence comprises the following steps:

[0044] S100, synchronize data with the academic management system, HR system and email in real time through interfaces;

[0045] S200, pre-processing the scanned paper files, using a convolutional neural network to convert text content in the image into editable text information;

[0046] S300, performing syntax parsing and semantic analysis on the input text information to extract key information and metadata;

[0047] S400, the administrator pre-configures the workflow, automatically assigns tasks for file information approval and monitors task execution, and automatically reminds relevant personnel of tasks that are about to expire or have expired;

[0048] S500, collects statistics on the execution data of various workflows, generates process analysis reports, provides API interfaces, and realizes task integration with other information systems;

[0049] S600, whenever the user needs to confirm the file information, the user file information is sent to the corresponding user for the user to confirm. If the user confirmation is incorrect, step S601 is executed, otherwise step S700 is executed;

[0050] S601, sending feedback information to the administrator, re-executing the automated workflow module for the information, confirming the correct archive information, and updating it to other systems;

[0051] S700, after each user confirms the test score, establish interaction between parents, teachers and users, and use decision tree algorithm to analyze and generate a user score analysis report;

[0052] S800. Use the random forest algorithm to combine and analyze the user's performance analysis report with the key information and metadata in the user's profile information, generate the comprehensive attention needs of different users and the attention needs of the users in different aspects, and send them to the corresponding class teachers.

[0053] Compared with the prior art, the beneficial effects achieved by the present invention are:

[0054] The present invention ensures timely updating and consistency of file information through real-time data synchronization with the educational management system, HR system and email interface, and reduces the workload of manual input and synchronization.

[0055] In the present invention, the administrator pre-configures the workflow, the system automatically assigns file information approval tasks, and monitors the execution status of tasks in real time, and can automatically remind relevant personnel of tasks that are about to expire or have expired, reducing omissions and delays in manual management.

[0056] The present invention combines the user performance analysis report with the key information and metadata in the archive information, and uses the random forest algorithm to generate the user's comprehensive attention needs and attention needs in different aspects, providing more accurate attention suggestions. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0058] Figure 1 It is a system structure diagram of an integrated management system for archival information storage based on artificial intelligence of the present invention;

[0059] Figure 2 It is a schematic diagram of the steps of an integrated management method for archival information storage based on artificial intelligence of the present invention. DETAILED DESCRIPTION

[0060] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0061] See also Figure 1 and Figure 2 , the present invention provides a technical solution:

[0062] According to one embodiment of the present invention, Figure 1 As shown in the system structure diagram of an integrated management system for archival information storage based on artificial intelligence, an integrated management system for archival information storage based on artificial intelligence includes:

[0063] Data entry module: synchronizes data with the academic affairs management system, HR system and email in real time through interfaces, pre-processes scanned paper archives, uses convolutional neural networks to convert text content in images into editable text information, performs syntax parsing and semantic analysis on the entered text information, and extracts key information and metadata;

[0064] Automated workflow module: Administrators pre-configure workflows, automatically assign tasks for file information approval, monitor task execution, and automatically remind relevant personnel of tasks that are about to expire or have expired; collect statistics on the execution data of various workflows, generate process analysis reports, and provide API interfaces to achieve task integration with other information systems;

[0065] Intelligent feedback communication module: Whenever the user needs to confirm the file information, the user file information will be sent to the corresponding user for confirmation. If the user's confirmation is incorrect, the feedback information will be sent to the administrator, and the automated workflow module will be re-executed for the information to confirm the correct file information and update it to other systems. After each user confirms the test score, the interaction between parents, teachers and users is established, and the decision tree algorithm is used to generate a user score analysis report;

[0066] User attention need calculation module: Use the random forest algorithm to combine and analyze the user's performance analysis report with the key information and metadata in the user's profile information, generate the comprehensive attention needs of different users and the user's attention needs in different aspects, and send them to the corresponding class teacher.

[0067] In this embodiment, the user represents a student.

[0068] The data entry module uses Gaussian filtering to remove noise from the image, converts the color image into a grayscale image, simplifies the information representation of the image, performs binarization processing on the grayscale image, converts the image into a black and white image, and enhances the image to increase contrast and clarity;

[0069] Annotate the preprocessed image, mark the position of each user's profile information to be entered in the image and the corresponding characters; use VGG as the model structure of the convolutional neural network, the input layer accepts the preprocessed user profile image as input, set the convolution kernel size, number of convolution kernels, activation function, padding and step size of different convolution layers, set the pooling window size and step size of different maximum pooling layers; stack several convolution layers, keep the convolution kernel size, activation function and step size settings unchanged, and increase the number of convolution kernels; set the number of neurons and activation function of the fully connected layer; the number of neurons in the output layer is the same as the character set size of the input content, use Adam optimizer as the loss function for compilation and training, use the test data set to evaluate the trained model, and perform text recognition after evaluation.

[0070] Extract attribute information of user profile from text, extract numerical information related to user profile, and extract the relationship between user profile information in text as metadata.

[0071] The administrator defines a specific process for file information approval in the automated workflow module, wherein the specific process for approval includes approval nodes, approval personnel, approval logic, approval time limit, and approval sequence;

[0072] When a new application for file information approval is submitted, the system automatically triggers the corresponding approval process instance; according to the defined approval process, the system automatically assigns the approval task to the corresponding approval personnel according to the preset rules; the system monitors the execution status of the approval task in real time, and the execution status includes the current task handler, approval progress, remaining time, task submission time, task type, approval history and approval result; the system monitors the remaining time of the approval task according to the set approval time limit, and triggers a reminder notification when the time limit is approaching or has expired; if the task has expired and has not been processed, the system will automatically send an emergency reminder notification to the relevant personnel; the system records the processing process and results of the approval task.

[0073] The automated workflow module analyzes the collected execution data, and the obtained analysis results include the execution efficiency of the workflow, the processing time of the approval task, and the approval pass rate, and generates a process analysis report based on the analysis results, and the process analysis report includes a text description and a chart display;

[0074] Design an API interface to provide query and integration functions for execution data to other information systems. The interface includes functions for querying task status, obtaining approval history, and triggering approval processes.

[0075] In the intelligent feedback communication module, when the user needs to confirm the profile information, the system triggers the profile information confirmation event according to the preset conditions, which include the start of a new semester, changes in student status, user graduation or leaving school, and changes in school policies; the system sends the profile information that needs to be confirmed to the corresponding user according to the triggered profile information confirmation event; the user receives the confirmation task notification, logs in to the system, browses the profile information and confirms it;

[0076] If the user confirms that the file information is incorrect, the system provides a feedback mechanism, allowing the user to mark the errors and provide explanations or remarks; after receiving the user's feedback, the system automatically sends feedback to the administrator to remind the administrator to deal with the incorrect file information; after the administrator receives the feedback, the system automatically triggers the automated workflow module to re-process the information; in the automated workflow module, the administrator re-checks the file information, confirms the correct information and updates it; after confirming the correct file information, the system automatically synchronizes the updated information to other systems.

[0077] The intelligent feedback communication module generates a performance analysis report based on the confirmed performance using a decision tree algorithm, wherein the performance analysis report includes a comparative analysis of the current and history subject performances, the current and history class rankings, the current and history grade rankings, and the current and history learning situations;

[0078] Information gain is used to select the split point in the decision tree and measure the information increment that a feature can bring when splitting on the data set. The information gain formula is as follows:

[0079] ;

[0080] Among them, D is the collected data set of the current user, A is the candidate split feature, including feature vectors extracted from the current and history subject scores, current and history class rankings, and current and history grade rankings. are all possible values ​​of feature A, is a subset of feature A whose value is v, is the entropy of the data set D;

[0081] in, is defined as follows:

[0082] ;

[0083] in, It is a collection of categories that classify the user's learning status. The categories include: excellent, good, medium, pass and fail. is the probability of category c in data set D;

[0084] The optimal splitting feature is selected based on information gain, and a subtree is recursively constructed. Information gain is recursively performed on the child nodes until the node purity is higher than the threshold or the maximum depth is reached. The decision tree model is trained using labeled data. Based on the model's prediction results, a comparative analysis of the current and historical learning situations is obtained, and a performance analysis report is generated together with the grades and rankings.

[0085] We have the following user data: current math score: 85, current Chinese score: 90, current English score: 80, past 3 math scores: 82, 78, 85, past 3 Chinese scores: 88, 89, 90, past 3 English scores: 75, 78, 80; Label: user overall score classification: excellent;

[0086] Through the decision tree algorithm, we can construct the following decision tree: root node: if the current math score > 80, yes: go to the next node, no: classify as medium; second-level node: if the current Chinese score > 85, yes: classify as excellent, no: go to the next node; third-level node: if the current English score > 75, yes: classify as good, no: classify as passing.

[0087] According to the above decision tree, the user's classification results and the generated performance analysis report can help parents, teachers and users better understand the user's learning situation and develop personalized learning plans and improvement strategies.

[0088] The user attention need calculation module extracts the user's performance characteristics, behavior characteristics, background characteristics and health characteristics from the user performance analysis report and the user profile information to obtain a feature set; uses the known user's characteristics and attention need labels as training data to train a random forest model for predicting the user's comprehensive attention need and various aspects of attention need;

[0089] In the random forest model, each tree is constructed using information gain to select the best split feature. When constructing a single decision tree, for a given training data set, a subset is randomly sampled with replacement, and a feature subset is randomly selected from the feature set for splitting. The information gain is used to select the optimal split point and construct a decision tree. The final prediction result is obtained by voting among all decision trees.

[0090] When calculating the comprehensive attention, the user's performance characteristics, behavior characteristics, background characteristics and health characteristics are integrated to form a comprehensive feature vector; the comprehensive feature vector is input into the trained random forest model to predict each user's comprehensive attention needs and various aspects of attention needs.

[0091] The final prediction result is obtained by voting of all decision trees. Assume that there are N decision trees in the random forest. , , ..., ; For an input sample x, each tree Give a prediction ; The final prediction result is the category with the most votes among all trees’ predictions:

[0092]

[0093] Among them, mode represents the mode, and the predicted category with the highest frequency is output;

[0094] When integrating the user's performance characteristics, behavior characteristics, background characteristics, and health characteristics, each characteristic needs to be standardized. Z-score standardization is used for processing. , calculate its mean and standard deviation , the formula is as follows:

[0095]

[0096] in, It is a feature The standardized value of

[0097] Assign a weight to each feature, and set the weights of performance features, behavior features, background features, and health features to be , , and , the score feature vector is , the behavior feature vector is , the background feature vector is , the health feature vector is , sum up all kinds of features according to their weights to form a comprehensive feature vector X:

[0098] ;

[0099] in, , , and They are the standardized values ​​of the performance feature vector, behavior feature vector, background feature vector and health feature vector respectively.

[0100] According to another embodiment of the present invention, Figure 2 A method for integrated management of archival information storage based on artificial intelligence is shown in a schematic diagram of steps. The method for integrated management of archival information storage based on artificial intelligence comprises the following steps:

[0101] S100, synchronize data with the academic management system, HR system and email in real time through interfaces;

[0102] S200, pre-processing the scanned paper files, using a convolutional neural network to convert text content in the image into editable text information;

[0103] S300, performing syntax parsing and semantic analysis on the input text information to extract key information and metadata;

[0104] S400, the administrator pre-configures the workflow, automatically assigns tasks for file information approval and monitors task execution, and automatically reminds relevant personnel of tasks that are about to expire or have expired;

[0105] S500, collects statistics on the execution data of various workflows, generates process analysis reports, provides API interfaces, and realizes task integration with other information systems;

[0106] S600, whenever the user needs to confirm the file information, the user file information is sent to the corresponding user for the user to confirm. If the user confirmation is incorrect, step S601 is executed, otherwise step S700 is executed;

[0107] S601, sending feedback information to the administrator, re-executing the automated workflow module for the information, confirming the correct archive information, and updating it to other systems;

[0108] S700, after each user confirms the test score, establish interaction between parents, teachers and users, and use decision tree algorithm to analyze and generate a user score analysis report;

[0109] S800. Use the random forest algorithm to combine and analyze the user's performance analysis report with the key information and metadata in the user's profile information, generate the comprehensive attention needs of different users and the attention needs of the users in different aspects, and send them to the corresponding class teachers.

[0110] In this example, the user represents a student:

[0111] The data is as follows: User's grades: Math 85, Chinese 90, English 80. Behavior record: Attendance rate 95%, high class participation, no disciplinary record. Family background: Family economic situation is average, parents have high education level. Health status: Physically healthy, mental health record is normal.

[0112] Feature vector X = [85, 90, 80, 0.95, 1, 0, 1, 1]; When performing random forest prediction, the training data set contains feature vectors of multiple users and corresponding attention labels. The random forest model is trained using several decision trees, and each tree randomly selects features and samples for training. Prediction results: comprehensive attention: medium, academic attention: low, behavior attention: low, family background attention: medium, health attention: low; the report generated by the system includes:

[0113] Comprehensive attention score: medium;

[0114] The scores of attention required in each aspect: academic (low), behavior (low), family background (medium), health (low);

[0115] Personalized recommendations: Personalized learning and development recommendations based on the level of attention needed, such as strengthening family communication, maintaining a healthy lifestyle, etc.

[0116] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0117] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An integrated archival information storage management system based on artificial intelligence, characterized in that: include: Data entry module: synchronizes data with the academic affairs management system, HR system and email in real time through interfaces, pre-processes scanned paper archives, uses convolutional neural networks to convert text content in images into editable text information, performs syntax parsing and semantic analysis on the entered text information, and extracts key information and metadata; Automated workflow module: Administrators pre-configure workflows, automatically assign tasks for file information approval, monitor task execution, and automatically remind relevant personnel of tasks that are about to expire or have expired; collect statistics on the execution data of various workflows, generate process analysis reports, and provide API interfaces to achieve task integration with other information systems; Intelligent feedback communication module: Whenever the user needs to confirm the file information, the user file information will be sent to the corresponding user for confirmation. If the user's confirmation is incorrect, the feedback information will be sent to the administrator, and the automated workflow module will be re-executed for the file information to confirm the correct file information and update it to other systems; after each user confirms the test score, the decision tree algorithm is used to generate a user score analysis report; User attention need calculation module: Use the random forest algorithm to combine the user performance analysis report with the key information and metadata in the user profile information to generate the comprehensive attention need of different users and the attention need of the user in different aspects; The user attention need calculation module extracts the user's performance characteristics, behavior characteristics, background characteristics and health characteristics from the user performance analysis report and the user profile information to obtain a feature set; uses the known user's characteristics and attention need labels as training data to train a random forest model for predicting the user's comprehensive attention need and various aspects of attention need; In the random forest model, each tree is constructed using information gain to select the best split feature. When constructing a single decision tree, for a given training data set, a subset is randomly sampled with replacement, and a feature subset is randomly selected from the feature set for splitting. The information gain is used to select the optimal split point and construct a decision tree. The final prediction result is obtained by voting among all decision trees. When calculating the comprehensive attention, the user's performance characteristics, behavior characteristics, background characteristics and health characteristics are integrated to form a comprehensive feature vector; the comprehensive feature vector is input into the trained random forest model to predict the comprehensive attention needs and various aspects of each user; The final prediction result is obtained through voting of all decision trees. When a new sample needs to be predicted, each decision tree will make an independent prediction for the sample and obtain a classification result. For classification problems, each tree will directly give a category label. When integrating the user's performance characteristics, behavior characteristics, background characteristics, and health characteristics, each characteristic needs to be standardized, using Z-score standardization; Assign a weight to each feature, set the weights of performance features, behavior features, background features and health features, combine the performance feature vector, behavior feature vector, background feature vector and health feature vector, sum up each type of feature according to the weight, and form a comprehensive feature vector.

2. According to claim 1, an integrated archival information storage management system based on artificial intelligence is characterized by: The data entry module uses Gaussian filtering to remove noise from the image, converts the color image into a grayscale image, simplifies the information representation of the image, performs binarization processing on the grayscale image, converts the image into a black and white image, and enhances the image to increase contrast and clarity; Annotate the preprocessed image, mark the position of each user's profile information to be entered in the image and the corresponding characters; use VGG as the model structure of the convolutional neural network, the input layer accepts the preprocessed user profile image as input, set the convolution kernel size, number of convolution kernels, activation function, padding and step size of different convolution layers, set the pooling window size and step size of different maximum pooling layers; stack several convolution layers, keep the convolution kernel size, activation function and step size settings unchanged, and increase the number of convolution kernels; set the number of neurons and activation function of the fully connected layer; the number of neurons in the output layer is the same as the character set size of the input content, use Adam optimizer as the loss function for compilation and training, use the test data set to evaluate the trained model, and perform text recognition after evaluation.

3. The integrated archival information storage management system based on artificial intelligence according to claim 1 is characterized in that: The data entry module identifies the sentence structure in the text through syntactic analysis, and identifies the sentences in the user profile information that contain the user's key information, where the user's key information is pre-set by the administrator; performs part-of-speech tagging on each word in the text, analyzes the dependency relationship between key information words in the text, constructs a dependency tree structure of the user's key information, and extracts the key information; Extract attribute information of user profile from text, extract numerical information related to user profile, and extract the relationship between user profile information in text as metadata.

4. The integrated archival information storage management system based on artificial intelligence according to claim 1 is characterized in that: The administrator defines a specific process for file information approval in the automated workflow module, wherein the specific process for approval includes approval nodes, approval personnel, approval logic, approval time limit, and approval sequence; When a new application for file information approval is submitted, the system automatically triggers the corresponding approval process instance; according to the defined approval process, the system automatically assigns the approval task to the corresponding approval personnel according to the preset rules; the system monitors the execution status of the approval task in real time, which includes the current task handler, approval progress, remaining time, task submission time, task type, approval history and approval result; the system monitors the remaining time of the approval task according to the set approval time limit, and triggers a reminder notification when the time limit is approaching or has expired; If a task has expired and has not been processed, the system will automatically send an urgent reminder notification to the relevant personnel; the system records the processing process and results of the approval task.

5. The integrated archival information storage management system based on artificial intelligence according to claim 1 is characterized in that: The automated workflow module analyzes the collected execution data, and the obtained analysis results include the execution efficiency of the workflow, the processing time of the approval task, and the approval pass rate, and generates a process analysis report based on the analysis results, and the process analysis report includes a text description and a chart display; Design an API interface to provide query and integration functions for execution data to other information systems. The interface includes functions for querying task status, obtaining approval history, and triggering approval processes.

6. The integrated archival information storage management system based on artificial intelligence according to claim 1 is characterized in that: In the intelligent feedback communication module, when the user needs to confirm the file information, the system triggers the file information confirmation event according to the preset conditions; the system sends the file information that needs to be confirmed to the corresponding user according to the triggered file information confirmation event; The user receives a confirmation task notification, logs into the system, browses the archive information and confirms; If the user confirms that the file information is incorrect, the system provides a feedback mechanism, allowing the user to mark the errors and provide explanations or remarks; after receiving the user's feedback, the system automatically sends feedback to the administrator to remind the administrator to deal with the incorrect file information; after the administrator receives the feedback, the system automatically triggers the automated workflow module to re-process the information; in the automated workflow module, the administrator re-checks the file information, confirms the correct information and updates it; after confirming the correct file information, the system automatically synchronizes the updated information to other systems.

7. The integrated archival information storage management system based on artificial intelligence according to claim 1 is characterized in that: The intelligent feedback communication module generates a performance analysis report based on the confirmed performance using a decision tree algorithm, wherein the performance analysis report includes a comparative analysis of the current and history subject performances, the current and history class rankings, the current and history grade rankings, and the current and history learning situations; Information gain is used to select the split point in the decision tree and measure the information increment that a feature can bring when splitting on the data set; Information gain is calculated by collecting the current user's data set, candidate split features, all possible values ​​of the candidate split features, a subset of the specified values ​​of the candidate split features, and data set entropy. The candidate split features include feature vectors extracted from the current and history subject scores, the current and history class rankings, and the current and history grade rankings. For each candidate split feature, the information gain after splitting is calculated, and the feature with the largest information gain is selected as the split point. The optimal splitting feature is selected based on information gain, and a subtree is recursively constructed. Information gain is recursively performed on the child nodes until the node purity is higher than the threshold or the maximum depth is reached. The decision tree model is trained using labeled data. Based on the model's prediction results, a comparative analysis of the current and historical learning situations is obtained, and a performance analysis report is generated together with the grades and rankings.

8. An integrated management method for archival information storage based on artificial intelligence, applied to an integrated management system for archival information storage based on artificial intelligence as claimed in any one of claims 1 to 7, characterized in that: The following steps are involved: S100, synchronize data with the academic management system, HR system and email in real time through interfaces; S200, pre-processing the scanned paper files, using a convolutional neural network to convert text content in the image into editable text information; S300, performing syntax parsing and semantic analysis on the input text information to extract key information and metadata; S400, the administrator pre-configures the workflow, automatically assigns tasks for file information approval and monitors task execution, and automatically reminds relevant personnel of tasks that are about to expire or have expired; S500, collects statistics on the execution data of various workflows, generates process analysis reports, provides API interfaces, and realizes task integration with other information systems; S600, whenever the user needs to confirm the file information, the user file information is sent to the corresponding user for the user to confirm. If the user confirmation is incorrect, step S601 is executed, otherwise step S700 is executed; S601, sending feedback information to the administrator, re-executing the automated workflow module for the information, confirming the correct archive information, and updating it to other systems; S700, after each user confirms the test score, a decision tree algorithm is used to analyze and generate a user score analysis report; S800. Use a random forest algorithm to combine and analyze the user performance analysis report with key information and metadata in the user profile information to generate the comprehensive attention needs of different users and the attention needs of the users in different aspects.

Citation Information

Patent Citations

  • OBE (Outcomes-based Education) course development teaching system for higher vocational school

    CN108596807A

  • Engineering archive intelligent collection method and system based on enabling thinking

    CN114117171A

  • Intelligent training integrated management system

    CN116227981A

  • Financial information management system

    CN117876139A

  • Learning early warning method based on improved random forest algorithm

    CN118505457A