Practical training teaching resource sharing method and system based on artificial intelligence

Abstract

This application relates to the field of data storage technology, specifically disclosing an artificial intelligence-based method and system for sharing practical training resources. The method includes: deploying multiple storage servers, each with a login management table; when a user uploads teaching resources, they select a storage server as the first server based on the login management table; selecting a second server from among the storage servers and synchronizing the teaching resources to the second server; the second server predicting access pressure over multiple future time periods based on its stored teaching resources; if multiple future time periods have access pressure exceeding a first threshold, selecting a third server and synchronizing a portion of the stored teaching resources to the third server; the second and third servers distributing access keys to the sharing target, and the sharing target accessing the teaching resources based on the access keys. This invention achieves intelligent allocation and optimization of teaching resources, significantly improving the user experience.

Description

Technical Field

[0001] This application relates to the field of data storage technology, and in particular to a method and system for sharing practical training and teaching resources based on artificial intelligence. Background Technology

[0002] Traditional methods of sharing practical training resources have many limitations, such as uneven resource allocation, low access efficiency, and the inability to effectively predict and cope with peak access times. These problems limit the maximum utilization of educational resources. To address these issues, researchers have begun to explore using artificial intelligence (AI) technology to optimize the sharing of practical training resources. AI technology can analyze access patterns of teaching resources and predict future access pressure, thereby achieving intelligent allocation and optimization of resources.

[0003] For example, Chinese patent document CN118862120A discloses a data access method and system for a teaching platform. This method utilizes a Hadoop cluster and streaming tools. The configuration of the Hadoop cluster and real-time monitoring of data access parameters ensure the stability and scalability of the teaching platform, adapting to different user sizes and data volumes. The application of streaming tools and data access optimization enables the teaching platform to respond to user data access requests in real time. Another example is Chinese patent document CN116028501A, which discloses a blockchain-based artificial intelligence data sharing method and system. This method obtains shared values ​​by analyzing shared parameters (synchronous value, common value, and time-sharing value) of shared data, and then divides the data into frequently used and infrequently used data, storing them in different blocks. The system can also intelligently display shared data based on shared values ​​or display coefficients, improving user experience.

[0004] However, while the methods described above improve the efficiency and user experience of sharing teaching resources to some extent, they still have some drawbacks. For example, Patent Document 1 may not be flexible and accurate enough in terms of resource allocation and access prediction, leading to ineffective handling during peak access periods and potential inefficiencies in resource allocation. While Patent Document 2 improves data security, it may face performance bottlenecks in handling large-scale data and real-time requirements. Summary of the Invention

[0005] To address the problems mentioned in the background, this application provides a method and system for sharing practical training and teaching resources based on artificial intelligence.

[0006] To achieve the aforementioned objectives, this invention proposes a method for sharing practical training resources based on artificial intelligence, comprising:

[0007] Multiple storage servers are deployed, and a login management table is set up in each of the storage servers. The login management table records the historical login information of each user.

[0008] When a user uploads teaching resources, they select one of the storage servers as the first server according to the login management table.

[0009] Based on the sharing goal of the teaching resources and the login management table, the first server selects a second server from the storage servers and synchronizes the teaching resources to the second server;

[0010] The second server, based on the teaching resources it has stored and the login management table, combined with a prediction model, predicts the access pressure for multiple future time periods;

[0011] If there are multiple future time periods where the access pressure exceeds the first threshold, a third server is selected from the storage servers, and a portion of the stored teaching resources are synchronized to the third server.

[0012] The second server and the third server distribute access keys to the shared target, and the shared target obtains the teaching resources based on the access keys.

[0013] Furthermore, predicting the access pressure for the future time period includes the following steps:

[0014] The login management table is used to extract the hourly access volume of each teaching resource. Based on the access volume, a daily access sequence is generated. The access sequence is clustered in units of years to obtain multiple clustering results. Based on the clustering results, multiple access patterns are divided for each teaching resource, and the center of the clustering results is used as the representative sequence of the access pattern.

[0015] The teaching resources stored on the second server are obtained as resources to be analyzed. Based on the date of the future time period, the access pattern of each resource to be analyzed is obtained, and the representative sequence in the access pattern is obtained as the first prediction sequence.

[0016] Obtain the monthly user change, revise the first prediction sequence based on the monthly user change to obtain a second prediction sequence, merge the second prediction sequences of different resources to be analyzed to obtain a third prediction sequence, and use the hourly access volume in the third prediction sequence as the access pressure for the future time period.

[0017] Furthermore, the clustering of the access sequences includes the following steps:

[0018] Different access sequences are traversed and combined to obtain multiple combination sequences. Each combination sequence includes two access sequences, which are defined as a first sequence and a second sequence, respectively. A difference sequence is calculated based on the first sequence and the second sequence. The difference sequence is the absolute value of the difference in access volume between the first sequence and the second sequence in the corresponding time period. A first feature is obtained after processing the difference sequence using the sliding window method.

[0019] Filter the top N largest access counts in the first sequence and the second sequence respectively, and the corresponding first time period and second time period. Extract the third sequence corresponding to the first time period and the fourth sequence corresponding to the second time period from the difference sequence. Calculate the average value of the third sequence and the fourth sequence, and use them as the second feature and the third feature respectively.

[0020] The access sequences for each month are clustered based on the first feature, the second feature, and the third feature.

[0021] Further, selecting the third server from the storage server includes the following steps:

[0022] Define the future time period in which the access pressure of the second server is greater than the first threshold as the migration time period. Take the sum of the access pressure of the storage servers in the migration time period as the first pressure value. Calculate the second pressure value of each storage server for the entire day in the migration time period. The login management table includes the user's access speed. Perform a weighted summation of the access speed, the first pressure value, and the second pressure value to obtain the evaluation value of the storage server. Select the storage server with the smallest evaluation value as the third server.

[0023] Furthermore, the process by which the shared target obtains the teaching resources based on the access key includes the following steps:

[0024] The second server and the third server generate a shared key and a one-dimensional initial matrix, generate a key sequence based on the shared key and the one-dimensional initial matrix, divide the key sequence into multiple sub-sequences, generate a storage address based on each sub-sequence, select one of the storage addresses as the target address, the second server and the third server generate a random key, encrypt the teaching resources based on the random key to obtain encrypted data, and store the encrypted data and the random key in different storage addresses, which are defined as the first address and the second address, respectively.

[0025] Based on the access time of the shared target, a target server is selected from the second server and the third server. The shared target obtains the teaching resources after decrypting the encrypted data of the first address based on the random key of the second address in the target server.

[0026] Furthermore, after the shared target obtains the teaching resources, the target server regenerates the random key, uses the random key to re-encrypt the teaching resources and stores them in the first address, selects a third address from the storage addresses, and stores the regenerated random key in the third address.

[0027] Furthermore, generating the storage address based on the subsequence includes the following steps:

[0028] The numerical values ​​included in the subsequence are converted into ASCII characters, and then the ASCII characters are converted into basic characters according to the corresponding multi-language character names. Based on the basic characters, a character sequence of the subsequence is generated, and a conversion rule table is established. The conversion rule table includes the correspondence between each character sequence and the storage address.

[0029] Furthermore, if the teaching resources are not present on the third server, a request is made to retrieve the teaching resources from the second server.

[0030] Furthermore, the access sequences are clustered using K-means or DBSCAN algorithms.

[0031] This invention also provides an artificial intelligence-based practical training resource sharing system for the above-described method, the system comprising:

[0032] The storage module includes multiple storage servers, and a login management table is set in each of the storage servers. The login management table records the historical login information of each user.

[0033] In the upload module, when a user uploads teaching resources, the upload module selects a storage server as the first server based on the login management table. The first server selects a second server from the storage servers based on the sharing target of the teaching resources and the login management table, and synchronizes the teaching resources to the second server.

[0034] The optimization module, based on the teaching resources it has stored and the login management table, predicts the access pressure for multiple future time periods using a prediction model; if there are multiple future time periods where the access pressure is greater than a first threshold, the optimization module selects a third server from the storage server and synchronizes a portion of the stored teaching resources to the third server.

[0035] An encryption module is provided, and the second server and the third server distribute access keys to the shared target based on the encryption module, and the shared target obtains the teaching resources based on the access keys. Beneficial effects

[0036] This invention deploys multiple storage servers and sets up a login management table on each server to record the login time and address of each user in detail. When a user uploads teaching resources, the system selects the most suitable storage server as the first server based on the user's login history and sharing goals. The first server then intelligently selects a second server based on the sharing goals of the teaching resources and the login management table, and synchronizes the resources to it, thereby optimizing resource distribution. By predicting access pressure in future time periods using a predictive model, it can more accurately predict and respond to access peaks, thus avoiding inefficient resource allocation and system pressure during peak access times. This invention's dynamic resource allocation mechanism based on user behavior and resource needs can more flexibly respond to the resource needs of different users and at different times, thereby improving resource utilization efficiency and user access experience. Attached Figure Description

[0037] Figure 1 This is a flowchart illustrating the steps of a practical training and teaching resource sharing method based on artificial intelligence, as described in this application.

[0038] Figure 2 A schematic diagram illustrating the principle of generating the difference sequence for this application;

[0039] Figure 3 A schematic diagram illustrating the principle of generating the first feature for this application;

[0040] Figure 4 This is a schematic diagram of the structure of a practical training and teaching resource sharing system based on artificial intelligence, as proposed in this application. Detailed Implementation

[0041] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0042] It is understood that the terms "first," "second," etc., used in this application may be used herein to describe various elements, but unless otherwise specified, these elements are not limited by these terms. These terms are used only to distinguish one element from another. For example, without departing from the scope of this application, a first script may be referred to as a second script, and similarly, a second script may be referred to as a first script.

[0043] like Figure 1 As shown, an artificial intelligence-based method for sharing practical training resources includes:

[0044] S1: Deploy multiple storage servers, and set up a login management table in each storage server. The login management table records the historical login information of each user.

[0045] S2: When a user uploads teaching resources, a storage server is selected as the first server based on the login management table.

[0046] In this embodiment, three storage servers are deployed, namely storage servers A, B, and C. Each storage server has a login management table, which records the user's login time and login address. The login time is accurate to the minute, and the login address includes the IP address and the actual location of the user. Each time a user logs in, storage servers A, B, and C record the login information in their login management tables and synchronize it with the other storage servers, ensuring that the login management tables of storage servers A, B, and C remain consistent.

[0047] When user A uploads teaching resources, the sharing target is first determined. The sharing target is other users who can access and download the teaching resources. For example, user A specifies the 3000 employees in company B as the sharing target. Then, based on the login management table, it is determined which storage server user A is frequently assigned to for data communication. For instance, the login management table shows that user A has been consistently assigned to storage server A for data communication over the past month; therefore, storage server A is designated as the primary server. The teaching resources uploaded by user A are first stored on the primary server.

[0048] S3: Based on the sharing goals of teaching resources and the login management table, the first server selects the second server from the storage servers and synchronizes the teaching resources to the second server.

[0049] The first server can determine which storage server each shared target frequently logs into through the login management table. For example, if it's determined that all shared targets frequently log into storage server B, storage server B can be designated as the second server, and the teaching resources can be copied to the second server. This method eliminates the need to copy the teaching resources to each storage server, thus reducing the resource consumption of the storage servers. In particular, the first server itself can also act as a second server, and multiple second servers can be selected.

[0050] S4: The second server uses its stored teaching resources and login management table, combined with a predictive model, to predict access pressure for multiple future time periods.

[0051] S5: If there are multiple future time periods where the access pressure exceeds the first threshold, select a third server from the storage servers and synchronize some of the stored teaching resources to the third server.

[0052] The login management table also includes the time point when each user accesses each type of teaching resource. The second server can predict the access pressure in future time periods based on its stored teaching resources and the login management table using a prediction model. By setting a first threshold, it can determine which time periods have high access pressure and which have low access pressure. For example, if the predicted access pressure between 19:00 and 21:00 tomorrow is greater than the first threshold, it indicates that there will be high access pressure between 19:00 and 21:00 tomorrow. In this case, the storage server with lower access pressure is selected as the third server. The second server sends some teaching resources to the third server and diverts users to access the third server to download teaching resources, thereby reducing its own access pressure. The specific method for predicting access pressure will be introduced later.

[0053] S6: The second and third servers distribute access keys to the shared target, and the shared target obtains teaching resources based on the access keys.

[0054] The second and third servers also send access keys to the shared target's account. Only the shared target holding the access key can decrypt and access the teaching resources, thus ensuring that unauthorized users cannot access the teaching resources.

[0055] Specifically, if the teaching resources are not available on the third server, a request is made to retrieve the teaching resources from the second server.

[0056] If the third server loses teaching resources due to an abnormal reason, the lost teaching resources can be retrieved from the second server.

[0057] This invention deploys multiple storage servers and sets up a login management table on each server to record the login time and address of each user in detail. When a user uploads teaching resources, the system selects the most suitable storage server as the first server based on the user's login history and sharing goals. The first server then intelligently selects a second server based on the sharing goals of the teaching resources and the login management table, and synchronizes the resources to it, thereby optimizing resource distribution. By predicting access pressure in future time periods using a predictive model, it can more accurately predict and respond to access peaks, thus avoiding inefficient resource allocation and system pressure during peak access times. This invention's dynamic resource allocation mechanism based on user behavior and resource needs can more flexibly respond to the resource needs of different users and at different times, thereby improving resource utilization efficiency and user access experience.

[0058] This solution effectively improves the ability to handle large-scale data and real-time demands by leveraging the collaborative work of multiple storage servers and the application of predictive models, thus avoiding performance bottlenecks. Furthermore, by distributing access keys to shared targets, this solution ensures the security of teaching resources and the precision of access control, further enhancing system security and user experience.

[0059] Of particular note is that this invention enables intelligent allocation and optimization of teaching resources, improves the ability to process large-scale data and in real time, and greatly enhances the user experience.

[0060] This embodiment predicts access pressure for a future time period by including the following steps:

[0061] Extract the hourly access volume of each teaching resource from the login management table, generate daily access sequences based on the access volume, cluster the access sequences on an annual basis to obtain multiple clustering results, divide each teaching resource into multiple access patterns based on the clustering results, and use the center of the clustering results as the representative sequence of the access pattern.

[0062] Among them, K-means or DBSCAN algorithms are used to cluster the access sequences.

[0063] The login management table records user access records for each type of teaching resource. By statistically analyzing these records, the hourly access volume for each teaching resource can be obtained. For example, the access volume for teaching resource A between 19:00 and 20:00 is 200. After obtaining the access volume, an access sequence for each teaching resource is generated on a daily basis, with each access sequence containing 24 access records. Then, the access sequences from the past year are clustered, allowing the identification of access patterns for various teaching resources based on the clustering results. For instance, if the clustering result includes 5 clusters, it indicates that all teaching resources have a total of 5 access patterns. The center of each cluster is then used as the representative sequence for each access pattern. Specifically, this embodiment uses the DBSCAN algorithm for clustering, thus omitting the step of determining the number of clusters.

[0064] The teaching resources stored on the second server are obtained as the resources to be analyzed. Based on the date of the future time period, the access patterns of each resource to be analyzed are obtained, and the representative sequence in the access patterns is obtained as the first prediction sequence.

[0065] In this embodiment, the teaching resources already stored on the second server are first obtained during the calculation. For ease of differentiation, these are defined as resources to be analyzed. This embodiment determines the access pattern of each teaching resource based on the following steps: First, the date for the future time period to be predicted is obtained. Then, dates with the same attribute from the past year are obtained. "Same attribute" means that both dates are in the same month, both are either weekdays or rest days, and the order of weekdays or rest days in the month is the same. For example, March 1, 2020, is determined to be the first rest day of March. Therefore, the date of the first rest day of March 2019 is obtained as March 2. Next, the access patterns of each resource to be analyzed for the five rest days prior to March 1, 2020 are obtained, defined here as the access pattern sequence for the current time period, and the access patterns for the five rest days prior to March 2, 2019 are defined here as the access pattern sequence for the past time period.

[0066] By comparing the access model sequences of the current time period and past time periods, the number of times the same pattern appears is determined. For example, by comparison, it is determined that the resource A to be analyzed has three identical access patterns in the current time period and past time periods, which is greater than or equal to the second threshold. The second threshold is set to 3. Then, the access pattern of teaching resource A on March 2, 2019 is set as the access pattern on March 1, 2020, and the representative sequence is used as the first prediction sequence for March 1.

[0067] If the number of times the same access pattern occurs is less than the second threshold, it indicates that the access to resource A to be analyzed has changed, possibly due to a change in the exam month. In this case, obtain the access volume of resource A to be analyzed every hour for each day during the five rest days before March 1, 2020. Add up the access volumes for the same time period, calculate the average, and multiply by a preset coefficient to obtain the access volume for March 1. For example, from February 25 to February 29, the access volumes between 18:00 and 19:00 are 15, 25, 30, 40, and 26 respectively. Then, the access volume between 18:00 and 19:00 on March 1, 2020 is (18+15+30+40+26) / 5*0.8=20.64.

[0068] Obtain the monthly user change, revise the first prediction sequence based on the monthly user change to obtain the second prediction sequence, merge the second prediction sequences of different resources to be analyzed to obtain the third prediction sequence, and use the hourly access volume in the third prediction sequence as the access pressure for future time periods.

[0069] The monthly user change is obtained based on the number of registered users in the system. For example, if there are 3,000 registered users on February 29th and 3,015 registered users on March 1st, the ratio is 3015 / 3000 = 1.005. Therefore, the access volume for each time period in the first prediction sequence is multiplied by 1.005 to obtain the second prediction sequence. If the second server contains teaching resources A, B, and C, the second prediction sequences for all three resources are summed (adding the access volumes for the same time period) to obtain the third prediction sequence. The third prediction sequence can then be used to obtain the hourly access volume of the second server on March 1st, thus providing the hourly access pressure.

[0070] This embodiment performs clustering of access sequences, including the following steps:

[0071] Different access sequences are traversed and combined to obtain multiple combination sequences. Each combination sequence includes two access sequences, which are defined as the first sequence and the second sequence, respectively. A difference sequence is calculated based on the first sequence and the second sequence. The difference sequence is the absolute value of the difference in access volume between the first sequence and the second sequence in the corresponding time period. The first feature is obtained after processing the difference sequence using the sliding window method.

[0072] For example, given the access sequences 1, 2, and 3, the combined sequences obtained after traversing and combining them are 1, 2, and 3. Combination sequence 1 includes access sequences 1 and 2, combination sequence 2 includes access sequences 2 and 3, and combination sequence 3 includes access sequences 1 and 3. Here, we choose one of these combination sequences for explanation, defining access sequences 1 and 2 as the first sequence and the second sequence, respectively. For ease of explanation, only a simple example is provided. Figure 2 As shown, assuming the first sequence is [10, 50, 30, 20, 60] and the second sequence is [20, 40, 70, 55, 25], then the difference sequence is [10, 10, 40, 35, 35]. Here, the window size is set to 3, and the sliding step is 1. Figure 3 As shown, the first feature obtained from the difference sequence is [20, 28.3, 36.7]. The closer the value in the first feature is to 0, the more similar the first sequence and the second sequence are.

[0073] Filter the top N visits with the largest values ​​in the first and second sequences respectively, and the corresponding first and second time periods. Extract the third sequence corresponding to the first time period and the fourth sequence corresponding to the second time period from the difference sequence. Calculate the average value of the third and fourth sequences, and use them as the second and third features respectively.

[0074] Setting N to 2, the first sequence contains 60 and 50, corresponding to the first time intervals of 5 and 2. The second sequence contains 70 and 55, corresponding to the second time intervals of 3 and 4. The third sequence is [10, 35] with a mean of 22.5, and the fourth sequence is [40, 35] with a mean of 37.5. We then use 22.5 and 37.5 as the second and third features, respectively. The smaller the values ​​of the second and third features, the more similar the first and second sequences are.

[0075] The access sequences for each month are clustered based on the first, second, and third features.

[0076] Using the extracted first, second, and third features as input features for the clustering algorithm can significantly improve the clustering speed when dealing with large amounts of data.

[0077] This embodiment of selecting a third server from a storage server includes the following steps:

[0078] Define the future time period in which the access pressure of the second server is greater than the first threshold as the migration time period. Take the sum of the access pressure of the storage servers during the migration time period as the first pressure value. Calculate the second pressure value of each storage server for the whole day during the migration time period. The login management table includes the user's access speed. Perform a weighted sum of the access speed, the first pressure value and the second pressure value to obtain the evaluation value of the storage server. Select the storage server with the smallest evaluation value as the third server.

[0079] Based on predictions, it was determined that the access pressure on the second server would exceed the first threshold between 18:00 and 20:00 tomorrow. Therefore, the migration period was defined as 18:00 to 20:00 tomorrow. Subsequently, the access pressure of the remaining storage servers between 18:00 and 20:00 tomorrow was obtained. The sum of the access pressure between 18:00 and 19:00 and between 19:00 and 20:00 was used as the first pressure value. The smaller the first pressure value, the lower the access pressure of the storage server during the migration period.

[0080] Next, the second stress value for each storage server throughout tomorrow is calculated. In this embodiment, the second stress value is calculated as follows: First, the third predicted sequence for each storage server tomorrow is obtained. Then, the third predicted sequence is fitted using the least squares method to obtain a function of access stress changing over time. Then obtain the function The reciprocal of the integral over the entire day, i.e. Thus, the second pressure value is obtained. In the above formula, P is the second pressure value. The starting time for tomorrow. The end time tomorrow. The higher the second pressure value, the greater the access pressure on the storage server throughout the day, and the smaller its reciprocal.

[0081] The lower the access speed value, the faster the access speed, and the faster users can access resources. Setting the access speed weight to 0.3, the first stress value weight to 0.5, and the second stress value weight to 0.2, the evaluation value is... Where S is the evaluation value, and A, B, and C are the access speed, the first stress value, and the second stress value, respectively. A smaller evaluation value indicates a lower overall access pressure on the storage server.

[0082] In this embodiment, the process of a shared target obtaining teaching resources based on an access key includes the following steps:

[0083] The second and third servers generate a shared key and a one-dimensional initial matrix. Based on the shared key and the one-dimensional initial matrix, they generate a key sequence, divide the key sequence into multiple sub-sequences, generate a storage address based on each sub-sequence, select a storage address as the target address, and the second and third servers generate a random key. Based on the random key, they encrypt the teaching resources to obtain encrypted data. The encrypted data and the random key are stored in different storage addresses, which are defined as the first address and the second address, respectively.

[0084] Generating a storage address based on a subsequence includes the following steps:

[0085] The numerical values ​​included in the subsequence are converted into ASCII characters. Then, the ASCII characters are converted into basic characters according to the corresponding multi-language character names. Based on the basic characters, the character sequence of the subsequence is generated, and a conversion rule table is established. The conversion rule table includes the correspondence between each character sequence and the storage address.

[0086] For ease of explanation in this embodiment, the shared key is set to... The one-dimensional initial matrix is The key sequence is: Then, the sequence is divided into subsequences of fixed length, for example, two elements per subsequence. Subsequence 1 is [8, 8], and subsequence 2 is [24, 0]. If the last digit is insufficient, it is padded with 0s. The elements of each subsequence are then converted to ASCII characters. For example, 8 is converted to BS (backspace, Ctrl / H), so it is replaced with H. Similarly, 24 is converted to X, and 0 is converted to @. Thus, the two subsequences are converted to HH and X@ respectively. Next, an address translation rule table is set, which includes the correspondence between each character and its memory address. For example, HH corresponds to memory address 1, and X@ corresponds to memory address 2.

[0087] The random keys generated by the second and third servers are random numbers. Encrypted data is obtained by XORing the random numbers with the teaching resources. In other embodiments, logical algorithms such as AND and OR can also be used for encryption. After encryption, the encrypted data is stored in storage address 1, and the random numbers are stored in storage address 2.

[0088] Based on the access time of the shared target, the target server is selected from the second and third servers. After the shared target decrypts the encrypted data of the first address using a random key based on the second address on the target server, the teaching resources are obtained.

[0089] When a shared target needs to access teaching resources, it first accesses the second server. If the second server determines that its current access pressure is low, it uses itself as the target server. If the access pressure on the second server is high, it uses the third server as the target server. The sharing user generates subsequence 1 and subsequence 2 using the previously described method to determine the possible storage addresses where the teaching resources and random keys are stored. Then, it retrieves the encrypted data and random key from the storage addresses and uses the random key to decrypt the encrypted data to obtain the teaching resources. If there are at least three storage addresses, after obtaining the encrypted data from one of the storage addresses, it sequentially retrieves keys from the remaining two storage addresses and attempts to decrypt them until the correct random key is obtained.

[0090] In this embodiment, after the shared target obtains the teaching resources, the target server regenerates a random key, uses the random key to re-encrypt the teaching resources and stores them in the first address, selects a third address from the storage addresses, and stores the regenerated random key in the third address.

[0091] Through this step, whenever a shared target accesses the teaching resources on the target server, the target server regenerates a random key, re-encrypts the teaching resources using the newly generated random key, obtains new encrypted data, stores the encrypted data in the original first address, and then selects a third address from the storage addresses to store the random key in the third address. If only two storage addresses exist, the random key is still stored in the second address.

[0092] like Figure 4 As shown, the present invention also provides an artificial intelligence-based practical training and teaching resource sharing system for the above-described method, the system comprising:

[0093] The storage module includes multiple storage servers, each with a login management table that records each user's historical login information.

[0094] In the upload module, when a user uploads teaching resources, the upload module selects a storage server as the first server based on the login management table. The first server then selects a second server from the storage servers based on the sharing target of the teaching resources and the login management table, and synchronizes the teaching resources to the second server.

[0095] The optimization module, based on its stored teaching resources and login management table, uses a prediction model to predict access pressure for multiple future time periods. If there are multiple future time periods with access pressure exceeding the first threshold, the optimization module selects a third server from the storage servers and synchronizes some of the stored teaching resources to the third server.

[0096] The encryption module, the second server, and the third server distribute access keys to the shared target based on the encryption module, and the shared target obtains teaching resources based on the access keys.

[0097] It should be understood that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments have been described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0098] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included outside the scope of protection of the present invention.

Claims

1. A method for sharing practical training resources based on artificial intelligence, characterized in that, Multiple storage servers are deployed, and a login management table is set up in each of the storage servers. The login management table records the historical login information of each user. When a user uploads teaching resources, they select one of the storage servers as the first server according to the login management table. Based on the sharing goal of the teaching resources and the login management table, the first server selects a second server from the storage servers and synchronizes the teaching resources to the second server; The second server, based on the teaching resources it has stored and the login management table, combined with a prediction model, predicts the access pressure for multiple future time periods; If there are multiple future time periods where the access pressure exceeds the first threshold, a third server is selected from the storage servers, and a portion of the stored teaching resources are synchronized to the third server. The second server and the third server distribute access keys to the shared target, and the shared target obtains the teaching resources based on the access keys; Predicting the access pressure for the future time period includes the following steps: The login management table is used to extract the hourly access volume of each teaching resource. Based on the access volume, a daily access sequence is generated. The access sequence is clustered in units of years to obtain multiple clustering results. Based on the clustering results, multiple access patterns are divided for each teaching resource, and the center of the clustering results is used as the representative sequence of the access pattern. The teaching resources stored on the second server are obtained as resources to be analyzed. Based on the date of the future time period, the access pattern of each resource to be analyzed is obtained, and the representative sequence in the access pattern is obtained as the first prediction sequence. The monthly user change is obtained, the first prediction sequence is corrected based on the monthly user change to obtain the second prediction sequence, the second prediction sequences of different resources to be analyzed are merged to obtain the third prediction sequence, and the access volume of each hour in the third prediction sequence is used as the access pressure of the future time period. Clustering the access sequences includes the following steps: Different access sequences are traversed and combined to obtain multiple combination sequences. Each combination sequence includes two access sequences, which are defined as a first sequence and a second sequence, respectively. A difference sequence is calculated based on the first sequence and the second sequence. The difference sequence is the absolute value of the difference in access volume between the first sequence and the second sequence in the corresponding time period. A first feature is obtained after processing the difference sequence using the sliding window method. Filter the top N largest access counts in the first sequence and the second sequence respectively, and the corresponding first time period and second time period. Extract the third sequence corresponding to the first time period and the fourth sequence corresponding to the second time period from the difference sequence. Calculate the average value of the third sequence and the fourth sequence, and use them as the second feature and the third feature respectively. The access sequences for each month are clustered based on the first feature, the second feature, and the third feature.

2. The method according to claim 1, characterized in that, Selecting the third server from the storage server includes the following steps: Define the future time period in which the access pressure of the second server is greater than the first threshold as the migration time period. Take the sum of the access pressure of the storage servers in the migration time period as the first pressure value. Calculate the second pressure value of each storage server for the entire day in the migration time period. The login management table includes the user's access speed. Perform a weighted summation of the access speed, the first pressure value, and the second pressure value to obtain the evaluation value of the storage server. Select the storage server with the smallest evaluation value as the third server.

3. The method according to claim 1, characterized in that, The shared target obtains the teaching resources based on the access key, including the following steps: The second server and the third server generate a shared key and a one-dimensional initial matrix, generate a key sequence based on the shared key and the one-dimensional initial matrix, divide the key sequence into multiple sub-sequences, generate a storage address based on each sub-sequence, select one of the storage addresses as the target address, the second server and the third server generate a random key, encrypt the teaching resources based on the random key to obtain encrypted data, and store the encrypted data and the random key in different storage addresses, which are defined as the first address and the second address, respectively. Based on the access time of the shared target, a target server is selected from the second server and the third server. The shared target obtains the teaching resources after decrypting the encrypted data of the first address based on the random key of the second address in the target server.

4. The method according to claim 3, characterized in that, After the shared target obtains the teaching resources, the target server regenerates the random key, uses the random key to re-encrypt the teaching resources and stores them in the first address, selects a third address from the storage addresses, and stores the regenerated random key in the third address.

5. The method according to claim 3, characterized in that, Generating the storage address based on the subsequence includes the following steps: The numerical values ​​included in the subsequence are converted into ASCII characters, and then the ASCII characters are converted into basic characters according to the corresponding multi-language character names. Based on the basic characters, a character sequence of the subsequence is generated, and a conversion rule table is established. The conversion rule table includes the correspondence between each character sequence and the storage address.

6. The method according to claim 1, characterized in that, If the teaching resources are not available on the third server, a request is made to retrieve the teaching resources from the second server.

7. The method according to claim 1, characterized in that, The access sequences are clustered using the K-means or DBSCAN algorithm.

8. A resource-sharing system for practical training based on artificial intelligence, used to implement the method as described in any one of claims 1-7, characterized in that, The storage module includes multiple storage servers, and a login management table is set in each of the storage servers. The login management table records the historical login information of each user. In the upload module, when a user uploads teaching resources, the upload module selects a storage server as the first server based on the login management table. The first server selects a second server from the storage servers based on the sharing target of the teaching resources and the login management table, and synchronizes the teaching resources to the second server. The optimization module, based on the teaching resources it has stored and the login management table, predicts the access pressure for multiple future time periods using a prediction model; if there are multiple future time periods where the access pressure is greater than a first threshold, the optimization module selects a third server from the storage server and synchronizes a portion of the stored teaching resources to the third server. An encryption module is provided, and the second server and the third server distribute access keys to the shared target based on the encryption module, and the shared target obtains the teaching resources based on the access keys.

Images