Optimization Method for Educational Resource Storage Scheduling Based on Multimodal Hybrid Reasoning Perception

Abstract

The present invention discloses an optimization method for educational resource storage scheduling based on multi-modal hybrid reasoning perception. The method includes: obtaining real-time hybrid teaching scenario data and constructing an organizational representation of the hybrid teaching scenario data; determining the logical symbols and knowledge symbols of the hybrid teaching scenario data according to the hybrid teaching scenario data; constructing an educational resource access feature inference model under complex scenarios according to the logical symbols and knowledge symbols of the hybrid teaching scenario data; and optimizing the storage scheduling of cloud-edge-end multi-level collaborative educational resources according to the educational resource access feature inference model. The present invention can optimize the storage scheduling of educational resources at cloud-edge-end with a wide geographical distribution and can be widely applied to the field of computer technology.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to an optimization method for storing and scheduling educational resources based on multi-modal hybrid reasoning perception. Background Art

[0002] With the rapid development of multimedia technology, educational digital resources have also shown a blowout growth. Due to the wide geographical distribution of end-users, there are significant differences in resource access types and access times. At the same time, in the case of limited data storage space and network bandwidth resources, adopting a multi-level heterogeneous cloud-edge-end data storage architecture is a good choice. Compared with other related fields, there are significant characteristics such as significant aggregation and topicality in user data access in the field of educational application scenarios, and these characteristics can be used to optimize the storage and scheduling of educational resources in the cloud-edge-end. However, most current studies are based on the perception of access pattern features from user online behavior data, which is difficult to support the optimization of the storage and scheduling of digital resources in the hybrid educational scenario of online learning and offline teaching, affecting the resource access efficiency in the hybrid learning environment. Summary of the Invention

[0003] In view of this, an embodiment of the present invention provides an optimization method for storing and scheduling educational resources based on multi-modal hybrid reasoning perception, which can optimize the storage and scheduling of educational resources in the cloud-edge-end with a wide geographical distribution.

[0004] An aspect of the embodiment of the present invention provides an optimization method for storing and scheduling educational resources based on multi-modal hybrid reasoning perception, including:

[0005] Obtain real-time hybrid teaching scenario data and construct an organizational representation of the hybrid teaching scenario data;

[0006] Determine the logical symbols and knowledge symbols of the hybrid teaching scenario data according to the hybrid teaching scenario data;

[0007] Construct an inference model for educational resource access characteristics in complex scenarios according to the logical symbols and knowledge symbols of the hybrid teaching scenario data;

[0008] Optimize the storage and scheduling of multi-level collaborative educational resources in the cloud-edge-end according to the inference model for educational resource access characteristics.

[0009] Optionally, the obtaining real-time hybrid teaching scenario data and constructing an organizational representation of the hybrid teaching scenario data includes:

[0010] Obtain course activity arrangement data, teaching behavior data, and learning behavior data in a hybrid learning environment as the first type of data; wherein, the content of the first type of data includes online behavior text logs, course arrangement data, and offline classroom multi-modal data;

[0011] Obtain online resource tag-related data and access-related data as the second type of data; wherein, the content of the second type of data includes tag information of online education resources and access information of online education resources.

[0012] According to audio-to-text technology, convert the audio data in the first type of data and the second type of data into text data.

[0013] According to the teaching domain ontology library, use the teaching event definition rules in the text data and the teaching event definition rules in the video data to define teaching events for the text data and video data in the first type of data and the second type of data.

[0014] According to the state logic expression mechanism, based on the defined teaching events and the first type of data and the second type of data, construct a data organization method for dynamic perception of cross-time-domain hybrid teaching scenarios.

[0015] Optionally, the determining the logical symbols and knowledge symbols of the hybrid teaching scenario data according to the hybrid teaching scenario data includes:

[0016] According to the hybrid teaching scenario data, determine the set of temporal logic descriptions supported by common multi-modalities in the field.

[0017] According to the tensorization requirements of the set of temporal logic descriptions, improve the logical tensor network to determine the logical symbols of the hybrid teaching scenario data.

[0018] Through the knowledge graph logicization method, perform knowledge enhancement design on the hybrid teaching scenario data to determine the knowledge symbols of the hybrid teaching scenario data.

[0019] Optionally, the constructing an educational resource access feature inference model in a complex scenario according to the logical symbols and knowledge symbols of the hybrid teaching scenario data includes:

[0020] According to the logical symbols and knowledge symbols of the hybrid scenario data, model the educational resource access feature process according to the timestamp information in the data.

[0021] According to the logical symbols and knowledge symbols of the hybrid scenario data, infer the educational resource access features to determine the potential educational resource access requirements for multi-level teaching event mapping.

[0022] According to the storage scheduling requirements of educational resources at different levels of cloud-edge-terminal, adjust the events and time granularity in the hybrid scenario data, and construct an educational resource access prediction model after obtaining the fine-grained results.

[0023] Optionally, the storage scheduling optimization of the cloud-edge multi-level collaborative educational resources according to the educational resource access feature inference model includes:

[0024] At the edge side, with the help of multi-media sensing devices, the real-time collection of the first type of data is carried out, and then the text teaching events and video teaching events are extracted by small-area intelligent computing devices. The corresponding original multi-modal data set is marked with corresponding timestamps and event tags, and then transmitted to the central area.

[0025] In the middle area, the real-time collection of the second type of data is carried out by using the log system. Combining the processing results of the first type of collected data transmitted from the small area, the data organization and processing of the educational resource access features are completed, and then the processing results are transmitted to the computing server in the large area.

[0026] The computing server in the large area counts the time and events of the processing results of the collected data from the middle area. If the time and event data of receiving the processing results are greater than the set first threshold, and the resource access inference accuracy rate is lower than the specific second threshold, then the solution task of the resource access statistical information impact metric function is run on the computing server in the large area. According to the solution results, counterfactual information extraction is carried out through a specific intention optimization method.

[0027] Optionally, the storage scheduling optimization of the cloud-edge multi-level collaborative educational resources according to the educational resource access feature inference model further includes:

[0028] Send the optimized educational resource access inference model to the computing servers in the central area and the small area to replace the original resource access prediction model.

[0029] When the computing servers in the central area and the small area obtain the educational resource access prediction model, the resource access times of the educational resource files in the resource storage service in their respective areas are predicted.

[0030] At the small area level, according to the prediction results of the educational resource prediction model, calculate the cache demand value of the resource file. According to the situation of the resource storage server in this area, the educational resources with a cache demand value greater than a specific value are stored and cached. When the storage capacity is insufficient, the old cached files are deleted.

[0031] At the middle area level, simultaneously converge the access prediction results of the educational resource files of the central storage server and the small area storage servers under its jurisdiction, calculate the replica factor value of the resource file, and generate and delete the replicas of the educational resource files in the middle area according to the size of the replica factor value of the file.

[0032] Set a specific time period and trigger conditions, and repeat to complete the dynamic optimization of the cloud-edge educational resource storage scheduling based on the multi-modal hybrid inference scenario perception.

[0033] Optionally, the calculation of the cache requirement value is performed based on the predicted number of accesses, the size of the file, or the access latency.

[0034] The calculation of the replica factor is performed based on the predicted number of accesses, the network bandwidth of the service center, the size of the file, or the storage capacity of the service center.

[0035] Another aspect of the embodiments of the present invention further provides an educational resource storage scheduling optimization device based on multi-modal hybrid inference perception, including:

[0036] A first module, configured to obtain real-time hybrid teaching scenario data and construct an organizational representation of the hybrid teaching scenario data;

[0037] A second module, configured to determine the logical symbols and knowledge symbols of the hybrid teaching scenario data according to the hybrid teaching scenario data;

[0038] A third module, configured to construct an educational resource access feature inference model in a complex scenario according to the logical symbols and knowledge symbols of the hybrid teaching scenario data;

[0039] A fourth module, configured to optimize the storage scheduling of cloud-edge multi-level collaborative educational resources according to the educational resource access feature inference model.

[0040] Another aspect of the embodiments of the present invention further provides an electronic device, including a processor and a memory;

[0041] The memory is used to store programs;

[0042] The processor executes the program to implement the method described above.

[0043] Another aspect of the embodiments of the present invention further provides a computer-readable storage medium, where the storage medium stores a program, and the program is executed by a processor to implement the method described above.

[0044] The embodiments of the present invention also disclose a computer program product or a computer program, which includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. The processor of the computer device can read the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions to enable the computer device to execute the method described above.

[0045] Embodiments of the present invention obtain real-time hybrid teaching scenario data and construct an organizational representation of the hybrid teaching scenario data; determine the logical symbols and knowledge symbols of the hybrid teaching scenario data according to the hybrid teaching scenario data; construct an inference model for educational resource access characteristics in complex scenarios according to the logical symbols and knowledge symbols of the hybrid teaching scenario data; and optimize the storage scheduling of multi-level collaborative educational resources at the cloud-edge-end according to the educational resource access characteristics inference model. The present invention can optimize the storage scheduling of geographically widespread cloud-edge-end educational resources. Brief Description of the Drawings

[0046] To more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the accompanying drawings required for the description of the embodiments. Obviously, the accompanying drawings in the following description are only some embodiments of the present application, and those of ordinary skill in the art can obtain other drawings based on these drawings without creative efforts.

[0047] Figure 1 It is the overall step flow chart of the embodiments of the present invention. Detailed Embodiments

[0048] To make the objectives, technical solutions, and advantages of the present application clearer, the following further elaborates on the present application in combination with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0049] In view of the problems existing in the prior art, one aspect of the embodiments of the present invention provides an educational resource storage scheduling optimization method based on multi-modal hybrid inference perception, including:

[0050] Obtain real-time hybrid teaching scenario data and construct an organizational representation of the hybrid teaching scenario data;

[0051] Determine the logical symbols and knowledge symbols of the hybrid teaching scenario data according to the hybrid teaching scenario data;

[0052] Construct an inference model for educational resource access characteristics in complex scenarios according to the logical symbols and knowledge symbols of the hybrid teaching scenario data;

[0053] Optimize the storage scheduling of multi-level collaborative educational resources at the cloud-edge-end according to the educational resource access characteristics inference model.

[0054] Optionally, the obtaining real-time hybrid teaching scenario data and constructing an organizational representation of the hybrid teaching scenario data includes:

[0055] Obtain course activity arrangement data, teaching behavior data, and learning behavior data in a blended learning environment as the first type of data; wherein, the content of the first type of data includes online behavior text logs, course arrangement data, and offline classroom multimodal data;

[0056] Obtain online resource tag-related data and access-related data as the second type of data; wherein, the content of the second type of data includes tag information of online education resources and access information of online education resources;

[0057] According to audio-to-text technology, convert the audio data in the first type of data and the second type of data into text data;

[0058] According to the teaching domain ontology library, use the teaching event definition rules in the text data and the teaching event definition rules in the video data to define teaching events for the text data and video data in the first type of data and the second type of data;

[0059] According to the state logic expression mechanism, based on the defined teaching events and the first type of data and the second type of data, construct a data organization method for dynamic perception of cross-temporal blended teaching scenarios.

[0060] Optionally, the determining the logical symbols and knowledge symbols of the blended teaching scenario data according to the blended teaching scenario data includes:

[0061] According to the blended teaching scenario data, determine the set of temporal logic descriptions supported by common multimodality in the field;

[0062] According to the tensorization requirements of the set of temporal logic descriptions, improve the logical tensor network to determine the logical symbols of the blended teaching scenario data;

[0063] Through the knowledge graph logicization method, perform knowledge enhancement design on the blended teaching scenario data to determine the knowledge symbols of the blended teaching scenario data.

[0064] Optionally, the constructing an educational resource access feature inference model in a complex scenario according to the logical symbols and knowledge symbols of the blended teaching scenario data includes:

[0065] According to the logical symbols and knowledge symbols of the blended scenario data, model the educational resource access feature process according to the timestamp information in the data;

[0066] According to the logical symbols and knowledge symbols of the blended scenario data, infer the educational resource access features to determine the potential educational resource access requirements for multi-level teaching event mapping;

[0067] According to the storage scheduling requirements of educational resources at different levels of cloud, edge, and terminal, adjust the events and time granularity in the hybrid scenario data, and construct an educational resource access prediction model after obtaining the fine-grained results.

[0068] Optionally, the storage scheduling optimization of multi-level collaborative educational resources at cloud, edge, and terminal according to the educational resource access feature inference model includes:

[0069] At the edge, use multi-media sensing devices to collect the first type of data in real time, and then complete the extraction of text teaching events and video teaching events on small-area intelligent computing devices. Mark the corresponding original multi-modal data set with corresponding timestamps and event tags, and transmit it to the central area;

[0070] In the middle area, use the logging system to collect the second type of data in real time, combine the processing results of the first type of collected data transmitted from the small area, complete the data organization and processing of educational resource access features, and then transmit the processing results to the computing server in the large area;

[0071] The computing server in the large area counts the time of the processing results of the collected data from the middle area. If the time of receiving the processing results and the data of the event are greater than the set first threshold, and the resource access inference accuracy rate is lower than the specific second threshold, then run the solution task of the resource access statistical information impact metric function on the computing server in the large area, and extract counterfactual information through a specific intention optimization method according to the solution results.

[0072] Optionally, the storage scheduling optimization of multi-level collaborative educational resources at cloud, edge, and terminal according to the educational resource access feature inference model further includes:

[0073] Send the optimized educational resource access inference model to the computing servers in the central area and the small area to replace the original resource access prediction model;

[0074] When the computing servers in the central area and the small area obtain the educational resource access prediction model, predict the number of resource accesses for the educational resource files in the resource storage service in their respective areas;

[0075] At the small area level, according to the prediction results of the educational resource prediction model, calculate the cache demand value of the resource file. According to the situation of the resource storage server in this area, store and cache the educational resources whose cache demand value is greater than a specific value, and delete the old cache files when the storage capacity is insufficient;

[0076] At the medium - area level, the access prediction results of the educational resource files of the central storage server and the small - area storage servers under its jurisdiction are aggregated simultaneously, the replica factor value of the resource files is calculated, and the replica generation and deletion of the educational resource files in the medium area are performed according to the value of the replica factor of the files;

[0077] Set a specific time period and trigger conditions, and repeat to complete the dynamic optimization of the cloud - edge - end educational resource storage scheduling based on the multi - modal hybrid - reasoning scenario awareness.

[0078] Optionally, the calculation of the cache demand value is calculated by the predicted access times, the size of the file, or the access latency;

[0079] The calculation of the replica factor is calculated by the predicted access times, the network bandwidth of the service center, the size of the file, or the storage capacity of the service center.

[0080] Another aspect of the embodiments of the present invention also provides an educational resource storage scheduling optimization device based on multi - modal hybrid - reasoning perception, including:

[0081] The first module is used to obtain real - time hybrid teaching scenario data and construct an organizational representation of the hybrid teaching scenario data;

[0082] The second module is used to determine the logical symbols and knowledge symbols of the hybrid teaching scenario data according to the hybrid teaching scenario data;

[0083] The third module is used to construct an educational resource access feature inference model in a complex scenario according to the logical symbols and knowledge symbols of the hybrid teaching scenario data;

[0084] The fourth module is used to optimize the storage scheduling of the cloud - edge - end multi - level collaborative educational resources according to the educational resource access feature inference model.

[0085] Another aspect of the embodiments of the present invention also provides an electronic device, including a processor and a memory;

[0086] The memory is used to store programs;

[0087] The processor executes the program to implement the method as described above.

[0088] Another aspect of the embodiments of the present invention also provides a computer - readable storage medium, where the storage medium stores a program, and the program is executed by a processor to implement the method as described above.

[0089] An embodiment of the present invention also discloses a computer program product or a computer program. The computer program product or the computer program includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. A processor of a computer device can read the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the foregoing method.

[0090] The following combines the description of the accompanying drawings of the specification to describe the specific implementation process of the present invention in detail:

[0091] Refer to Figure 1 , in order to solve the problems existing in the prior art, the present invention proposes a cloud-edge-end education resource storage scheduling optimization method based on multi-modal hybrid inference scene perception. With the help of multi-media sensing devices, the access rules of user resources are perceived through a multi-modal hybrid inference method, and the storage scheduling of cloud-edge-end education resources with a wide geographical distribution is optimized.

[0092] The following separately expands and explains each step:

[0093] 1. Real-time perception, acquisition, organization, and representation of hybrid teaching scenario data:

[0094] 1). Collection of multiple types of data:

[0095] The data obtained in real time by the present invention mainly includes two categories. The first category is data such as course activity arrangements and teaching and learning behavior data in a blended learning environment, and the second category is data related to online resource tags and access-related data.

[0096] The collection of the first category of data includes online behavior text logs (such as searched keywords and corresponding timestamps), course arrangement data (such as course outlines), and offline classroom multi-modal data. The collection of these multiple types of data in the hybrid teaching scenario is mainly completed by terminal multi-media sensing devices.

[0097] The collection of the second category of data includes relevant tag information of online education resources (such as text information such as resource types, resource themes, resource names, and resource introductions), and the access situation of online education resources (such as the number of accesses, access client address information, resource distribution, etc.). The collection of this type of data is mainly completed at the edge server of educational resources.

[0098] 2). Processing and organization of multiple types of data:

[0099] The present invention constructs a teaching and learning event tracking mechanism to perceive the blended teaching scenario, providing an information basis for inferring the access characteristics of educational resources in this scenario.

[0100] The teaching events of the present invention are mainly obtained from the text data and video data in the first type of collected data. Therefore, the audio will be directly texturized through the commonly used audio-to-text technology in the field. In the present invention, the set of teaching events extracted from the text data is denoted as The set of events extracted from the teaching video data is denoted as The event-event relationship set between the text events and the video events is denoted as where r k takes values from {same relationship, hierarchical relationship}.

[0101] Regarding the definition of teaching events for text data, in the present invention, it is mainly combined with the ontology library of the teaching field and refers to the event definition method of ACE in the field to define the text teaching events. Its annotation is mainly completed based on the BERT pre-trained model with tags.

[0102] Regarding the definition of teaching events for video data, it is also mainly based on the ontology library of the teaching field, and the CLIP model in the field is used to complete the label extraction work of the video teaching events.

[0103] Combining the above teaching event definition and extraction methods, a data organization method for dynamically perceiving cross-temporal hybrid teaching scenarios is constructed.

[0104] The present invention combines the Topos theory and the homotopy type theory, and uses the commonly used state logic expression mechanism in the field to establish a cross-temporal multi-modal data chain organization method centered on the inference of educational resource access characteristics. The data organization unit is abstractly described by the following equation,

[0105]

[0106] where, I t represents the teaching event description information of this data unit, mainly including information such as text teaching event and video teaching event tags, event relationship tags, etc. is the multi-modal perception data set, represents the set of temporal logic descriptions supported by commonly used multi-modal in the field, represents and The association information of is used for state logic expression, represents different The connection information between.

[0107] 2. Symbolic neural computing design of hybrid teaching scenario data

[0108] The educational resource access feature inference of the present invention involves a large amount of symbolic computation, mainly including two categories. One is the computation of logical symbols (mainly from state logical descriptions), and the other is the computation of knowledge symbols (mainly from knowledge graphs). The following introduces the design of the present invention for these two types of symbolic computations.

[0109] Logical symbol neural computation design:

[0110] For the element tensoring requirement in the equation in improve the logical tensor network, such as containing quantifiers and k free teaching behavior related factors x i The tensoring process of the logical sentence ψ can be expressed as:

[0111]

[0112] Among them, represents the tensoring function. The Gather operator processes the first k variables in each loop. Gather represents the gathering operator. See the equation specifically:

[0113]

[0114] Among them, represents the association with the quantifier

[0115] Knowledge symbol neural computation design:

[0116] Due to the cross-time and space characteristics of the hybrid teaching scenario, there is a problem of incomplete perception of the information in the hybrid learning process. The experience knowledge such as the rule base and the knowledge graph can be used to construct a domain knowledge enhanced hybrid teaching scenario perception and reasoning mechanism to provide fine-grained diagnostic information for the educational resource access feature inference.

[0117] The present invention mainly completes the knowledge enhancement design through the logicalization of the knowledge graph. Denote the relationship triple of the teaching behavior knowledge graph as Based on the Curry-Howard isomorphism theory, its logical proposition can be simply described as:

[0118]

[0119] Among them, represents the entity edge connection, represents the relationship, represents the predicate. It can be seen that ​The logical proposition has been completed. According to the characteristics of logicalization of such domain knowledge, the present invention performs propositional logic neural network expression based on a symmetric network and uses a content-addressable memory network architecture to achieve large-scale prior semantic knowledge-constrained regularization learning, which can provide solution space search optimization for the backward chaining inference mode in the following educational resource access feature inference process.

[0120] 3. Construction of an Educational Resource Access Feature Inference Model in Complex Scenarios

[0121] This part mainly solves the inference of educational resource access features, obtains the spatio-temporal laws of educational resource access, and provides support for the optimization of educational resource storage and scheduling at the cloud, edge, and terminal.

[0122] 1). Modeling of the Educational Resource Access Feature Process:

[0123] First, the second type of collected data is matched with the data organization unit of the equation Mainly using their respective timestamp information, the association between teaching events and educational resource access in the blended teaching scenario is completed. The data organization unit after association is denoted as

[0124]

[0125] Among them, y t represents the number of times a specific online educational resource is accessed in this data unit.

[0126] Then, combined with the forward chaining inference information, a cross-temporal online educational resource access feature inference method in the blended teaching scenario is constructed. Its general process can be described as follows.

[0127]

[0128] Among them, X T represents the time series data chain composed of, Y T represents the educational resource access statistical feature, Φ, Θ, W are learnable parameters, f logtensor represents the logical tensorization process of X t and so on, forward represents the forward chaining inference process (forwardchaining), f backward represents combining the forward chaining inference information, ROI(·) represents the input of the region of interest, which represents the teaching event that triggers the access to the online educational resource in the present invention.

[0129] 2). Inference of Educational Resource Access Features:

[0130] The present invention constructs an equation

[0131]

[0132] The fine-grained information acquisition method of ROI(·) analyzes the potential educational resource access requirements mapped by multi-level teaching events, and realizes the inference and analysis of educational resource access characteristics adapted to multiple types of educational application scenarios.

[0133] First, according to the requirements of resource access characteristic information granularity for different-level educational resource scheduling tasks, a resource access statistical information impact metric function for regular teaching events is constructed, and the function is defined as:

[0134] I s = Impact(Y T , X s , W, y * )

[0135] Among them, I s represents the impact metric with a granularity of s, W represents the optimization parameter, Y T represents the educational resource access statistical characteristics, such as the spatio-temporal aggregation or topicality of educational resource access, X s ∈X T represents the set of associated teaching event data to be evaluated, and y * represents the counterfactual inference expectation.

[0136] Since the present invention adopts the design of symbolic neural computing, many counterfactual information is relatively easy to calculate through counterfactual logical expressions. For the counterfactual information that is difficult to obtain, it is extracted through a specific intention optimization method, and the main extraction process is as follows:

[0137] Among them, represents the dynamic causal probability distribution at time t of the mixed teaching scenario data chain X s before counterfactual processing, represents the probability distribution after counterfactual processing, and l is the intention function. Based on a specific intention optimization algorithm, counterfactual information can be extracted from the experiment.

[0138] Combining equation and equation I s = Impact(Y T , X s , W, y * ), based on the neural-symbolic hybrid reasoning knowledge extraction mechanism and parameter optimization algorithm, the logical expression set of the impact on the current educational resource access after sorting can be output, and the impact time series logical expression set combined with the changes in the educational resource access status within T time instants can be obtained, completing the acquisition of ROI(X T ).

[0139] Educational resource access prediction model:

[0140] The above solution of the present invention can adjust X according to the storage scheduling requirements of educational resources at different levels of the cloud edge. s Event and time granularity to obtain fine-grained ROI (X T ) result. Combined with ROI(X T ) can be used to obtain the result of equation Perform supervised training and share the optimized parameters to the following equation:

[0141]

[0142] Among them, X′ T+1 It represents the embedding vector of the set of teaching events at the next moment obtained based on historical data or future teaching-related information (such as course outline), and MLP represents a multi-layer fully connected neural network.

[0143] 4. Optimization of storage scheduling of multi-level collaborative education resources in the cloud, edge and terminal:

[0144] The educational resource management system of the present invention adopts a three-level storage architecture of cloud, edge and end, and combines the above-mentioned hybrid teaching scene perception and educational resource access feature reasoning method to propose a cloud-edge-end collaborative educational resource storage scheduling optimization method.

[0145] The cloud-edge storage architecture in the present invention mainly refers to the three-level method of "small area storage", "medium area storage" and "large area storage". For example, the small area can be the town center, the medium area can be the city center, and the large area can be the provincial center. Generally, the larger the geographical coverage area, the stronger the network bandwidth and storage capacity of its data center. The educational resource storage scheduling optimization of the present invention utilizes the educational resource access feature inference information in the hybrid teaching environment to optimize the cloud-edge collaborative educational resource storage. Its main scheduling optimization process is as follows:

[0146] 1) At the edge, with the help of multi-media sensing equipment, the first type of data is collected in real time, mainly including mixed teaching multi-type data. The collected data and audio files are directly converted into text, and then the text teaching events and video teaching events are extracted on the small area intelligent computing device. The corresponding original multi-modal data set is marked with corresponding timestamps and event tags and transmitted to the central area;

[0147] 2) In the middle area, the log system is used to collect the second type of data in real time, and combined with the processed results of the first type of data transmitted from the small area, the equation is completed. The data is organized and processed, and then the processing results are transmitted to the computing servers in a large area;

[0148] 3) The computing server in the large area performs timing and counting on the processing results of the collected data from the middle area. If the time to receive the processing results and the data of the event are greater than the set threshold, and the resource access inference accuracy rate of the equation is lower than a specific threshold, then the computing server in the large area runs the solution task of the equation and performs incremental optimization training of the equation according to the results, and stops training when a specific prediction accuracy rate is met. Some of the above thresholds can be adjusted according to the actual application scenario;

[0149] 4) The optimized educational resource access inference model is sent to the computing servers in the central area and the small area to replace the original resource access prediction model;

[0150] 5) After the computing servers in the central area and the small area obtain the educational resource access prediction model, they predict the number of resource accesses for the educational resource files in the resource storage service in their respective areas;

[0151] 6) At the small area level, according to the prediction results of the educational resource prediction model, calculate the cache demand value of the resource file. The calculation of the cache demand value is based on various data such as the predicted number of accesses, the size of the file, and the access delay (the calculation method can be various, such as simple weighted average). According to the situation of the resource storage server in this area, the educational resources with a cache demand value greater than a specific value are stored in the cache, and the old cache files are deleted when the storage capacity is insufficient;

[0152] 7) At the middle area level, simultaneously aggregate the educational resource file access prediction results of the central storage server and the small area storage servers under its jurisdiction, and calculate the replica factor value of the resource file. The calculation of the replica factor is based on various data such as the predicted number of accesses, the network bandwidth of the service center, the size of the file, and the storage capacity of the service center (the calculation method can be various, such as simple weighted average). According to the value of the replica factor of the file, generate and delete the replicas of the educational resource files in the middle area according to the common practices in the field;

[0153] 8) Set a specific time period and trigger conditions, and repeat steps 1)-7) to complete the dynamic optimization of the cloud-edge-end educational resource storage scheduling based on multi-modal hybrid inference scenario perception.

[0154] In summary, the present invention proposes an optimization method for cloud-edge-end educational resource storage scheduling based on multi-modal hybrid inference scenario perception. With the help of multi-media sensing devices, it perceives the resource access rules of users through multi-modal hybrid inference methods for online and offline teaching behaviors, and optimizes the cloud-edge-end educational resource storage scheduling with a wide geographical distribution.

[0155] In some alternative embodiments, the functions / operations recited in the block diagrams may not occur in the order presented in the operational illustrations. For example, depending on the functions / operations involved, two consecutive blocks shown may actually be executed substantially simultaneously or the blocks may sometimes be executed in reverse order. Further, the embodiments presented and described in the flowcharts of the present invention are provided by way of example to provide a more thorough understanding of the technology. The disclosed methods are not limited to the operations and logical flows presented herein. Alternative embodiments are envisioned in which the order of various operations is altered and where sub-operations described as part of a larger operation are performed independently.

[0156] Moreover, while the present invention has been described in the context of functional modules, it should be understood that, unless otherwise stated to the contrary, one or more of the described functions and / or features may be integrated in a single physical device and / or software module, or one or more functions and / or features may be implemented in separate physical devices or software modules. It should also be understood that a detailed discussion of the actual implementation of each module is not necessary for an understanding of the present invention. Rather, given the attributes, functions, and internal relationships of the various functional modules in the devices disclosed herein, the actual implementation of the modules will be understood within the ordinary skill of an engineer. Thus, those of ordinary skill in the art will be able to implement the present invention as set forth in the claims without undue experimentation. It should also be understood that the particular concepts disclosed are illustrative only and are not intended to limit the scope of the present invention, the scope of which is determined by the full scope of the appended claims and their equivalents.

[0157] If the functions are implemented in the form of software functional units and sold or used as independent products, they may be stored in a computer-readable storage medium. Based on such understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of such technical solution, may be embodied in the form of a software product stored in a storage medium, including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical discs.

[0158] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus or device (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus or device), or used in combination with these instruction execution systems, apparatus or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate or transport a program for use by or in connection with an instruction execution system, apparatus or device.

[0159] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection part (electronic device) having one or more wirings, a portable computer disk cartridge (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, then editing, interpreting or otherwise processing it as appropriate, and then storing it in a computer memory.

[0160] It should be understood that various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0161] In the description of this specification, the description with reference to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples", etc. means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0162] Although embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the claims and their equivalents.

[0163] The above has specifically described the preferred embodiments of the present invention, but the present invention is not limited to the described embodiments. Those skilled in the art can also make various equivalent deformations or substitutions without departing from the spirit of the present invention, and these equivalent deformations or substitutions are all included within the scope defined by the claims of this application.

Claims

1. An optimization method for educational resource storage scheduling based on multi-modal hybrid reasoning perception, characterized in that Including: Obtain real-time hybrid teaching scenario data and construct an organizational representation of the hybrid teaching scenario data; Determine the logical symbols and knowledge symbols of the hybrid teaching scenario data according to the hybrid teaching scenario data; Construct an educational resource access feature inference model under complex scenarios according to the logical symbols and knowledge symbols of the hybrid teaching scenario data; Optimize the storage scheduling of cloud-edge multi-level collaborative educational resources according to the educational resource access feature inference model; The obtaining of real-time hybrid teaching scenario data and the construction of the organizational representation of the hybrid teaching scenario data include: Obtain curriculum activity arrangement data, teaching behavior data, and learning behavior data in a blended learning environment as the first type of data; wherein, the content of the first type of data includes online behavior text logs, curriculum arrangement data, and offline classroom multimodal data; Obtain online resource label-related data and access-related data as the second type of data; wherein, the content of the second type of data includes label information of online educational resources and access information of online educational resources; Convert the audio data in the first type of data and the second type of data into text data according to audio-to-text technology; According to the teaching domain ontology library, define teaching events for the text data and video data in the first type of data and the second type of data by using the teaching event definition rules in the text data and the teaching event definition rules in the video data; Construct a data organization method for dynamic perception of cross-time-domain hybrid teaching scenarios based on the defined teaching events and the first type of data and the second type of data according to the state logic expression mechanism; The determining of the logical symbols and knowledge symbols of the hybrid teaching scenario data according to the hybrid teaching scenario data includes: Determine a set of temporal logic descriptions supported by commonly used multimodals in the field according to the hybrid teaching scenario data; Improve the logical tensor network according to the tensoring requirements of the set of temporal logic descriptions to determine the logical symbols of the hybrid teaching scenario data; Determine the knowledge symbols of the hybrid teaching scenario data through knowledge graph logicalization for knowledge enhancement design of the hybrid teaching scenario data; The constructing of an educational resource access feature inference model under complex scenarios according to the logical symbols and knowledge symbols of the hybrid teaching scenario data includes: Model the educational resource access feature process according to the logical symbols and knowledge symbols of the hybrid scenario data and the timestamp information in the data; Infer the educational resource access features according to the logical symbols and knowledge symbols of the hybrid scenario data to determine the potential educational resource access requirements for multi-level teaching event mapping; Adjust the events and time granularity in the hybrid scenario data according to the storage scheduling requirements of different levels of cloud-edge educational resources, and construct an educational resource access prediction model after obtaining the fine-grained results.

2. The optimized method for storing and scheduling educational resources based on multi-modal hybrid reasoning perception according to claim 1, wherein The optimizing of the storage scheduling of cloud-edge multi-level collaborative educational resources according to the educational resource access feature inference model includes: At the edge side, a multi-media sensing device is used to collect the first type of data in real time. Then, in the small-area intelligent computing device, text teaching events and video teaching events are extracted, and the corresponding original multi-modal data set is marked with corresponding timestamps and event tags, and then transmitted to the central area; In the middle area, a log system is used to collect the second type of data in real time. Combining the processing results of the first type of collected data transmitted from the small area, the data organization and processing of the education resource access characteristics are completed, and then the processing results are transmitted to the computing server in the large area; The computing server in the large area performs timing and counting on the processing results of the collected data from the middle area. If the time and event data of receiving the processing results are greater than the set first threshold, and the resource access inference accuracy rate is lower than the specific second threshold, then the computing server in the large area runs the solution task of the resource access statistical information impact measurement function. According to the solution results, counterfactual information extraction is performed through a specific intention optimization method.

3. The optimization method for educational resource storage scheduling based on multi-modal hybrid reasoning perception according to claim 2, wherein According to the education resource access characteristic inference model, optimizing the storage scheduling of cloud-edge multi-level collaborative education resources further includes: Sending the optimized education resource access inference model to the computing servers in the central area and the small area to replace the original resource access prediction model; When the computing servers in the central area and the small area obtain the education resource access prediction model, they predict the number of resource accesses for the education resource files in the resource storage service in their respective areas; At the small area level, according to the prediction results of the education resource prediction model, the cache demand value of the resource file is calculated. According to the situation of the resource storage server in this area, the education resources with a cache demand value greater than a specific value are stored in the cache, and the old cache files are deleted when the storage capacity is insufficient; At the middle area level, while aggregating the education resource file access prediction results of the central storage server and the small area storage servers under its jurisdiction, the replica factor value of the resource file is calculated, and the replica generation and deletion of the education resource files in the middle area are performed according to the size of the replica factor value of the file; Set a specific time period and trigger conditions, and repeat to complete the dynamic optimization of the cloud-edge education resource storage scheduling based on multi-modal hybrid inference scenario perception.

4. The education resource storage scheduling optimization method based on multi-modal hybrid inference perception according to claim 3, wherein The calculation of the cache demand value is calculated by the predicted number of accesses, the size of the file or the access delay; The calculation of the replica factor is calculated by the predicted number of accesses, the network bandwidth of the service center, the size of the file or the storage capacity of the service center.

5. An apparatus for implementing an optimized method for storing and scheduling educational resources based on multi-modal hybrid inference perception as described in any one of claims 1-4, characterized in that, It includes: The first module is used to obtain real-time hybrid teaching scenario data and construct an organizational representation of the hybrid teaching scenario data; The second module is used to determine the logical symbols and knowledge symbols of the hybrid teaching scenario data according to the hybrid teaching scenario data; The third module is used to construct an education resource access characteristic inference model in a complex scenario according to the logical symbols and knowledge symbols of the hybrid teaching scenario data; The fourth module is used to optimize the storage scheduling of cloud-edge multi-level collaborative education resources according to the education resource access characteristic inference model.

6. An electronic device, characterized in that, It includes a processor and a memory; The memory is used to store programs; The processor executes the program to implement the method described in any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, The storage medium stores a program, and the program is executed by the processor to implement the method described in any one of claims 1 to 4.

Images