Method and system for scheduling cloud terminal call center data based on edge computing

Abstract

The invention relates to the technical field of data processing, and discloses a scheduling method and system for realizing cloud terminal call center data based on edge computing, and the method comprises the steps: carrying out the feature extraction of a data scheduling request, obtaining a feature descriptor, and calling request association data from a database of a cloud terminal call center; calculating the adaptation degree between the multimode integrated semantics and the edge computing nodes, and screening out data execution nodes corresponding to the request associated data from the edge computing nodes; calculating a channel noise ratio between the data scheduling equipment and the node server, and calculating a communication throughput rate between the data scheduling equipment and the node server; calculating the total load of task data when the request associated data is transmitted, and constructing a data transmission path between the data scheduling equipment and the node server; and executing scheduling processing on the request associated data by using the data scheduling equipment and the node server to obtain a scheduling result. The scheduling efficiency of the cloud terminal call center data can be improved.

Description

Technical Field

[0001] The present invention relates to a scheduling method and system for cloud terminal call center data based on edge computing, belonging to the technical field of data processing. Background Art

[0002] Cloud terminal call refers to a call center system built based on cloud computing technology. By deploying the hardware devices, computing resources, and software functions of the traditional call center in the cloud, various forms of customer call interaction services such as telephone, online customer service, and video calls are realized, mainly for data scheduling according to user requests.

[0003] The existing scheduling method for cloud terminal call center data is a centralized cloud computing scheduling method. This method transmits all data such as voice, text, and video generated by the call center to the cloud server, which is uniformly stored, analyzed, and task-allocated by the cloud. It relies on pre-set scheduling rules, such as evenly allocating tasks according to the polling strategy, or simply classifying and scheduling according to the skill tags of customer service personnel. At the same time, resource allocation is also centrally managed by the cloud, and computing and storage resources are planned based on the traffic volume estimated from historical data. However, this method does not consider the scheduling adaptability during data transmission, which leads to unreasonable resource allocation. It does not consider the transmission performance of the server during the transmission process, resulting in low data scheduling efficiency. Therefore, a method capable of improving the data scheduling efficiency of the cloud terminal call center is needed. Summary of the Invention

[0004] The present invention provides a scheduling method and system for cloud terminal call center data based on edge computing, and its main purpose is to improve the data scheduling efficiency of the cloud terminal call center.

[0005] To achieve the above purpose, a scheduling method for cloud terminal call center data based on edge computing provided by the present invention includes: Obtain the data scheduling request and edge computing nodes of the cloud terminal call center, extract features from the data scheduling request to obtain a feature descriptor, and based on the feature descriptor, retrieve the request-related data from the database of the cloud terminal call center; Perform semantic fusion processing on the feature descriptor and the request-related data to obtain a multi-modal integrated semantics, calculate the fitness between the multi-modal integrated semantics and the edge computing nodes, and based on the fitness, screen out the data execution nodes corresponding to the request-related data from the edge computing nodes; Query the data scheduling device and node server corresponding to the data execution node, calculate the channel noise ratio between the data scheduling device and the node server, and based on the channel noise ratio, calculate the communication throughput rate between the data scheduling device and the node server; Calculate the total task data payload of the request-related data during transmission, and construct a data transmission path between the data scheduling device and the node server in combination with the total task data payload and the communication throughput rate; Set the scheduling priority corresponding to the request-related data, and perform scheduling processing on the request-related data by using the data scheduling device and the node server in combination with the scheduling priority and the data transmission path to obtain a scheduling result.

[0006] Optionally, the feature extraction of the data scheduling request to obtain a feature descriptor includes: Perform protocol format parsing processing on the data scheduling request to obtain a structured request protocol; Analyze the request protocol parameters in the structured request protocol and identify the request attribute identifiers in the request protocol parameters; Perform standardization conversion processing on the request attribute identifiers to obtain a normalized feature value; Perform structure encapsulation processing on the normalized feature value to obtain a feature descriptor.

[0007] Optionally, the retrieving the request-related data from the database of the cloud terminal call center based on the feature descriptor includes: Obtain the data index identifier in the database and set the retrieval rule of the feature descriptor in the database; Perform compliance verification processing on the retrieval rule to obtain a compliant retrieval rule; Set the retrieval query instruction corresponding to the compliant retrieval rule, and perform an association mapping process on the retrieval query instruction and the data index identifier to obtain an index-associated query condition; Perform execution optimization processing on the index-associated query condition to obtain an optimal query execution condition; Perform distributed routing processing on the optimal query execution condition to obtain parallel retrieval subtasks; Retrieve the request-related data from the database based on the parallel retrieval subtasks.

[0008] Optionally, the semantic fusion processing of the feature descriptor and the request-related data to obtain a multi-modal integrated semantics includes: Perform vector space mapping processing on the feature descriptor to obtain a service feature vector; Perform multi-dimensional feature extraction on the request-related data to obtain multi-dimensional data features; Calculate the semantic association weight between the service feature vector and the multi-dimensional data features; Based on the semantic association weights, perform feature alignment processing on the service feature vector and the multi-dimensional data features to obtain heterogeneous alignment features; Perform semantic fusion processing on the heterogeneous alignment features to obtain multi-modal integrated semantics.

[0009] Optionally, calculating the fitness between the multi-modal integrated semantics and the edge computing node includes: Query the node multi-dimensional information corresponding to the edge computing node, and extract the key node information in the node multi-dimensional information; Filter out the node core features in the key node information, perform requirement parsing processing on the multi-modal integrated semantics to obtain semantic service requirements; Calculate the similarity between the node core features and the semantic service requirements to obtain a similarity matrix; Based on the similarity matrix, calculate the fitness between the multi-modal integrated semantics and the edge computing node.

[0010] Optionally, calculating the channel noise ratio between the data scheduling device and the node server includes: Calculate the channel noise ratio between the data scheduling device and the node server through the following formula: ; where A represents the channel noise ratio between the data scheduling device and the node server, represents the transmission power between the i-th data scheduling device and the e-th node server, represents the channel coefficient between the i-th data scheduling device and the e-th node server, represents the maximum transmission power, represents the transmission power of other links except the link between the i-th data scheduling device and the e-th node server, E represents the noise power, and i and e respectively represent the serial numbers corresponding to the data scheduling device and the node server.

[0011] Optionally, calculating the communication throughput between the data scheduling device and the node server based on the channel noise ratio includes: Calculate the communication throughput between the data scheduling device and the node server through the following formula: ; where, represents the communication throughput between the data scheduling device and the node server, F represents the total bandwidth between the data scheduling device and the node server, G represents the number of associated devices corresponding to the node server, and A represents the channel noise ratio.

[0012] Optionally, calculating the total task data payload of the request-associated data during transmission includes: Structurally parsing the request-associated data to obtain a data field list; Sensitively identifying the data field list to obtain sensitive data fields; Based on the sensitive data fields, determining the data to be encrypted in the request-associated data; Based on the data to be encrypted, encrypting the request-associated data to obtain encrypted associated data; Calculating the task data volume corresponding to the encrypted associated data and analyzing the data key length corresponding to the encrypted associated data; Combining the task data volume and the data key length, calculating the total task data payload of the request-associated data during transmission through the following formula: ; where H represents the total task data payload of the request-associated data during transmission, represents the task data volume corresponding to the h-th encrypted associated data, represents the data equal-share amount corresponding to the h-th encrypted associated data, represents the data key length corresponding to the h-th encrypted associated data.

[0013] Optionally, setting the scheduling priority corresponding to the request-associated data includes: Performing metadata parsing on the request-associated data to obtain a basic attribute set; Based on a preset time-sensitive rule library, filtering the basic attribute set to obtain target sensitive attributes; Extracting the time constraint words in the target sensitive attributes and analyzing the semantic meaning of the constraint words corresponding to the time constraint words; Based on the semantic meaning of the constraint words, determining the emergency quantization index corresponding to the request-associated data; Extracting the data timestamp corresponding to the request-associated data and calculating the data timeliness index corresponding to the request-associated data based on the data timestamp; Combining the emergency quantization index and the data expiration index, setting the scheduling priority corresponding to the request-associated data.

[0014] To solve the above problems, the present invention also provides a scheduling system for cloud terminal call center data based on edge computing, and the system includes: A data retrieval module, configured to obtain a data scheduling request and an edge computing node of a cloud terminal call center, extract features from the data scheduling request to obtain a feature descriptor, and retrieve request-associated data from a database of the cloud terminal call center based on the feature descriptor; An edge node screening module, configured to perform semantic fusion processing on the feature descriptor and the request-associated data to obtain a multi-modal integrated semantics, calculate a fitness between the multi-modal integrated semantics and the edge computing node, and screen out a data execution node corresponding to the request-associated data from the edge computing nodes based on the fitness; A communication throughput rate calculation module, configured to query a data scheduling device and a node server corresponding to the data execution node, calculate a channel noise ratio between the data scheduling device and the node server, and calculate a communication throughput rate between the data scheduling device and the node server based on the channel noise ratio; A data transmission path construction module, configured to calculate a total task data payload during transmission of the request-associated data, and construct a data transmission path between the data scheduling device and the node server in combination with the task data payload and the communication throughput rate; A data scheduling processing module, configured to set a scheduling priority corresponding to the request-associated data, and perform scheduling processing on the request-associated data by using the data scheduling device and the node server in combination with the scheduling priority and the data transmission path to obtain a scheduling result.

[0015] Compared with the problems described in the background art, the present invention can obtain a feature descriptor of the core attributes of the scheduling requirements by extracting features from the data scheduling request, thereby providing a scientific basis for efficiently retrieving request-associated data. Further, the present invention performs semantic fusion processing on the feature descriptor and the request-associated data to obtain a multi-modal integrated semantics, which can achieve a deep semantic alignment between the feature descriptor and the request-associated data, providing data support for the precise adaptation of edge computing nodes and improving the intelligent level of data scheduling. By calculating the channel noise ratio between the data scheduling device and the node server, the present invention can understand the signal interference degree of the link during the data transmission between the data scheduling device and the node server, providing data support for the subsequent calculation of the communication throughput rate. Further, by calculating the total task data payload of the request-associated data during transmission, the present invention can understand the total amount of data in the data transmission of the request-associated data, providing data support for constructing the data transmission path between the data scheduling device and the node server. By setting the scheduling priority corresponding to the request-associated data, the present invention can ensure the priority scheduling of high-timeliness and high-urgency data, effectively reducing the key service delay and improving the overall system response efficiency, thereby achieving the purpose of improving the data transmission efficiency. Therefore, the scheduling method and system for cloud terminal call center data based on edge computing provided by the embodiments of the present invention can improve the scheduling efficiency of cloud terminal call center data. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 FIG. is a schematic flowchart of a scheduling method for cloud terminal call center data based on edge computing provided by an embodiment of the present invention; Figure 2 FIG. is a schematic diagram of scheduling processing in the scheduling method for cloud terminal call center data based on edge computing provided by the present invention; Figure 3 FIG. is a schematic diagram of a module for implementing the scheduling method for cloud terminal call center data based on edge computing provided by an embodiment of the present invention.

[0017] The implementation, functional features, and advantages of the present invention will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0019] An embodiment of the present application provides a method for scheduling cloud terminal call center data based on edge computing. The execution subject of the method for scheduling cloud terminal call center data based on edge computing includes, but is not limited to, at least one of electronic devices such as a server, a terminal, etc. that can be configured to execute the method provided by the embodiment of the present application. In other words, the method for scheduling cloud terminal call center data based on edge computing can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to: a single server, a server cluster, a cloud server, or a cloud server cluster, etc.

[0020] Embodiment 1: Refer to Figure 1 As shown, it is a schematic flowchart of a method for scheduling cloud terminal call center data based on edge computing provided by an embodiment of the present invention. In this embodiment, the method for scheduling cloud terminal call center data based on edge computing includes: S1. Obtain a data scheduling request and an edge computing node of the cloud terminal call center, extract features from the data scheduling request to obtain a feature descriptor, and based on the feature descriptor, retrieve request-related data from the database of the cloud terminal call center.

[0021] By extracting features from the data scheduling request, the present invention can obtain a feature descriptor of the core attributes of the scheduling requirements, thereby providing a scientific basis for efficiently retrieving request-related data. Among them, the data scheduling request is a data processing requirement generated during the operation of the cloud terminal call center, the edge computing node is a distributed computing unit deployed by the cloud terminal call center on the network edge side, and the feature descriptor is a multi-dimensional structured feature vector corresponding to the data scheduling request.

[0022] As an embodiment of the present invention, the extracting features from the data scheduling request to obtain a feature descriptor includes: Perform protocol format parsing processing on the data scheduling request to obtain a structured request protocol; Analyze the request protocol parameters in the structured request protocol and identify the request attribute identifiers in the request protocol parameters; Perform standardized conversion processing on the request attribute identifiers to obtain normalized feature values; Perform structured encapsulation processing on the normalized feature values to obtain a feature descriptor.

[0023] Among them, the structure request protocol is a set of structured parameters obtained after parsing the protocol format of the data scheduling request; the request protocol parameter is the specific field content in the structure request protocol (such as request type, data volume, etc.); the request attribute identifier is the service attribute and performance index identifier in the request protocol parameter (such as priority, real-time requirement, etc.); the normalized feature value is a unified format feature value obtained after standardizing and converting the request attribute identifier.

[0024] Furthermore, the protocol format of the data scheduling request can be parsed through a custom protocol parser to obtain the structure request protocol. The custom protocol parser includes a JSON parser; the request protocol parameters in the structure request protocol can be analyzed through field mapping rules. For example, reqType can be mapped to the request type; the request attribute identifier in the request protocol parameter can be identified through a pre-trained text classification model, such as the LSTM financial text classifier; the request attribute identifier can be standardized and converted through a feature standardization engine to obtain a normalized feature value. The feature standardization engine, such as a combination of a rule engine and a unit converter, or a machine learning standardization component, converts the request attribute identifier into a normalized feature value in a unified format through a preset mapping rule (such as priority 1-5 corresponding to text descriptions) and unit conversion (KB→B); the normalized feature value can be structurally encapsulated through a multi-dimensional vector encapsulation algorithm to obtain a feature descriptor. For example, the numerical feature [3, 512000, 1000] is encapsulated into a structured vector according to the (priority - data volume - delay threshold) dimension, or organized into a two-dimensional matrix [[3], [512000, 1000]] according to the service / performance dimension, so that the scattered feature values form a standardized feature vector that can be calculated by a machine, and thus a feature descriptor is obtained.

[0025] Based on the feature descriptor, the present invention can retrieve the request-associated data from the database of the cloud terminal call center, accurately locate the target data that matches the scheduling requirements, and avoid the waste of resources caused by full-scale query. Among them, the request-associated data is a set of service data directly matching the data scheduling request retrieved from the database of the cloud terminal call center based on the feature descriptor.

[0026] As an embodiment of the present invention, retrieving the request-associated data from the database of the cloud terminal call center based on the feature descriptor includes: Obtaining the data index identifier in the database and setting the retrieval rule of the feature descriptor in the database; Performing compliance verification processing on the retrieval rule to obtain a compliant retrieval rule; Set the retrieval query instruction corresponding to the compliance retrieval rule, and perform an association mapping process on the retrieval query instruction and the data index identifier to obtain an index-associated query condition; Perform an execution optimization process on the index-associated query condition to obtain an optimal query execution condition; Perform a distributed routing process on the optimal query execution condition to obtain parallel retrieval subtasks; Based on the parallel retrieval subtasks, retrieve the request-associated data from the database.

[0027] Among them, the data index identifier is the physical storage structure metadata in the database (such as partition key, index type); the retrieval rule is the query syntax mapping rule of the feature descriptor in the database; the compliance retrieval rule is the constraint expression that meets the data security and privacy policies after the retrieval rule passes the compliance verification process; the retrieval query instruction is the executable SQL statement template corresponding to the compliance retrieval rule; the index-associated query condition is the query expression with index optimization hints generated after the association mapping process of the retrieval query instruction and the data index identifier; the optimal query execution condition is the parameter set that selects the lowest-cost execution path after the index-associated query condition passes the execution optimization process; the parallel retrieval subtasks are the localized query task units disassembled according to the data sharding strategy after the optimal query execution condition passes the distributed routing process.

[0028] Further, the data index identifier in the database can be obtained through the database metadata interface, such as calling the pg_indexes system view of PostgreSQL to query the index structure; the retrieval rule of the feature descriptor in the database can be set through the rule mapping engine, such as mapping high_priority in the feature descriptor to priority>3 in SQL; the compliance check module can perform compliance check processing on the retrieval rule to obtain a compliant retrieval rule, such as verifying whether the query condition contains sensitive fields (such as ID card numbers) and automatically desensitizing them; the query instruction generator can set the retrieval query instruction corresponding to the compliant retrieval rule; the index mapping parser can perform an association mapping process on the retrieval query instruction and the data index identifier to obtain an index association query condition, such as associating the query field amount with the B-tree index idx_amount; the query optimizer can perform execution optimization processing on the index association query condition to obtain an optimal query execution condition, such as selecting a nested loop join instead of a hash join and estimating the cost based on statistical information; the distributed routing scheduler can perform distributed routing processing on the optimal query execution condition to obtain parallel retrieval subtasks, such as decomposing the time range query into a parallel task of 30 nodes according to date partitions; based on the parallel retrieval subtasks, the requested associated data can be retrieved from the database through multi-node parallel execution and result aggregator, such as after each node returns the data, removing duplicates and sorting by time to generate a complete result set.

[0029] S2. Perform semantic fusion processing on the feature descriptor and the requested associated data to obtain a multi-modal integrated semantics, calculate the fitness between the multi-modal integrated semantics and the edge computing nodes, and based on the fitness, screen out the data execution nodes corresponding to the requested associated data from the edge computing nodes.

[0030] Through semantic fusion processing of the feature descriptor and the requested associated data, the present invention obtains a multi-modal integrated semantics, which can achieve deep semantic alignment between the feature descriptor and the requested associated data, provide data support for the precise adaptation of edge computing nodes, and improve the intelligent level of data scheduling. Among them, the multi-modal integrated semantics refers to a unified semantic representation formed by fusing the semantics in the feature descriptor and the semantics in the requested associated data.

[0031] As an embodiment of the present invention, the performing semantic fusion processing on the feature descriptor and the requested associated data to obtain a multi-modal integrated semantics includes: Perform vector space mapping processing on the feature descriptor to obtain a service feature vector; Perform multi-dimensional feature extraction on the requested associated data to obtain multi-dimensional data features; Calculate the semantic association weight between the service feature vector and the multi-dimensional data feature; Based on the semantic association weight, perform feature alignment processing on the service feature vector and the multi-dimensional data feature to obtain heterogeneous alignment features; Perform semantic fusion processing on the heterogeneous alignment features to obtain multi-modal integrated semantics.

[0032] Among them, the service feature vector is a low-dimensional dense semantic vector converted after the feature descriptor is processed by vector space mapping; the multi-dimensional data feature is a multi-type feature set such as text, time series, and structure obtained after the request-associated data is subjected to multi-dimensional feature extraction; the semantic association weight represents the strength of the semantic correlation between the service feature vector and the multi-dimensional data feature; the heterogeneous alignment feature is a cross-modal semantic consistent feature representation obtained by performing feature alignment processing on the service feature vector and the multi-dimensional data feature based on the semantic association weight.

[0033] Furthermore, the service feature vector can be obtained by processing the feature descriptor through a pre-trained financial word vector model (such as FinBERT) for vector space mapping; the multi-dimensional data feature can be obtained by performing multi-dimensional feature extraction on the request-associated data through a combined feature extraction engine (TF-IDF + Fourier transform + statistical analysis); the semantic association weight between the service feature vector and the multi-dimensional data feature can be calculated through a multi-head attention mechanism module; based on the semantic association weight, the service feature vector and the multi-dimensional data feature are subjected to feature alignment processing through a cross-modal feature aligner to obtain heterogeneous alignment features; the heterogeneous alignment features can be subjected to semantic fusion processing through a gated semantic fusion network to obtain multi-modal integrated semantics.

[0034] By calculating the fitness between the multi-modal integrated semantics and the edge computing node, the present invention can accurately understand the matching ability of the edge computing node to business requirements, so as to facilitate subsequent intelligent scheduling of data processing tasks and computing resources, avoid processing delays caused by resource mismatches, and improve the efficiency and reliability of data execution in the edge computing scenario. Among them, the fitness represents a comprehensive measure of the semantic matching degree and resource carrying capacity between the multi-modal integrated semantics and the edge computing node, and reflects the fit degree of the semantic features of business requirements with the node computing resources, storage capacity, and network bandwidth.

[0035] As an embodiment of the present invention, calculating the fitness between the multi-modal integrated semantics and the edge computing node includes: Query the node multi-dimensional information corresponding to the edge computing node, and extract the key node information in the node multi-dimensional information; Filter out the node core features in the key node information, perform requirement analysis on the multi-modal integrated semantics, and obtain semantic service requirements; Calculate the similarity between the node core features and the semantic service requirements to obtain a similarity matrix; Based on the similarity matrix, calculate the fitness between the multi-modal integrated semantics and the edge computing node.

[0036] Among them, the node multi-dimensional information is various attributes corresponding to the edge computing node (such as computing power, storage capacity, network bandwidth, etc.), the key node information is the part that plays a decisive role in scheduling in the multi-dimensional information (such as real-time load, hardware performance), the node core feature is the quantifiable core index in the key node information (such as CPU usage rate, remaining storage), the semantic service requirement is the specific service requirement obtained after parsing the multi-modal integrated semantics (such as low latency, high security), and the similarity matrix is a quantization matrix formed by calculating the matching degree between the node core features and the semantic service requirements.

[0037] Furthermore, the node multi-dimensional information corresponding to the edge computing node can be queried through the node information database; the key node information in the multi-dimensional information can be extracted through the key index screening algorithm, such as attributes that play a core role in data scheduling decisions such as computing resource occupancy rate, remaining storage capacity, network bandwidth utilization rate, and hardware health status score; the node core features in the key node information can be screened out with the help of the feature quantization model, such as the PCA (Principal Component Analysis) dimensionality reduction model; the semantic service requirements can be obtained by using the semantic parsing engine to perform requirement analysis on the multi-modal integrated semantics; the similarity between the node core features and the semantic service requirements can be calculated through the similarity calculation algorithm to form a matrix; based on the similarity matrix, calculate the fitness between the multi-modal integrated semantics and the edge computing node, and sum the matrix elements by the weighted average method to obtain the fitness between the multi-modal integrated semantics and the edge computing node.

[0038] According to the present invention, based on the fitness, the data execution node corresponding to the requested associated data is screened out from the edge computing nodes, and an edge node that highly matches the semantic requirements of the requested associated data can be obtained, realizing the precise coupling of computing resources and service requirements. Among them, the data execution node is the edge computing node that is screened out from the edge computing nodes and is most suitable for executing the data processing task corresponding to the requested associated data. Furthermore, based on the value of the fitness, the data execution node corresponding to the requested associated data is screened out from the edge computing nodes, such as screening out the nodes higher than the threshold by setting a fitness threshold.

[0039] S3. Query the data scheduling device and the node server corresponding to the data execution node, calculate the channel noise ratio between the data scheduling device and the node server, and calculate the communication throughput rate between the data scheduling device and the node server based on the channel noise ratio.

[0040] By calculating the channel noise ratio between the data scheduling device and the node server, the present invention can understand the signal interference degree of the link during the data transmission process between the data scheduling device and the node server, providing data support for the subsequent calculation of the communication throughput rate. Among them, the data scheduling device is the data transfer management device corresponding to the data execution node, responsible for data distribution, scheduling, and transmission control. The node server is the physical or logical computing unit corresponding to the data execution node, carrying the core functions of data processing and service execution. The channel noise ratio is the ratio of the signal power to the noise power between the data scheduling device and the node server, used to measure the anti-interference ability and communication quality of the data transmission link. Further, the data scheduling device and the node server corresponding to the data execution node can be queried through the node resource mapping table.

[0041] As an embodiment of the present invention, the calculation of the channel noise ratio between the data scheduling device and the node server includes: Calculate the channel noise ratio between the data scheduling device and the node server through the following formula: ; where A represents the channel noise ratio between the data scheduling device and the node server, represents the transmission power between the i-th data scheduling device and the e-th node server, represents the channel coefficient between the i-th data scheduling device and the e-th node server, represents the maximum transmission power, represents the transmission power of other links except the link between the i-th data scheduling device and the e-th node server. E represents the noise power, and i and e respectively represent the serial numbers corresponding to the data scheduling device and the node server.

[0042] Among them, the transmission power represents the magnitude of the initial energy during signal transmission between the data scheduling device and the node server, and determines the signal propagation distance and intensity; the channel coefficient represents the intensity change factor of the signal affected by the channel during transmission from the data scheduling device to the node server; the noise power represents the magnitude of the interference signal introduced by the environment, equipment, etc. in the transmission link between the data scheduling device and the node server. Further, the transmission power can be obtained by reading device configuration parameters, measuring with a power measurement instrument, or according to a preset value of a communication protocol; the channel coefficient can be calculated by channel estimation techniques (such as the estimation method based on pilot symbols), statistical analysis of historical data, or determined by a dedicated channel measurement device; the noise power can be measured by a spectrum analyzer, calculated by a noise model (combining environmental parameters and device characteristics), or obtained by the device's own measurement when there is no signal input.

[0043] The present invention calculates the communication throughput between the data scheduling device and the node server based on the channel noise ratio, so as to understand the data transmission ability between the data scheduling device and the node server, and further facilitate the construction and processing of the data transmission path between the data scheduling device and the node server. Among them, the communication throughput represents the quantitative expression of the data transmission ability between the data scheduling device and the node server.

[0044] As an embodiment of the present invention, calculating the communication throughput between the data scheduling device and the node server based on the channel noise ratio includes: Calculating the communication throughput between the data scheduling device and the node server through the following formula: ; Among them, represents the communication throughput between the data scheduling device and the node server, F represents the total bandwidth between the data scheduling device and the node server, G represents the number of associated devices corresponding to the node server, and A represents the channel noise ratio.

[0045] Among them, the total bandwidth represents the total amount of frequency resources available for data transmission between the data scheduling device and the node server, reflecting the potential data transmission ability of the link; the number of associated devices is the number of terminals such as mobile devices that are connected to the node server and need to share communication resources for data interaction. Further, the total bandwidth can be obtained through network configuration parameters or measured with the help of professional network testing tools (such as network analyzers); the number of associated devices can be obtained by counting the number of devices currently connected to the edge server through its connection management module.

[0046] S4. Calculate the total task data payload of the request-associated data during transmission, and construct a data transmission path between the data scheduling device and the node server by combining the total task data payload and the communication throughput rate.

[0047] By calculating the total task data payload of the request-associated data during transmission, the present invention can understand the total amount of data of the request-associated data in data transmission, providing data support for constructing a data transmission path between the data scheduling device and the node server. Herein, the total task data payload represents the total amount of data of the request-associated data during transmission. As an embodiment of the present invention, the calculation of the total task data payload of the request-associated data during transmission includes: Perform structured parsing on the request-associated data to obtain a data field list; Perform sensitive identification on the data field list to obtain sensitive data fields; Based on the sensitive data fields, determine the data to be encrypted in the request-associated data; Based on the data to be encrypted, perform encryption processing on the request-associated data to obtain encrypted associated data; Calculate the task data volume corresponding to the encrypted associated data, and analyze the data key length corresponding to the encrypted associated data; By combining the task data volume and the data key length, calculate the total task data payload of the request-associated data during transmission through the following formula: ; wherein, H represents the total task data payload of the request-associated data during transmission, represents the task data volume corresponding to the hth encrypted associated data, represents the data equal share amount corresponding to the hth encrypted associated data, represents the data key length corresponding to the hth encrypted associated data.

[0048] Among them, the data field list is a list containing all field names, data types, and original values obtained after the structured parsing of the request-associated data; the sensitive data fields are the fields marked as involving privacy or confidential information after the sensitive identification of the data field list; the data to be encrypted is the data part that needs to be encrypted based on the sensitive data fields in the request-associated data; the task data volume is the data volume size corresponding to the encrypted associated data (i.e., the total payload of the encrypted data); the data key length represents the number of bits of the key used in the encryption algorithm corresponding to the encrypted associated data (reflecting the encryption strength); the data equal division quantity represents the equal quantity or capacity of each portion of the encrypted associated data after being divided according to specific rules (such as encryption block size, transmission fragmentation requirements), and can be obtained by a method of calculating the number of division portions and the data volume of each portion through the network transmission MTU limit (such as 1500 bytes for Ethernet).

[0049] Furthermore, the request-associated data can be structurally parsed through protocol parsing tools (such as Wireshark, Postman) to obtain a data field list; the data field list can be sensitively identified through regular expression matching to obtain sensitive data fields; based on the sensitive data fields, combined with business logic and compliance requirements, the data to be encrypted in the request-associated data can be determined. For example, by judging whether the field involves personal privacy (such as ID number), financial information (such as bank card number), or corporate secrets (such as an unpublished contract), and comparing with the requirements of regulations such as the Personal Information Protection Law and the Cybersecurity Law, the scope of fields that must be encrypted can be delineated; based on the data to be encrypted, the request-associated data can be encrypted through an encryption algorithm (such as AES, RSA) to obtain encrypted associated data; the task data volume corresponding to the encrypted associated data can be obtained by counting the total number of binary bytes of the encrypted associated data; the data key length corresponding to the encrypted associated data can be analyzed through the corresponding key file.

[0050] The present invention combines the total task data payload and the communication throughput rate to construct a data transmission path between the data scheduling device and the node server, and an optimal transmission path that takes into account both data transmission efficiency and link load balancing can be obtained, thereby reducing data transmission latency and improving the overall system throughput. Among them, the data transmission path is a combination of the data transmission channel and routing rules between the data scheduling device and the node server. Furthermore, by combining the total task data payload and the communication throughput rate, a data transmission path between the data scheduling device and the node server is constructed. First, the required transmission bandwidth is determined based on the total task data payload, then the available capacity of each link is calculated in combination with the communication throughput rate, and finally the optimal path combination is selected through a load balancing algorithm.

[0051] S5. Set the scheduling priority corresponding to the request-associated data, and in combination with the scheduling priority and the data transmission path, use the data scheduling device and the node server to perform scheduling processing on the request-associated data to obtain a scheduling result.

[0052] By setting the scheduling priority corresponding to the request-associated data, the present invention can ensure that high-timeliness and high-urgency data are scheduled first, effectively reduce the latency of critical services, and improve the overall response efficiency of the system, thereby achieving the purpose of improving data transmission efficiency. Among them, the scheduling priority is the order of data processing determined by the request-associated data based on factors such as time-sensitive attributes and business urgency.

[0053] As an embodiment of the present invention, setting the scheduling priority corresponding to the request-associated data includes: Perform metadata parsing on the request-associated data to obtain a set of basic attributes; Based on a preset time-sensitive rule library, perform screening processing on the set of basic attributes to obtain target sensitive attributes; Extract the time constraint words in the target sensitive attributes and analyze the semantic meaning of the constraint words corresponding to the time constraint words; Based on the semantic meaning of the constraint words, determine the emergency quantification index corresponding to the request-associated data; Extract the data timestamp corresponding to the request-associated data, and based on the data timestamp, calculate the data timeliness index corresponding to the request-associated data; In combination with the emergency quantification index and the data timeliness index, set the scheduling priority corresponding to the request-associated data.

[0054] Among them, the basic attribute set is a combination of basic features extracted after parsing the request-related data (such as original attributes like timestamp, business type, user level, etc.); the preset time-sensitive rule library is a set of predefined time-sensitive judgment criteria (including rules such as business scenario time limit requirements, SLA agreements, etc.); the target sensitive attribute is a subset of attributes that meet the time-sensitive conditions obtained by screening the basic attribute set according to the preset time-sensitive rule library (such as screening out key attributes with time constraints); the time constraint word is a keyword representing time limit in the target sensitive attribute (such as time-sensitive description words like "immediately", "within 24 hours", "same day", etc.); the semantic meaning of the constraint word is a specific meaning explanation of the time constraint word corresponding to the time constraint word (such as "immediately" corresponding to "≤10 seconds for processing"); the emergency quantification index represents the urgency of data processing corresponding to the request-related data (a priority identifier determined by factors such as business impact and time deadline requirements); the data timestamp is a time mark when the record data corresponding to the request-related data is generated or updated (including start time and end time); the data timeliness index is a quantitative index for measuring data timeliness calculated based on the data timestamp for the request-related data (such as calculating data freshness or expiration degree through time difference).

[0055] Further, metadata parsing can be performed on the request-associated data through regular expressions to obtain a set of basic attributes; based on a preset time-sensitive rule library, the set of basic attributes can be screened through a rule matching algorithm to obtain target sensitive attributes; time constraint words in the target sensitive attributes can be extracted through natural language processing (NLP) techniques; the constraint word semantics corresponding to the time constraint words can be analyzed through a semantic analysis engine; based on the constraint word semantics, an emergency quantification index corresponding to the request-associated data can be determined. For example, semantics such as "take effect immediately" and "be completed on the same day" are converted into specific scores. For example, immediately corresponds to 100 points and the same day corresponds to 80 points. If the constraint word semantics is the same day, the corresponding emergency quantification index is 0.8; a data timestamp corresponding to the request-associated data can be extracted through a data extraction tool; based on the data timestamp, a data timeliness index corresponding to the request-associated data can be calculated. The calculation steps are as follows: subtract the start time from the end time, and map the obtained time difference to a timeliness score according to business scenario rules (for example, if the difference ≤ 10 minutes, the score is 100 points; if it is 10 - 60 minutes, the score is 80 points), and finally obtain a data timeliness index representing data freshness; combining the emergency quantification index and the data timeliness index, set the scheduling priority corresponding to the request-associated data. Combining the emergency quantification index and the data timeliness index, the two indicators are weighted and calculated to obtain a comprehensive score according to a preset weight (for example, the emergency index accounts for 60% and the timeliness index accounts for 40%), and then the score interval (for example, 90 - 100 points is the highest priority) is mapped to a scheduling priority level, and finally an identification of the order of data processing is generated.

[0056] The present invention combines the scheduling priority and the data transmission path, and uses the data scheduling device and the node server to perform scheduling processing on the request-associated data to obtain a scheduling result, thereby realizing the transmission of high-priority data along the shortest path or a low-latency link, reducing the time-consuming of critical business processing and optimizing the utilization efficiency of system resources. Specifically, to more intuitively understand the process of using the data scheduling device and the node server to perform scheduling processing on the request-associated data in this application, reference can be made to Figure 2 As shown, it is a schematic diagram of scheduling processing in the scheduling method for cloud terminal call center data based on edge computing provided by the present invention. It should be noted that in the present invention, Figure 2 The presented schematic diagram of scheduling processing is only used for the scheduling processing of the scheduling method for cloud terminal call center data based on edge computing, and does not limit the scheduling processing of the scheduling method for cloud terminal call center data based on edge computing in actual different application scenarios.

[0057] Compared with the problems described in the background art, the present invention can obtain a feature descriptor of the core attributes of the scheduling requirements by extracting features from the data scheduling request, thereby providing a scientific basis for efficiently retrieving request-associated data. Further, the present invention performs semantic fusion processing on the feature descriptor and the request-associated data to obtain a multi-modal integrated semantics, which can achieve deep semantic alignment between the feature descriptor and the request-associated data, provide data support for the precise adaptation of edge computing nodes, and improve the intelligent level of data scheduling. The present invention calculates the channel noise ratio between the data scheduling device and the node server, and can understand the signal interference degree of the link during the data transmission between the data scheduling device and the node server, providing data support for the subsequent calculation of the communication throughput rate. Further, the present invention calculates the total task data payload of the request-associated data during transmission, and can understand the total amount of data of the request-associated data in data transmission, providing data support for constructing the data transmission path between the data scheduling device and the node server. The present invention sets the scheduling priority corresponding to the request-associated data, which can ensure the priority scheduling of high-timeliness and high-urgency data, effectively reduce the delay of critical services and improve the overall response efficiency of the system, thereby achieving the purpose of improving data transmission efficiency. Therefore, the scheduling method and system for cloud terminal call center data based on edge computing provided by the embodiments of the present invention can improve the scheduling efficiency of cloud terminal call center data.

[0058] Embodiment 2: As Figure 3 shown, it is a functional module diagram of a scheduling system for cloud terminal call center data based on edge computing according to the present invention.

[0059] The scheduling system 300 for cloud terminal call center data based on edge computing according to the present invention can be installed in an electronic device. According to the implemented functions, the scheduling system for cloud terminal call center data based on edge computing may include a data retrieval module 301, an edge node screening module 302, a communication throughput rate calculation module 303, a data transmission path construction module 304, and a data scheduling processing module 305. The modules of the present invention may also be referred to as units, which refer to a series of computer program segments that can be executed by a processor of an electronic device and can complete fixed functions, and are stored in the memory of the electronic device.

[0060] In the embodiments of the present invention, the functions of each module / unit are as follows: The data retrieval module 301 is configured to obtain a data scheduling request and an edge computing node of a cloud terminal call center, extract features from the data scheduling request to obtain a feature descriptor, and retrieve request-associated data from the database of the cloud terminal call center based on the feature descriptor; The edge node screening module 302 is configured to perform semantic fusion processing on the feature descriptor and the request-associated data to obtain multimodal integrated semantics, calculate the fitness between the multimodal integrated semantics and the edge computing nodes, and based on the fitness, screen out the data execution nodes corresponding to the request-associated data from the edge computing nodes; The communication throughput rate calculation module 303 is configured to query the data scheduling device and the node server corresponding to the data execution node, calculate the channel noise ratio between the data scheduling device and the node server, and based on the channel noise ratio, calculate the communication throughput rate between the data scheduling device and the node server; The data transmission path construction module 304 is configured to calculate the total task data payload during the transmission of the request-associated data, and combine the task data payload and the communication throughput rate to construct a data transmission path between the data scheduling device and the node server; The data scheduling processing module 305 is configured to set the scheduling priority corresponding to the request-associated data, combine the scheduling priority and the data transmission path, and use the data scheduling device and the node server to perform scheduling processing on the request-associated data to obtain a scheduling result.

[0061] Specifically, each module in the scheduling system 300 for cloud terminal call center data based on edge computing in the embodiments of the present invention adopts the same technical means as those in the above-mentioned Figure 1 scheduling method for cloud terminal call center data based on edge computing, and can produce the same technical effects, which will not be elaborated here.

[0062] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and without departing from the spirit or basic characteristics of the present invention, the present invention can be implemented in other specific forms.

[0063] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A scheduling method for cloud terminal call center data based on edge computing, characterized in that, The method includes: Obtain the data scheduling request and the edge computing node of the cloud terminal call center, extract features from the data scheduling request to obtain a feature descriptor, and retrieve the request-associated data from the database of the cloud terminal call center based on the feature descriptor; Perform semantic fusion processing on the feature descriptor and the request-associated data to obtain a multi-modal integrated semantics, calculate the fitness between the multi-modal integrated semantics and the edge computing node, and based on the fitness, screen out the data execution node corresponding to the request-associated data from the edge computing nodes; Query the data scheduling device and the node server corresponding to the data execution node, calculate the channel noise ratio between the data scheduling device and the node server, and based on the channel noise ratio, calculate the communication throughput rate between the data scheduling device and the node server; Calculate the total task data payload during the transmission of the request-associated data, and combine the task data payload and the communication throughput rate to construct a data transmission path between the data scheduling device and the node server; Set the scheduling priority corresponding to the request-associated data, combine the scheduling priority and the data transmission path, and use the data scheduling device and the node server to perform scheduling processing on the request-associated data to obtain a scheduling result.

2. The scheduling method for cloud terminal call center data implemented based on edge computing according to claim 1, wherein, The extracting features from the data scheduling request to obtain a feature descriptor includes: Perform protocol format parsing processing on the data scheduling request to obtain a structured request protocol; Analyze the request protocol parameters in the structured request protocol and identify the request attribute identifiers in the request protocol parameters; Perform standardized conversion processing on the request attribute identifiers to obtain a normalized feature value; Perform structured encapsulation processing on the normalized feature value to obtain a feature descriptor.

3. The scheduling method for cloud terminal call center data implemented based on edge computing according to claim 1, wherein, The retrieving the request-associated data from the database of the cloud terminal call center based on the feature descriptor includes: Obtain the data index identifier in the database and set the retrieval rule of the feature descriptor in the database; Perform compliance verification processing on the retrieval rule to obtain a compliant retrieval rule; Set the retrieval query instruction corresponding to the compliant retrieval rule, and perform associated mapping processing on the retrieval query instruction and the data index identifier to obtain an index-associated query condition; Perform execution optimization processing on the index-associated query condition to obtain an optimal query execution condition; Perform distributed routing processing on the optimal query execution condition to obtain parallel retrieval subtasks; Retrieve the request-associated data from the database based on the parallel retrieval subtasks.

4. The scheduling method for cloud terminal call center data implemented based on edge computing according to claim 1, wherein The performing semantic fusion processing on the feature descriptor and the request-associated data to obtain a multi-modal integrated semantics includes: Perform vector space mapping processing on the feature descriptor to obtain a service feature vector; Perform multi-dimensional feature extraction on the request-associated data to obtain multi-dimensional data features; Calculate the semantic association weight between the service feature vector and the multi-dimensional data features; Based on the semantic association weight, perform feature alignment processing on the service feature vector and the multi-dimensional data features to obtain heterogeneous alignment features; Perform semantic fusion processing on the heterogeneous alignment features to obtain multimodal integrated semantics.

5. The scheduling method for cloud terminal call center data implemented based on edge computing according to claim 1, wherein The calculation of the fitness between the multimodal integrated semantics and the edge computing node includes: Query the node multi-dimensional information corresponding to the edge computing node, and extract the key node information from the node multi-dimensional information; Filter out the node core features in the key node information, perform requirement parsing processing on the multimodal integrated semantics to obtain semantic service requirements; Calculate the similarity between the node core features and the semantic service requirements to obtain a similarity matrix; Based on the similarity matrix, calculate the fitness between the multimodal integrated semantics and the edge computing node.

6. The scheduling method for cloud terminal call center data implemented based on edge computing according to claim 1, wherein, The calculation of the channel noise ratio between the data scheduling device and the node server includes: Calculate the channel noise ratio between the data scheduling device and the node server through the following formula: ； Among them, A represents the channel noise ratio between the data scheduling device and the node server, represents the transmission power between the i-th data scheduling device and the e-th node server, represents the channel coefficient between the i-th data scheduling device and the e-th node server, represents the maximum transmission power, represents the transmission power of other links except the link between the i-th data scheduling device and the e-th node server. E represents the noise power. i and e respectively represent the serial numbers corresponding to the data scheduling device and the node server.

7. The scheduling method for cloud terminal call center data implemented based on edge computing according to claim 1, characterized in that, The calculation of the communication throughput rate between the data scheduling device and the node server based on the channel noise ratio includes: Calculate the communication throughput rate between the data scheduling device and the node server through the following formula: ； Among them, represents the communication throughput rate between the data scheduling device and the node server, F represents the total bandwidth between the data scheduling device and the node server, G represents the number of associated devices corresponding to the node server, and A represents the channel noise ratio.

8. The scheduling method for cloud terminal call center data implemented based on edge computing according to claim 1, characterized in that The calculation of the total task data payload during the transmission of the request-associated data includes: Perform structured parsing on the request-associated data to obtain a data field list; Perform sensitive identification on the data field list to obtain sensitive data fields; Based on the sensitive data fields, determine the data to be encrypted in the request-associated data; Based on the data to be encrypted, perform encryption processing on the request-associated data to obtain encrypted associated data; Calculate the task data volume corresponding to the encrypted associated data, and analyze the data key length corresponding to the encrypted associated data; Combine the task data volume and the data key length, and calculate the total task data payload during the transmission of the request-associated data through the following formula: ； Among them, H represents the total task data payload during the transmission of the request-associated data, represents the task data volume corresponding to the h-th encrypted associated data, represents the data equal-division quantity corresponding to the h-th encrypted associated data, represents the data secret key length corresponding to the h-th encrypted associated data.

9. The scheduling method for cloud terminal call center data implemented based on edge computing according to claim 1, characterized in that, The setting of the scheduling priority corresponding to the request-associated data includes: Perform metadata parsing on the request-associated data to obtain a basic attribute set; Based on a preset time-sensitive rule library, perform screening processing on the basic attribute set to obtain target sensitive attributes; Extract the time constraint words in the target sensitive attributes, and analyze the semantic meaning of the constraint words corresponding to the time constraint words; Based on the semantic meaning of the constraint words, determine the emergency quantization index corresponding to the request-associated data; Extract the data timestamp corresponding to the request-associated data, and calculate the data timeliness index corresponding to the request-associated data based on the data timestamp; Combine the emergency quantization index and the data expiration index to set the scheduling priority corresponding to the request-associated data.

10. A scheduling system for cloud terminal call center data implemented based on edge computing, characterized in that, The system includes: A data retrieval module, configured to obtain a data scheduling request and an edge computing node of a cloud terminal call center, perform feature extraction on the data scheduling request to obtain a feature descriptor, and retrieve request-associated data from the database of the cloud terminal call center based on the feature descriptor; An edge node screening module, which is used to perform semantic fusion processing on the feature descriptor and the request-associated data to obtain a multimodal integrated semantics, calculate the fitness between the multimodal integrated semantics and the edge computing node, and based on the fitness, screen out the data execution nodes corresponding to the request-associated data from the edge computing nodes; A communication throughput rate calculation module, which is used to query the data scheduling device and the node server corresponding to the data execution node, calculate the channel noise ratio between the data scheduling device and the node server, and based on the channel noise ratio, calculate the communication throughput rate between the data scheduling device and the node server; A data transmission path construction module, which is used to calculate the total task data payload during the transmission of the request-associated data, and combine the task data payload and the communication throughput rate to construct a data transmission path between the data scheduling device and the node server; A data scheduling processing module, which is used to set the scheduling priority corresponding to the request-associated data, combine the scheduling priority and the data transmission path, and use the data scheduling device and the node server to perform scheduling processing on the request-associated data to obtain a scheduling result.

Images