CN116010370A

CN116010370A - Digital business information processing method and server combined with edge calculation

Info

Publication number: CN116010370A
Application number: CN202310313859.7A
Authority: CN
Inventors: 孙家祥; 郭乐乐; 李代艳
Original assignee: Taicang City Lvdian Information Technology Co ltd
Current assignee: Taicang City Lvdian Information Technology Co ltd
Priority date: 2023-03-28
Filing date: 2023-03-28
Publication date: 2023-04-25
Anticipated expiration: 2043-03-28
Also published as: CN116010370B

Abstract

According to the digital business information processing method and the edge computing server combined with edge computing, through the contrast learning mapping unit for extracting upper macro characterization knowledge and the behavior tendency mapping unit for extracting behavior tendency characterization knowledge, a network capable of learning two types of knowledge is obtained, when the behavior tendency analysis is carried out on the digital business information log according to the digital business information processing network, the information extraction can be carried out on the digital business information by combining the upper macro characterization knowledge and the behavior tendency characterization knowledge, the collaborative learning capability is enhanced, the extraction speed of the characterization knowledge is improved, the behavior tendency analysis of the digital business information is carried out according to the obtained digital business information log interaction characterization knowledge, and the obtained multi-element classification tendency result is more accurate and efficient.

Description

Digital business information processing method and server combined with edge calculation

Technical Field

The present application relates to the field of data processing, and in particular, to the field of artificial intelligence, and in particular, to a method and a server for processing digital service information in combination with edge computing.

Background

With the great development of cloud services, edge computing is accompanied with the spring wind of AI, and plays an important role in the field of Internet of things. Edge computing is the provision of cloud services and IT environment services for application developers and service providers on the edge side of the network. For example, in automobile auxiliary driving and intelligent production monitoring control, the problems of high delay, unstable network, low bandwidth and the like in a cloud computing mode can be solved by performing nearby computation through equipment data collected at the edge side and efficiently responding. The large amount of digital business information generated on the edge side generally conceals the meaning of the behavior tendency of the user, for example, in automobile driving, and the driving behavior tendency of the user can be analyzed through the analysis of the driving behavior of the user so as to carry out targeted auxiliary driving. How to accurately analyze and identify the behavior tendency of the user from the digital service information is a technical problem to be solved, and there is room for improvement at present. It should be noted that the foregoing is merely for the purpose of facilitating understanding of the technology of the present application, and is not an admission that it is prior art basis.

Disclosure of Invention

The application provides a digital service information processing method and a server combined with edge calculation.

According to an aspect of the present application, there is provided a digital service information processing method in combination with edge computation, applied to an edge computation server, the method including:

acquiring target digital service data through data acquisition equipment at an edge end to generate a digital service information log;

extracting upper macro characterization knowledge G-K1 of the digital service information log according to a contrast learning mapping unit in a pre-optimized digital service information processing network; the upper layer macro characterization knowledge G-K1 is used for reflecting macro dimension data meanings of the digital service information log;

extracting behavior tendency characterization knowledge I-K1 of the digital service information log according to a behavior tendency mapping unit in the digital service information processing network; the behavior tendency characterization knowledge I-K1 is used for reflecting the data behavior tendency meaning of the digital business information log;

performing characterization knowledge interaction on the upper macro characterization knowledge G-K1 and the behavior tendency characterization knowledge I-K1 according to a behavior tendency analysis unit in the digital service information processing network, and performing multi-element classification on the digital service information log according to the obtained first interaction characterization knowledge to obtain a multi-element classification tendency result of the digital service information log; the digital business information processing network is obtained based on the combination optimization of the comparison learning mapping unit and the behavior tendency mapping unit.

As an implementation manner, the multi-element classification of the digital service information log according to the obtained first interaction characterization knowledge further includes:

respectively extracting upper macro characterization knowledge G-K2 of each alternative digital service information log in a log library according to the contrast learning mapping unit;

according to the behavior tendency mapping unit, behavior tendency characterization knowledge I-K2 of the digital business information log is respectively extracted;

according to the behavior trend analysis unit, performing characterization knowledge interaction on each upper-layer macro characterization knowledge G-K2 and the corresponding behavior trend characterization knowledge I-K2 to obtain second interaction characterization knowledge respectively corresponding to each alternative digital service information log;

determining similarity scores of each alternative digital service information log and the digital service information log according to the first interaction characterization knowledge and the second interaction characterization knowledge corresponding to each alternative digital service information log respectively;

and determining a pairing digital service information log corresponding to the digital service information log according to each similarity score.

As an embodiment, the digital service information processing network further includes a data filtering unit; before extracting the upper macro characterization knowledge G-K1 of the digital service information log according to the comparison learning mapping unit in the pre-optimized digital service information processing network, the method further comprises the following steps:

Loading the digital service information log into a data filtering unit in the digital service information processing network, and extracting macroscopic representation knowledge from the digital service information log according to the data filtering unit to obtain a macroscopic representation knowledge relation network corresponding to the digital service information log;

the extracting the upper macro characterization knowledge G-K1 of the digital service information log according to the comparison learning mapping unit in the pre-optimized digital service information processing network comprises the following steps: performing mapping dimension reduction processing on the macroscopic representation knowledge relation network according to the contrast learning mapping unit to obtain upper macroscopic representation knowledge G-K1 corresponding to the digital service information log;

wherein the behavior tendency mapping unit comprises a feature extraction subunit and a feature coding subunit;

the step of extracting the behavior trend characterization knowledge I-K1 of the digital service information log according to the behavior trend mapping unit in the digital service information processing network comprises the following steps: extracting characterization knowledge from the behavior tendency information in the macroscopic characterization knowledge relationship network according to a feature extraction subunit in the behavior tendency mapping unit, and carrying out mapping dimension reduction processing on the extracted behavior tendency information according to a feature coding subunit in the behavior tendency mapping unit to obtain behavior tendency characterization knowledge I-K1 corresponding to the digital service information log;

The convergence efficiency parameter of the behavior trend analysis unit is larger than that of other units, and the other units comprise the data filtering unit, the contrast learning mapping unit and the behavior trend mapping unit.

As an embodiment, the data filtering unit comprises a plurality of feature extraction subunits; the parameters corresponding to the feature extraction subunit in the data filtering unit are obtained by carrying out preliminary assignment on weights and offsets according to the parameters which are calibrated by the pre-deployed example log library, and the parameters corresponding to the feature extraction subunit in the behavior tendency mapping unit are obtained by carrying out preliminary assignment on random weights and offsets.

As an embodiment, the method further includes an optimization process of the digital service information processing network, including:

acquiring an example log set, and screening example log groups in the example log set;

loading the filtered example log packet into the calibrated digital service information processing network, and acquiring an estimated knowledge field which represents an estimated credibility coefficient corresponding to each classification trend result of the example log according to upper-layer macro characterization knowledge G-K3 output by a comparison learning mapping unit, behavior trend characterization knowledge I-K3 output by the behavior trend mapping unit and estimated knowledge field output by the behavior trend analysis unit in the digital service information processing network;

Generating a target network quality evaluation factor according to the upper macro characterization knowledge G-K3, the behavior tendency characterization knowledge I-K3 and the estimated knowledge field, and repeatedly adjusting parameters of the digital service information processing network according to the target network quality evaluation factor until the digital service information processing network meets the preset adjustment requirement, so as to obtain the adjusted digital service information processing network.

As one embodiment, the generating a network quality assessment factor according to the upper macro characterization knowledge G-K3, the behavioral tendency characterization knowledge I-K3 and the estimated knowledge field includes:

generating a first multi-element example log quality assessment factor according to upper-layer macro characterization knowledge G-K3 corresponding to each example log in the example log group;

according to the behavior tendency characterization knowledge I-K3 corresponding to each example log, generating a second multi-element example log quality evaluation factor;

generating a multi-element classification quality evaluation factor according to the estimated knowledge field and the multi-element classification knowledge field corresponding to each example log, wherein the multi-element classification knowledge field represents the real credibility coefficient corresponding to each classification trend result of the example log;

Obtaining estimated entropy quality assessment factors according to the estimated knowledge fields corresponding to each example log;

and integrating the first multi-element example log quality assessment factor, the second multi-element example log quality assessment factor, the multi-element classification quality assessment factor and the estimated entropy quality assessment factor to obtain a target network quality assessment factor.

As one embodiment, each of the example log groupings encompasses a comparative example log, a true example log, and a false example log;

the generating a first multi-element example log quality assessment factor according to the upper macro characterization knowledge G-K3 corresponding to each example log in the example log group comprises the following steps:

determining the spatial similarity D1 between the upper-layer macro characterization knowledge G-K3 corresponding to the comparison example log and the upper-layer macro characterization knowledge G-K3 corresponding to the true example log in the example log group, and determining the spatial similarity D2 between the upper-layer macro characterization knowledge G-K3 corresponding to the comparison example log and the upper-layer macro characterization knowledge G-K3 corresponding to the false example log;

adding the difference result of the spatial similarity D1 and the spatial similarity D2 and a set critical variable, wherein the added result is used as a target variable V1, and the set critical variable represents a difference result limit of similarity scores between a true value example log and a false value example log;

And determining the maximum variable of the target variable V1 and the alternative variable as the first multi-element example log quality assessment factor.

As one embodiment, each of the example log groupings encompasses a comparative example log, a true example log, and a false example log; the step of generating a second multi-element example log quality evaluation factor according to the behavior tendency characterization knowledge I-K3 corresponding to each example log comprises the following steps:

according to the spatial similarity D3 between the behavior trend characterization knowledge I-K3 corresponding to the comparison example log and the behavior trend characterization knowledge I-K3 corresponding to the true value example log in the example log group, the spatial similarity D4 between the behavior trend characterization knowledge I-K3 corresponding to the comparison example log and the behavior trend characterization knowledge I-K3 corresponding to the false value example log;

adding the difference result of the spatial similarity D3 and the spatial similarity D4 and a set critical variable, wherein the added result is used as a target variable V2, and the set critical variable represents a difference result limit of similarity scores between a true value example log and a false value example log;

and determining the maximum variable of the target variable V2 and the alternative variable as the second multi-element example log quality assessment factor.

As an embodiment, before the repeated adjustment of the parameters of the digital service information processing network according to the target network quality assessment factor, the method further includes:

integrating the first multi-element example log quality assessment factor, the multi-element classification quality assessment factor and the estimated entropy quality assessment factor to obtain an intermediate network quality assessment factor;

repeatedly adjusting parameters of a behavior tendency mapping unit in the digital service information processing network according to the set adjustment times according to the intermediate network quality evaluation factors;

the method further includes a process of generating an example log packet in the example log set, including:

establishing each two digital business information example logs with similarity scores larger than the set similarity scores as true value example log pairs through the obtained similarity scores among the plurality of digital business information example logs;

performing example log mixing according to each true value example log pair to obtain a plurality of example log groups for generating the example log set, wherein each example log group comprises a comparison example log, a true value example log and a false value example log, the comparison example log and the true value example log are digital business information example logs with similarity scores greater than a set similarity score, and the false value example log and the true value example log are digital business information example logs with similarity scores not greater than the set similarity score;

The example log mixing based on the true value example log pair, to obtain a plurality of example log groups for generating the example log set, includes:

determining one digital service information example log in a true value example log pair, determining the true value example log as a target example log, and respectively determining one digital service information example log in other example log pairs as an intermediate example log;

determining spatial similarity between the target example log and each of the intermediate example logs;

sequentially arranging the plurality of intermediate example logs according to the corresponding spatial similarity, screening one or more intermediate example logs with preset sequences, and determining the intermediate example logs as false value example logs corresponding to the target example logs;

and respectively constructing each screened false value example log and the true value example log pair into example log groups, wherein a target example log in the true value example log pair is a true value example log in the example log groups, and the rest example logs in the true value example log pair are comparison example logs in the example log groups.

According to another aspect of the present application, there is provided an edge computing server including:

At least one processor;

and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method described above.

The application at least comprises the following beneficial effects:

It should be understood that the description of this section is not intended to identify key or critical features of the embodiments of the application or to delineate the scope of the application. Other features of the present application will become apparent from the description that follows.

Drawings

The accompanying drawings illustrate exemplary embodiments and, together with the description, serve to explain exemplary implementations of the embodiments. The illustrated embodiments are for exemplary purposes only and do not limit the scope of the claims. Throughout the drawings, identical reference numerals designate similar, but not necessarily identical, elements.

Fig. 1 shows an application scenario schematic diagram of a digital service information processing method in combination with edge computation according to an embodiment of the present application.

Fig. 2 shows a flow chart of a digital service information processing method in connection with edge computation according to an embodiment of the present application.

Fig. 3 shows a functional block architecture diagram of a digital service information processing apparatus according to an embodiment of the present application.

Fig. 4 shows a schematic composition diagram of an edge computing server according to an embodiment of the present application.

Description of the embodiments

Exemplary embodiments of the present application are described below in conjunction with the accompanying drawings, which include various details of the embodiments of the present application to facilitate understanding, and should be considered as merely exemplary. Accordingly, one of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope of the present application. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.

In this application, unless otherwise indicated, the use of the terms "first," "second," etc. to describe various elements is not intended to limit the positional relationship, timing relationship, or importance of the elements, but is merely used to distinguish one element from another. In some examples, a first element and a second element may refer to the same instance of the element, and in some cases, they may also refer to different instances based on the description of the context.

The terminology used in the description of the various illustrated examples in this application is for the purpose of describing particular examples only and is not intended to be limiting. Unless the context clearly indicates otherwise, the elements may be one or more if the number of the elements is not specifically limited. Furthermore, the term "and/or" as used in this application encompasses any and all possible combinations of the listed items.

Fig. 1 shows an application scenario schematic diagram of a digital service information processing method combined with edge computation according to an embodiment of the present application. Including one or more data acquisition devices 101, an edge computing server 120, and one or more communication networks 200 coupling the one or more data acquisition devices 101 to the edge computing server 120.

In embodiments of the present application, the edge computation server 120 may run one or more services or software applications that enable execution of digital business information processing methods in conjunction with edge computation.

In the configuration shown in fig. 1, edge computing server 120 may include one or more components that implement the functions performed by edge computing server 120. These components may include software components, hardware components, or a combination thereof that are executable by one or more processors.

Network 200 may be any type of network known to those skilled in the art that may support data communications using any of a number of available protocols, including but not limited to TCP/IP, SNA, IPX, etc. For example only, the one or more networks 110 may be a Local Area Network (LAN), an ethernet-based network, a token ring, a Wide Area Network (WAN), the internet, a virtual network, a Virtual Private Network (VPN), an intranet, an extranet, a blockchain network, a Public Switched Telephone Network (PSTN), an infrared network, a wireless network (e.g., bluetooth, WIFI), and/or any combination of these and/or other networks.

The edge computing server 120 may include one or more special purpose server computers (e.g., PC (personal computer) servers, UNIX servers, middleend servers), blade servers, mainframe computers, server clusters, or any other suitable arrangement and/or combination. Edge computing server 120 may include one or more virtual machines running a virtual operating system, or other computing architecture that involves virtualization (e.g., one or more flexible pools of logical storage devices that may be virtualized to maintain virtual storage devices of the server). In various embodiments, edge computing server 120 may run one or more services or software applications that provide the functionality described below.

The computing units in edge computing server 120 may run one or more operating systems including any of the operating systems described above as well as any commercially available server operating systems. The edge computing server 120 may also run any of a variety of additional server applications and/or middle tier applications, including HTTP servers, FTP servers, CGI servers, JAVA servers, database servers, etc.

In some implementations, the edge computing server 120 may include one or more applications to analyze and consolidate data feeds and/or event updates received from users of the data acquisition devices 101. The edge computing server 120 may also include one or more applications to display data feeds and/or real-time events via one or more display devices of the data acquisition device 101.

In addition, the overall system may also include one or more databases 130. In some embodiments, these databases may be used to store data and other information. Database 130 may reside in various locations. For example, the database used by the edge computing server 120 may be local to the edge computing server 120, or may be remote from the edge computing server 120 and may communicate with the edge computing server 120 via a network-based or dedicated connection. Database 130 may be of different types. In some embodiments, the database used by the edge computing server 120 may be, for example, a relational database. One or more of these databases may store, update, and retrieve the databases and data from the databases in response to the commands.

In some embodiments, one or more of databases 130 may also be used by applications to store application data. The databases used by the application may be different types of databases, such as key value stores, object stores, or conventional stores supported by the file system.

Referring to fig. 2, a flowchart of a method provided in an embodiment of the present application is executed by an edge computing server, and specifically includes the following steps:

STEP101, obtain the goal digital business data through the data acquisition equipment of the edge, produce the digital business information log.

The digital business information processing method combined with the edge calculation can be applied to scenes such as driving assistance, collaborative production and the like, and the data acquisition equipment of the edge end can be sensing equipment on an automobile, sensing equipment in an industrial system and the like. Accordingly, the target digital service data are, for example, running data (such as acceleration, speed, inclination angle, rotation speed, control data, hall data, distance data, etc.) of the automobile, production operation data, etc. The digital service information log is a standard data set obtained by classifying and preprocessing various data, and the included data dimension is determined according to actual conditions, so that the application is not limited. The digital business information log is processed and analyzed, so that the behavior tendency of the user can be identified, targeted auxiliary cooperation can be conveniently carried out based on the identified behavior tendency, for example, in auxiliary driving, comprehensive analysis is carried out according to the set weight based on the required analysis dimension, the driving behavior tendency of the user is identified as dangerous driving, and reasonable intervention is carried out for road safety and personal safety.

STEP102 extracts upper macro characterization knowledge G-K1 of the digital service information log according to a comparison learning mapping unit in the pre-optimized digital service information processing network, wherein the upper macro characterization knowledge G-K1 is used for reflecting the macro dimension data meaning of the digital service information log.

STEP103 extracts behavior trend characterization knowledge I-K1 of the digital service information log according to a behavior trend mapping unit in the digital service information processing network, wherein the behavior trend characterization knowledge I-K1 is used for reflecting the meaning of the data behavior trend of the digital service information log.

In the embodiment of the application, the upper layer macro characterization knowledge reflects macro information (global features) of a shallow layer of the digital service information log, and the behavior tendency characterization knowledge is obtained by extracting deeper knowledge from the whole coding information comprising the digital service information log.

The digital service information processing network combines shallow macroscopic characterization knowledge and deep behavior tendency characterization knowledge, the model comprises a comparison learning mapping unit and a behavior tendency mapping unit, and corresponding characterization knowledge (the characterization knowledge is vector characterization of service information according to different data dimensions, the expression forms of the characterization knowledge are different, such as one-dimensional vector, two-dimensional matrix, multi-order tensor and the like, and the dimension contained in different forms is not limited) is extracted according to the two units.

STEP104 carries out characterization knowledge interaction on upper macro characterization knowledge G-K1 and behavior tendency characterization knowledge I-K1 according to a behavior tendency analysis unit in the digital service information processing network, carries out multi-element classification on the digital service information log according to the obtained first interaction characterization knowledge, and obtains a multi-element classification tendency result of the digital service information log.

In this embodiment of the present application, the digital service information processing network is obtained by performing a combination optimization on the comparison learning mapping unit and the behavior tendency mapping unit, in other words, performing a combination learning on the comparison learning mapping unit and the behavior tendency mapping unit.

Based on the above, the embodiment of the application provides a comparison learning mapping unit for extracting upper macro characterization knowledge and a behavior tendency mapping unit for extracting behavior tendency characterization knowledge, so that a network capable of learning two types of knowledge is obtained, when the behavior tendency analysis is performed on a digital service information log according to the digital service information processing network, the information extraction can be performed on the digital service information by combining the upper macro characterization knowledge and the behavior tendency characterization knowledge, the collaborative learning capability is enhanced, the extraction speed of the characterization knowledge is improved, the behavior tendency analysis of the digital service information is performed according to the obtained digital service information log interaction characterization knowledge, and the obtained multi-element classification tendency result is more accurate and efficient.

In some embodiments, the digital service information log is classified in multiple according to the obtained first interaction characterization knowledge, and a log similar to the digital service information log may be found in a log library, so as to invoke a control policy matched with the similar log to perform subsequent processing, where a specific process of finding the log similar to the digital service information log may include the following steps:

STEP110 extracts upper macro characterization knowledge G-K2 of each alternative digital service information log in the log library according to the comparison learning mapping unit, and extracts behavior tendency characterization knowledge I-K2 of the digital service information log according to the behavior tendency mapping unit.

In the embodiment of the application, the data bottom layer meaning of the upper layer macro characterization knowledge G-K2 and the data bottom layer meaning of the upper layer macro characterization knowledge G-K1 are the same, namely the macro dimension data meaning of the digital service information log is embodied, the upper layer macro characterization knowledge G-K1 aims at the digital service information log, the upper layer macro characterization knowledge G-K2 aims at the alternative digital service information log, and similarly, the data bottom layer meaning of the behavior tendency characterization knowledge I-K2 and the behavior tendency characterization knowledge I-K1 is the same, namely the data behavior tendency meaning of the digital service information log is embodied.

STEP120 performs characterization knowledge interaction on each upper macro characterization knowledge G-K2 and the corresponding behavior tendency characterization knowledge I-K2 according to the behavior tendency analysis unit, and obtains second interaction characterization knowledge corresponding to each alternative digital service information log.

In the process of representing knowledge interaction, feature fusion of upper macro representation knowledge G-K2 and corresponding behavior tendency representation knowledge I-K2 is completed.

STEP130 determines similarity scores of each alternative digital service information log and the digital service information log according to the first interaction characterization knowledge and the second interaction characterization knowledge corresponding to each alternative digital service information log, respectively.

The similarity score is used to characterize the degree of approximation of the alternative digital service information log and the digital service information log, respectively.

STEP140, determining a pairing digital service information log corresponding to the digital service information log according to each similarity score.

Based on this, it is determined that paired digital service information logs corresponding to the digital service information logs in the log library include digital service information logs that are very similar to the digital service information logs, in other words, alternative digital service information logs having a similarity score higher than a preset score. The macroscopic dimension data meaning and the behavior tendency of the digital business information log are related to each other, and certain behavior tendency only appears in a fixed macroscopic data environment, for example, overspeed dangerous driving is usually related to high rotating speed and high acceleration.

Optionally, the digital service information processing network provided in the embodiment of the present application further includes a data filtering unit, and then the digital service information processing network includes at least a data filtering unit, a contrast learning mapping unit, a behavior tendency mapping unit, and a behavior tendency analysis unit.

Optionally, the data filtering unit includes a plurality of feature extraction subunits (i.e. a plurality of convolution layers, the filtering kernel size of each convolution layer is not limited), the structure of the data filtering unit may employ a res net (residual network), the behavior trend mapping unit includes, for example, a feature extraction subunit based on res net and a feature coding subunit, the feature extraction subunit is used to implement deep convolution, the feature coding subunit is used to implement code embedding, the feature extraction subunit constructed by the residual network may extract information to a deeper level, and the stacking layer number thereof is set according to actual needs.

In the embodiment of the application, the parameters corresponding to the feature extraction subunit in the data filtering unit are obtained by performing preliminary assignment on weights and offsets according to the parameters which are calibrated by the pre-deployed example log library, and the initial assignment of the parameters is completed. The data filtering unit may include a plurality of feature extraction subunits connected in sequence, where a filtering core size, a stride, a res net stacking number, a pooling core size, and the like of each feature extraction subunit are set according to actual needs, and parameters of residual units debugged by the plurality of feature extraction subunits are determined as initial parameters.

The contrast learning mapping unit provided by the embodiment of the application comprises a maximum pooling mapping unit and a classification mapping unit, wherein the output sizes of the maximum pooling mapping unit and the classification mapping unit are inconsistent, and the output size of the classification mapping unit is the coding dimension of upper-layer macro characterization knowledge. Optionally, the parameters corresponding to the feature extraction subunit of the behavior tendency mapping unit are obtained according to the random weights and the bias preliminary assignment, so as to complete the initial assignment of the parameters. And the behavior trend mapping unit and the behavior trend analysis unit are used for forming the framework and the parameter weight and selecting according to actual conditions.

According to the method, deep residual error knowledge extraction is added to a knowledge extraction network in a common form, so that a network capable of learning under two types of knowledge is obtained, optimization is performed in a step-by-step multi-level mode based on pre-optimization and continuous correction, a new unit is converged into an identification process, iteration speeds of different learning processes are considered, and finally upper macro characterization knowledge and behavior tendency characterization knowledge are combined to express digital business information, so that accurate analysis of behavior tendency is facilitated.

Optionally, the embodiments of the present application further provide a mechanism for different convergence efficiency parameters (Learning Rate), so as to further enhance the performance of the network. For example, assuming that the convergence efficiency parameter of the data filtering unit, the comparison learning mapping unit, and the behavior tendency mapping unit is a, the convergence efficiency parameter of the behavior tendency analysis unit may be set to 10A, it can be seen that the convergence efficiency parameter of the behavior tendency analysis unit is greater than the convergence efficiency parameter of the other units, that is, the data filtering unit, the comparison learning mapping unit, and the behavior tendency mapping unit.

The multiple classification is to make the example logs with the same classification result output the same estimated classification result, and for the full-connection layer, the fitting phenomenon often occurs, so that embedding fitting is caused, and embedding between the digital service information logs cannot be easily identified.

The following description is made in connection with the process of performing behavior trend analysis on the digital service information log by the digital service information processing network:

STEP10, load the digital business information log into the data filtering unit in the digital business information processing network, carry on the macroscopical representation knowledge extraction to the digital business information log on the basis of the data filtering unit, obtain the macroscopical representation knowledge relation network that the digital business information log corresponds. The macro characterization knowledge relationship network is a stack of multiple two-dimensional tables, each called a feature map.

STEP20, according to comparing the learning mapping unit, map and reduce the dimension to the knowledge relation network of macroscopic representation, obtain the correspondent upper strata macroscopic representation knowledge G-K1 of the information log of digital business. And the embedded coding is completed by mapping and dimension reduction processing on the macroscopic representation knowledge relation network, so that dimension compression is realized, and the data quantity is reduced.

STEP30, extracting behavior trend characterization knowledge I-K1 of the digital service information log according to the behavior trend mapping unit in the digital service information processing network.

STEP40, according to the characteristic extraction subunit in the behavior trend mapping unit, extracting the characterization knowledge of the behavior trend information in the macroscopic characterization knowledge relationship network, and according to the characteristic coding subunit in the behavior trend mapping unit, carrying out mapping dimension reduction processing on the extracted behavior trend information to obtain behavior trend characterization knowledge I-K1 corresponding to the digital service information log.

STEP50 carries out characterization knowledge interaction on upper macro characterization knowledge G-K1 and behavior tendency characterization knowledge I-K1 according to a behavior tendency analysis unit in the digital service information processing network, carries out multi-element classification on the digital service information log according to the obtained first interaction characterization knowledge, and obtains a multi-element classification tendency result of the digital service information log.

In the embodiment of the application, the digital service information processing network performs tuning through the following description.

The digital service information processing network adjusts the example logs, wherein the example logs are multiple example logs, and form an example log group, and one example log group comprises three example logs: comparative example log, true example log, and false example log. The comparative example log and the true example log are the same or similar plant data sets, i.e., the comparative example log and the true example log are digital business information example logs having a similarity score greater than the set similarity score, and the comparative example log and the false example log are dissimilar digital business information logs, i.e., the false example log and the true example log are digital business information example logs having a similarity score not greater than the set similarity score.

In the method, the example logs of the multi-example log learning are grouped and marked, and in the example log marking process of the digital service information log comparison learning, the marked process is multi-example log grouping constructed by three example logs of a preset strategy determined in the whole digital service information log. However, the randomly generated multi-element example log contains more simple and easily-identified simple example logs, which can help the network to perform contrast learning at first, but the network can generate higher distinguishing capability on the simple example logs soon, and the loss of the simple example logs is far beyond the difficult example logs, so that the difficult example logs are covered, and on the basis, more difficult example logs are needed later.

Then, for the generation manner of the example log set, please refer to the following description:

STEP100 constructs, as a true value example log pair, every two digital service information example logs with similarity scores greater than the set similarity score by obtaining similarity scores between the plurality of digital service information example logs.

STEP200 mixes the example logs according to each true value example log pair to obtain a plurality of example log packets for generating an example log set.

In other words, the plurality of obtained digital business information example logs are divided into a plurality of truth example log pairs, and since two example logs in each truth example log pair are the same, or similar, one of them can be determined as a comparative example log in the multiple example log, and the other as a truth example log in the multiple example log. However, the two example logs obtained in different truth example log pairs may be example logs with lower similarity scores, so that example log mixing can be performed on multiple truth example log pairs to complete reorganization, multiple example log groups can be generated according to one truth example log pair, and a sufficient number of example log groups can be owned to generate an example log set after the multiple truth example log pairs are combined.

Optionally, generating a plurality of example log packets according to one truth example log pair may specifically include:

STEP210 determines one digital service information example log in one true example log pair as a target example log, and determines one digital service information example log in the other example log pair as an intermediate example log, respectively.

STEP220 determines spatial similarities between the target example log and each intermediate example log, respectively.

The spatial similarity characterizes the similarity of the target example log and each intermediate example log, which may be determined by calculating a vector distance between the two logs, e.g., cosine distance, euclidean distance, etc., the smaller the vector distance, the higher the spatial similarity.

STEP230, arranging the plurality of intermediate example logs in sequence according to the corresponding spatial similarity, and screening one or more intermediate example logs in a preset sequence as false value example logs corresponding to the target example log.

The preset order is, for example, to arrange the plurality of intermediate example logs in descending order according to the corresponding spatial similarity, and to determine the M latter arranged as false value example logs.

STEP240 respectively constructs each determined false value example log and true value example log pair as an example log group, wherein the target example log in the true value example log pair is the true value example log in the example log group, and the remaining example logs in the true value example log pair are comparative example logs in the example log group.

Each false value example log can be constructed with an example log pair to form an example log group, M false value example logs are determined for one true value example log pair to obtain M example log groups, and M true value example log pairs are included to obtain m×M example log groups. For example, for the truth example log pair I, two digital service information example logs are included: digital service information example log 1 and digital service information example log 2. And determining the digital service information example log 1 as a target example log, and determining 5 digital service information example logs in a truth value example log pair 2-a truth value example log pair 6 as alternative digital service information logs: digital service information example log 3, digital service information example log 4, digital service information example log 6, digital service information example log 7 and digital service information example log 9. Obtaining spatial similarity between each digital service information example log 1 and 5 alternative digital service information logs, setting M=3, and the digital service information example log 3, the digital service information example log 4 and the digital service information example log 6 with minimum spatial similarity between the digital service information example logs 1, and generating three example log groups according to the 3 intermediate example logs to obtain [ digital service information example log 1, digital service information example log 2 and digital service information example log 3]; [ digital service information example log 1, digital service information example log 2, digital service information example log 4]; [ digital service information example log 1, digital service information example log 2, digital service information example log 6].

Specifically, the multi-instance log annotation can be modified as: marking only true value example log pairs results in similar example log pairs, such as by extracting in each batch of example log pairs based on the following process, a multiple example log is obtained, including: and for the target example logs in a truth example log pair, randomly determining a digital business information log in the rest truth example log pairs to serve as intermediate example logs, acquiring the spatial similarity between each intermediate example log and the target example log, determining P intermediate example logs with the minimum spatial similarity as false value example logs according to the size of the spatial similarity, and respectively forming a plurality of example logs with the truth example logs of the target example log group, wherein each example log obtains P plurality of example logs, and the whole batch obtains M times of batch of multi-element example logs.

After the example log is generated, the digital service information processing network is started to be optimally calibrated, and the method specifically comprises the following steps:

STEP310, obtain an example log set, and screen an example log packet from the example log set.

STEP320 loads the filtered example log packet into the calibrated digital service information processing network, and obtains upper macro characterization knowledge G-K3 output by a contrast learning mapping unit, behavior tendency characterization knowledge I-K3 output by a behavior tendency mapping unit and estimated knowledge fields output by a behavior tendency analysis unit in the digital service information processing network. The pre-estimated knowledge field is used for behavior trend analysis, and may be a vector.

STEP330 generates an intermediate network quality evaluation factor according to the upper macro characterization knowledge G-K3 and the estimated knowledge field, and repeatedly adjusts parameters of the behavior tendency mapping unit in the digital service information processing network according to the set adjustment times based on the intermediate network quality evaluation factor. The network quality assessment factor is used to assess the prediction level of the network, which may be a loss function or an error function.

Generating a first multi-element example log quality evaluation factor according to the upper macro characterization knowledge G-K3 and the estimated knowledge field, integrating the first multi-element example log quality evaluation factor, the multi-element classification quality evaluation factor and the estimated entropy quality evaluation factor (for example, directly adding or weighting according to the weight corresponding to each quality evaluation factor and then adding) to obtain an intermediate network quality evaluation factor.

STEP340 generates a target network quality evaluation factor according to the upper macro characterization knowledge G-K3, the behavior tendency characterization knowledge I-K3 and the estimated knowledge field, repeatedly adjusts parameters of the digital service information processing network according to the target network quality evaluation factor until the digital service information processing network meets the preset adjustment requirement, and outputs the adjusted digital service information processing network.

Specifically, a first multi-element example log quality assessment factor, a second multi-element example log quality assessment factor, a multi-element classification quality assessment factor and an estimated entropy quality assessment factor are generated according to upper macro-characterization knowledge G-K3, behavior tendency characterization knowledge I-K3 and estimated knowledge fields, and the first multi-element example log quality assessment factor, the second multi-element example log quality assessment factor, the multi-element classification quality assessment factor and the estimated entropy quality assessment factor are integrated to obtain a target network quality assessment factor. For example, according to the above process, all the example logs in the example log set are repeatedly calibrated for j generations, each generation performs a round of processing on all the example logs, and j is a super parameter, which characterizes the iteration times in all the example log sets. Each generation performs the following process: the full example logs are grouped into one batch per bs of example logs, for each batch: setting all parameters of a network to be learned, carrying out forward calculation on a loaded digital business information log to obtain an estimated knowledge field, and obtaining a multi-element example log error, wherein the multi-element example log error comprises a first multi-element example log quality evaluation factor, a second multi-element example log quality evaluation factor, a multi-element classification quality evaluation factor and an estimated entropy quality evaluation factor, and obtaining a total network quality evaluation factor, and optimizing network parameters according to an SGD algorithm.

In the above STEPs, STEP330 is not essential, and if STEP330 is executed, the full connection layer is optimized as follows: and repeatedly adjusting and straightening the network according to the integration of the three network quality assessment factors until the network quality assessment factors meet the preset adjustment requirement (such as convergence).

In the embodiment of the present application, the network quality assessment factors include three types of network quality assessment factors: a multivariate example log quality assessment factor, a pre-estimated entropy quality assessment factor, and a multivariate classification quality assessment factor. According to the upper macro characterization knowledge and the behavior tendency characterization knowledge, a first multi-element example log quality evaluation factor and a second multi-element example log quality evaluation factor are respectively generated, and in the embodiment of the application, the target network quality evaluation factors are integrated by aiming at the four network quality evaluation factors. Specifically, the network quality assessment factor is obtained by:

STEP (1), according to the upper macro characterization knowledge G-K3 corresponding to each example log in the example log group, a first multi-element example log quality assessment factor is generated.

The upper layer macro characterization knowledge G-K3 and the upper layer macro characterization knowledge G-K1 are consistent with the upper layer macro characterization knowledge G-K2 in terms of embodied information, the upper layer macro characterization knowledge G-K3 is corresponding to the example log, and the behavior tendency characterization knowledge I-K3 is consistent with the meaning of the estimated knowledge field. Generating a first multi-element example log quality assessment factor may specifically include:

STEP (11), the spatial similarity D1 between the upper macro characterization knowledge G-K3 corresponding to the comparison example log and the upper macro characterization knowledge G-K3 corresponding to the true example log in the example log group and the spatial similarity D2 between the upper macro characterization knowledge G-K3 corresponding to the comparison example log and the upper macro characterization knowledge G-K3 corresponding to the false example log are screened.

STEP (12), add the difference result of the spatial similarity D1 and the spatial similarity D2 and a set critical variable, and use the added result as the target variable V1, and set the critical variable to represent the difference result limit of the similarity score between the true value example log and the false value example log.

STEP (13), determining the maximum variable of the target variable V1 and the alternative variable as a first multi-element example log quality assessment factor.

After the multiple example logs are determined in the batch example log, determining a first multiple example log quality assessment factor according to upper layer macro characterization knowledge of the multiple example logs. For example, the multivariate example log quality assessment factor may be determined using the following formula:

loss=jd1-d2|+β; if I D1-D2I +beta < 0

Loss=0; if I D1-D2I +beta is more than or equal to 0

Loss is a quality evaluation factor of a first multi-element example log, beta is a set critical variable, D1 is the spatial similarity of upper-layer macro-characterization knowledge G-K3 corresponding to a comparison example log and upper-layer macro-characterization knowledge G-K3 corresponding to a true example log; d2 is the spatial similarity between the upper layer macro characterization knowledge G-K3 corresponding to the comparative example log and the upper layer macro characterization knowledge G-K3 corresponding to the false value example log. Wherein the candidate variable (reference variable) is equal to 0, if the target variable i D1-D2 i+β (i.e., the target variable V1) is less than 0, the first multi-element example log quality assessment factor is the target variable i D1-D2 i+β, and if the target variable i D1-D2 i+β is not less than 0, the first multi-element example log quality assessment factor is equal to 0.

The first multi-example log quality assessment factor is configured to make a spatial similarity of the upper layer macro-characterization knowledge of the comparison example log and the false value example log smaller than β than a spatial similarity of the upper layer macro-characterization knowledge of the comparison example log and the true value example log. According to the network quality evaluation factor, the situation that the upper-layer macro characterization knowledge of the comparison example log is close to that of the true value example log can be ensured, and the upper-layer macro characterization knowledge of the comparison example log is far away from that of the false value example log, so that the upper-layer macro characterization knowledge between the true value example log and the false value example log has low similarity, and the distinguishing degree of trend recognition is enhanced.

STEP (2), according to the behavior tendency characterization knowledge I-K3 corresponding to each example log, generating a second multi-element example log quality evaluation factor.

The process of generating the second multi-example log quality assessment factor may refer to the process of generating the first multi-example log quality assessment factor, and specifically includes the following steps:

STEP (21), according to the spatial similarity D3 between the behavior trend characterization knowledge I-K3 corresponding to the comparison example log and the behavior trend characterization knowledge I-K3 corresponding to the true example log in the example log group, the spatial similarity D4 between the behavior trend characterization knowledge I-K3 corresponding to the comparison example log and the behavior trend characterization knowledge I-K3 corresponding to the false example log.

STEP (22), adding the difference result of the spatial similarity D3 and the spatial similarity D4 and a set critical variable, taking the added result as a target variable V2, and a third set critical variable is used for representing the difference result limit of the similarity score between the true value example log and the false value example log.

STEP (23), determining the maximum variable of the target variable V2 and the alternative variable as a second multi-element example log quality assessment factor.

The calculation process adopts a formula referring to the calculation formula of the log quality evaluation factor Loss of the first multi-element example. The second multi-element example log quality evaluation factor is used for enabling the spatial similarity of the behavior trend characterization knowledge of the comparison example log and the behavior trend characterization knowledge of the false value example log to be smaller than the spatial similarity of the behavior trend characterization knowledge of the comparison example log and the behavior trend characterization knowledge of the true value example log by beta, according to the network quality evaluation factor, the comparison example log and the behavior trend characterization knowledge of the true value example log can be ensured to be close, the behavior trend characterization knowledge of the comparison example log and the behavior trend characterization knowledge of the false value example log are far away, the behavior trend characterization knowledge between the true value example log and the false value example log is ensured to have smaller similarity, and the distinguishing degree of trend recognition is enhanced.

STEP (3), according to the estimated knowledge field and the corresponding multi-element classification knowledge field corresponding to each example log, generating multi-element classification quality evaluation factors, wherein the estimated knowledge field represents the estimated credibility coefficient corresponding to each classification trend result of the example log, and the multi-element classification knowledge field represents the actual credibility coefficient corresponding to each classification trend result of the example log.

And determining the multi-classification quality evaluation factors of the multi-classification knowledge fields obtained based on multi-classification labeling and the pre-estimated knowledge fields (namely the classification knowledge fields) output by the full connection layer. The purpose of full-connected class learning is to constrain the class-embedded code so that it contains multiple class-prone results. Because the over-fitting condition is easy to occur to cause the insufficient identification of embedding discrimination, the bias vector of the full-connection layer is only 10% of the original gradient when the bias vector of the full-connection layer is reversely transferred and updated to the behavior trend information coding through setting the convergence efficiency parameter of the full-connection layer which is ten times that of other units, and the over-fitting is stopped.

STEP (4), according to the estimated knowledge field corresponding to each example log, obtaining estimated entropy quality evaluation factors.

The estimated entropy quality evaluation factor is used for avoiding the overfitting problem in the prediction of the behavior tendency, the estimated result of the classification of the full-connection layer is subjected to entropy calculation, the scattering cost is added, the expected entropy value is maximized, the estimated result of the network is ensured not to be concentrated on the single behavior tendency labels to be close, the estimated result of the plurality of behavior tendency labels is expected to be balanced instead, and meanwhile, the disturbance on classification is ensured to be small. The calculation formula can refer to the following formula:

Wherein, loss' is an estimated entropy quality evaluation factor; x is the number of example log pairs; s is S _mn The estimated result of the nth behavior tendency of the example log m is shown.

And STEP (5) integrating the first multi-element example log quality assessment factor, the second multi-element example log quality assessment factor, the multi-element classification quality assessment factor and the estimated entropy quality assessment factor to obtain a target network quality assessment factor.

The integration method is, for example, directly adding, or multiplying each quality evaluation factor by a corresponding weight, and then adding to obtain the target network quality evaluation factor.

According to the digital business information processing method combining edge calculation, through the contrast learning mapping unit for extracting upper macro characterization knowledge and the behavior tendency mapping unit for extracting behavior tendency characterization knowledge, a network capable of learning two types of knowledge is obtained, when behavior tendency analysis is carried out on the digital business information log according to the digital business information processing network, information extraction can be carried out on the digital business information by combining the upper macro characterization knowledge and the behavior tendency characterization knowledge, collaborative learning capacity is enhanced, the speed of characterization knowledge extraction is improved, and digital business information behavior tendency analysis is carried out according to the obtained digital business information log interaction characterization knowledge, so that the obtained multi-element classification tendency result is more accurate and efficient.

According to another aspect of the present application, there is also provided a digital service information processing apparatus 900, please refer to fig. 3, the digital service information processing apparatus 900 includes:

the service log obtaining module 910 is configured to obtain target digital service data through a data acquisition device at an edge end, and generate a digital service information log;

the upper knowledge extraction module 920 is configured to extract upper macro characterization knowledge G-K1 of the digital service information log according to a comparison learning mapping unit in the pre-optimized digital service information processing network; the upper layer macro characterization knowledge G-K1 is used for reflecting macro dimension data meanings of the digital service information log;

a trend knowledge extraction module 930, configured to extract a behavior trend characterization knowledge I-K1 of the digital service information log according to a behavior trend mapping unit in the digital service information processing network; the behavior tendency characterization knowledge I-K1 is used for reflecting the data behavior tendency meaning of the digital business information log;

the behavior trend identification module 940 is configured to perform characterization knowledge interaction on the upper macro characterization knowledge G-K1 and the behavior trend characterization knowledge I-K1 according to a behavior trend analysis unit in the digital service information processing network, and perform multiple classification on the digital service information log according to the obtained first interaction characterization knowledge, so as to obtain a multiple classification trend result of the digital service information log; the digital business information processing network is obtained based on the combination optimization of the comparison learning mapping unit and the behavior tendency mapping unit.

According to embodiments of the present application, there is also provided an edge computing server, a readable storage medium, and a computer program product.

Referring to fig. 4, which is a block diagram of the structure of the edge computing server 1000 of the present application, the edge computing server 1000 includes a computing unit 1001 that can perform various appropriate actions and processes according to a computer program stored in a Read Only Memory (ROM) 1002 or a computer program loaded from a storage unit 1008 into a Random Access Memory (RAM) 1003. In the RAM 1003, various programs and data required for the operation of the edge calculation server 1000 can also be stored. The computing unit 1001, the ROM 1002, and the RAM 1003 are connected to each other by a bus 1004. An input/output (I/O) interface 1005 is also connected to bus 1004.

The various components in edge computing server 1000 are connected to I/O interface 1005, including: an input unit 1006, an output unit 1007, a storage unit 1008, and a communication unit 10010. The input unit 1006 may be any type of device capable of inputting information to the edge computing server 1000, the input unit 1006 may receive input numeric or character information and generate key signal inputs related to user settings and/or function controls of the edge computing server, and may include, but is not limited to, a mouse, a keyboard, a touch screen, a track pad, a track ball, a joystick, a microphone, and/or a remote control. The output unit 1007 may be any type of device capable of presenting information and may include, but is not limited to, a display, speakers, video/audio output terminals, vibrators, and/or printers. Storage unit 1008 may include, but is not limited to, magnetic disks, optical disks. The communication unit 10010 allows the edge computing server 1000 to exchange information/data with other devices via a computer network, such as the internet, and/or various telecommunications networks, and may include, but is not limited to, modems, network cards, infrared communication devices, wireless communication transceivers, and/or chipsets, such as bluetooth (TM) devices, 802.11 devices, wiFi devices, wiMax devices, cellular communication devices, and/or the like.

The computing unit 1001 may be a variety of general and/or special purpose processing components having processing and computing capabilities. Some examples of computing unit 1001 include, but are not limited to, a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), various specialized Artificial Intelligence (AI) computing chips, various computing units running machine learning model algorithms, a Digital Signal Processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 1001 performs the various methods and processes described above, such as method 200. For example, in some embodiments, the method 200 may be implemented as a computer software program tangibly embodied on a machine-readable medium, such as the storage unit 1008. In some embodiments, part or all of the computer program may be loaded and/or installed onto the edge computing server 1000 via the ROM 1002 and/or the communication unit 1009. One or more of the steps of the method 200 described above may be performed when the computer program is loaded into RAM 1003 and executed by the computing unit 1001. Alternatively, in other embodiments, the computing unit 1001 may be configured to perform the method 200 in any other suitable way (e.g., by means of firmware).

Various implementations of the systems and techniques described here above may be implemented in digital electronic circuitry, integrated circuit systems, field Programmable Gate Arrays (FPGAs), application Specific Integrated Circuits (ASICs), application Specific Standard Products (ASSPs), systems On Chip (SOCs), complex Programmable Logic Devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs, the one or more computer programs may be executed and/or interpreted on a programmable system including at least one programmable processor, which may be a special purpose or general-purpose programmable processor, that may receive data and instructions from, and transmit data and instructions to, a storage system, at least one input device, and at least one output device.

Program code for carrying out methods of the present application may be written in any combination of one or more programming languages. These program code may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus such that the program code, when executed by the processor or controller, causes the functions/operations specified in the flowchart and/or block diagram to be implemented. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.

In the context of this application, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. The machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and pointing device (e.g., a mouse or trackball) by which a user can provide input to the computer. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form, including acoustic input, speech input, or tactile input.

The systems and techniques described here can be implemented in a computing system that includes a background component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such background, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local Area Networks (LANs), wide Area Networks (WANs), and the internet.

The computer system may include a client and a server. The client and server are typically remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server may be a cloud server, a server of a distributed system, or a server incorporating a blockchain.

It should be appreciated that various forms of the flows shown above may be used to reorder, add, or delete steps. For example, the steps recited in the present application may be performed in parallel, sequentially or in a different order, provided that the desired results of the technical solutions disclosed herein are achieved, and are not limited herein.

Although embodiments or examples of the present application have been described with reference to the accompanying drawings, it is to be understood that the foregoing methods, systems, and apparatus are merely exemplary embodiments or examples, and that the scope of the present invention is not limited by these embodiments or examples but only by the claims following the grant and their equivalents. Various elements of the embodiments or examples may be omitted or replaced with equivalent elements thereof. Furthermore, the steps may be performed in a different order than described in the present application. Further, various elements of the embodiments or examples may be combined in various ways. It is important that as technology evolves, many of the elements described herein may be replaced by equivalent elements that appear after the application.

Claims

1. A digital service information processing method in combination with edge computation, applied to an edge computation server, the method comprising:

2. The method of claim 1, wherein the multi-categorizing the digital business information log based on the obtained first interaction characterization knowledge further comprises:

3. The method of claim 1, wherein the digital service information processing network further comprises a data filtering unit; before extracting the upper macro characterization knowledge G-K1 of the digital service information log according to the comparison learning mapping unit in the pre-optimized digital service information processing network, the method further comprises the following steps:

4. A method according to claim 3, wherein the data filtering unit comprises a plurality of feature extraction subunits; the parameters corresponding to the feature extraction subunit in the data filtering unit are obtained by carrying out preliminary assignment on weights and offsets according to the parameters which are calibrated by the pre-deployed example log library, and the parameters corresponding to the feature extraction subunit in the behavior tendency mapping unit are obtained by carrying out preliminary assignment on random weights and offsets.

5. The method of claim 1, further comprising an optimization process of the digital service information processing network, comprising:

6. The method of claim 5, wherein generating the network quality assessment factor from the upper layer macro characterization knowledge G-K3, the behavioral trend characterization knowledge I-K3, and the predictive knowledge field comprises:

7. The method of claim 6, wherein each of the example log packets encompasses a comparative example log, a true example log, and a false example log;

8. The method of claim 6, wherein each of the example log packets encompasses a comparative example log, a true example log, and a false example log; the step of generating a second multi-element example log quality evaluation factor according to the behavior tendency characterization knowledge I-K3 corresponding to each example log comprises the following steps:

9. The method of claim 6, wherein prior to iteratively adjusting parameters of the digital service information processing network in accordance with the target network quality assessment factor, the method further comprises:

10. An edge computing server, comprising:

At least one processor;

and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the at least one processor to cause the at least one processor to perform the method of any one of claims 1-9.