CN115292475A

CN115292475A - Cloud computing service information processing method and system based on smart city

Info

Publication number: CN115292475A
Application number: CN202210747073.1A
Authority: CN
Inventors: 李宁
Original assignee: Individual
Current assignee: Individual
Priority date: 2022-06-29
Filing date: 2022-06-29
Publication date: 2022-11-04

Abstract

The embodiment of the application discloses a cloud computing service information processing method and system based on a smart city; the method comprises the steps that in response to a service demand identification instruction, X-type user service demand identification is carried out on selected intelligent service session records to obtain X-type demand knowledge phrase relationship networks Map1, and then each demand knowledge phrase relationship network Map1 and the demand knowledge phrase relationship network in contact with the demand knowledge phrase relationship network Map1 are sorted to obtain X-type demand knowledge phrase relationship networks Map2; the cloud service preference analysis is carried out on the X-type requirement knowledge phrase relation network Map2 to obtain user preference information, then the multi-type requirement knowledge phrases of the X-type requirement knowledge phrase relation network Map1 can be sorted and combined to obtain the comprehensive and accurate X-type requirement knowledge phrase relation network Map2, when the cloud service preference analysis is carried out on the basis of the requirement knowledge phrase relation network Map2, the relation among different requirements can be considered, the user preference is deeply mined, and the richness and accuracy of the obtained user preference information are guaranteed.

Description

Cloud computing service information processing method and system based on smart city

Technical Field

The application relates to the technical field of cloud computing, in particular to a cloud computing service information processing method and system based on a smart city.

Background

Smart City (Smart City) is a City informationization advanced form which fully applies a new generation of information technology to various industries in the City and is based on the innovation of the next generation of knowledge society, realizes the deep integration of informatization, industrialization and urbanization, is beneficial to relieving the large City diseases, improving the urbanization quality, realizing the fine and dynamic management, improving the City management effect and improving the life quality of citizens.

Currently, various cloud services in a smart city mode can provide a lot of convenience for work and life of people, and in order to further realize service function upgrading of the smart city, related technologies are focused on user analysis and mining, but the richness and the accuracy of user preference analysis are difficult to guarantee in the implementation process of the technologies.

Disclosure of Invention

An object of the present application is to provide a cloud computing service information processing method and system based on a smart city.

The technical scheme of the application is realized by at least some of the following embodiments.

A cloud computing service information processing method based on a smart city is applied to a cloud computing service processing system, and comprises the following steps:

responding to a service requirement identification instruction, carrying out X-type user service requirement identification on the selected intelligent service session record, and determining an X-type requirement knowledge phrase relation network Map1 of the selected intelligent service session record; the relation network dimensionality of each type of requirement knowledge phrase relation network Map1 in the X type of requirement knowledge phrase relation network Map1 is different, X is greater than 1 and belongs to Z;

respectively carrying out relation network dimension change and arrangement on the demand knowledge phrase relation network sets corresponding to the various demand knowledge phrase relation networks Map1 to obtain X-type demand knowledge phrase relation networks Map2; each demand knowledge phrase relationship network set comprises the demand knowledge phrase relationship network Map1 and a demand knowledge phrase relationship network Map1' which is in contact with the demand knowledge phrase relationship network Map1;

and carrying out cloud service preference analysis on the X-type requirement knowledge phrase relation network Map2, and determining the user preference information of the selected intelligent service session record.

In one possible embodiment, the requirement knowledge phrase relationship network set corresponding to the g-th class requirement knowledge phrase relationship network Map1 comprises a g-1-th class requirement knowledge phrase relationship network Map1, a g-th class requirement knowledge phrase relationship network Map1 and a g + 1-th class requirement knowledge phrase relationship network Map1, 1-g-s-X and g ∈ X; the method for respectively changing and sorting the relation network dimensions of the demand knowledge phrase relation network sets corresponding to the various demand knowledge phrase relation networks Map1 to obtain the X-type demand knowledge phrase relation network Map2 comprises the following steps:

performing relational network dimension compression on the g-1 type demand knowledge phrase relational network Map1 to obtain a first g type demand knowledge phrase relational network Map3;

carrying out relation network dimension adjustment on the g-th class requirement knowledge phrase relation network Map1 to obtain a second g-th class requirement knowledge phrase relation network Map3;

carrying out relation network dimension derivation on the g +1 th class requirement knowledge phrase relation network Map1 to obtain a third g class requirement knowledge phrase relation network Map3;

sorting the first g type requirement knowledge phrase relation network Map3, the second g type requirement knowledge phrase relation network Map3 and the third g type requirement knowledge phrase relation network Map3 to obtain a g type requirement knowledge phrase relation network Map2; the dimensionality of the relationship network of the first g-th class requirement knowledge phrase relationship network Map3, the second g-th class requirement knowledge phrase relationship network Map3 and the third g-th class requirement knowledge phrase relationship network Map3 is the same.

In a possible embodiment, the requirement knowledge phrase relationship network set corresponding to the class 1 requirement knowledge phrase relationship network Map1 includes the class 1 requirement knowledge phrase relationship network Map1 and the class 2 requirement knowledge phrase relationship network Map1, and the step of performing relationship network dimension change and sorting on the requirement knowledge phrase relationship network set corresponding to the class 1 requirement knowledge phrase relationship network Map1 respectively to obtain the class X requirement knowledge phrase relationship network Map2 includes:

carrying out relation network dimension adjustment on the class 1 requirement knowledge phrase relation network Map1 to obtain a first class 1 requirement knowledge phrase relation network Map3;

carrying out relation network dimension derivation on the 2 nd type requirement knowledge phrase relation network Map1 to obtain a second 1 st type requirement knowledge phrase relation network Map3;

sorting the first class 1 requirement knowledge phrase relationship network Map3 and the second class 1 requirement knowledge phrase relationship network Map3 to obtain a class 1 requirement knowledge phrase relationship network Map2; and the dimension of the relationship network of the first class 1 requirement knowledge phrase relationship network Map3 is the same as that of the second class 1 requirement knowledge phrase relationship network Map3.

In a possible embodiment, the requirement knowledge phrase relationship network set corresponding to the X-th type requirement knowledge phrase relationship network Map1 includes an X-1-th type requirement knowledge phrase relationship network Map1 and the X-th type requirement knowledge phrase relationship network Map1, and the relationship network dimension change and arrangement are respectively performed on the requirement knowledge phrase relationship network set corresponding to the various types of requirement knowledge phrase relationship networks Map1 to obtain an X-type requirement knowledge phrase relationship network Map2, which includes:

carrying out relational network dimension compression on the X-1 type requirement knowledge phrase relational network Map1 to obtain a first X type requirement knowledge phrase relational network Map3;

carrying out relation network dimension adjustment on the X-th class requirement knowledge phrase relation network Map1 to obtain a second X-th class requirement knowledge phrase relation network Map3;

sorting the first class X requirement knowledge phrase relationship network Map3 and the second class X requirement knowledge phrase relationship network Map3 to obtain a class X requirement knowledge phrase relationship network Map2; the first xth-type requirement knowledge phrase relationship network Map3 and the second xth-type requirement knowledge phrase relationship network Map3 have the same relationship network dimension.

In a possible embodiment, the performing the relationship network dimension compression on the g-1 th class requirement knowledge phrase relationship network Map1 to obtain a first g-th class requirement knowledge phrase relationship network Map3 includes: processing the g-1 type requirement knowledge phrase relationship net Map1 through a first feature filtering unit to determine a first g type requirement knowledge phrase relationship net Map3, wherein the feature filtering node size of the first feature filtering unit is S, the sliding variable is S, S and S are greater than 1 and belongs to Z, and the relationship net dimension of the g-1 type requirement knowledge phrase relationship net Map1 is S times of the relationship net dimension of the g type requirement knowledge phrase relationship net Map1;

performing relationship network dimension adjustment on the g-th class requirement knowledge phrase relationship network Map1 to obtain a second g-th class requirement knowledge phrase relationship network Map3, including: processing the g-th class requirement knowledge phrase relationship network Map1 through a second feature filtering unit to determine a second g-th class requirement knowledge phrase relationship network Map3, wherein the feature filtering node size of the second feature filtering unit is S, and a sliding variable is a set value;

performing relational network dimension derivation on the g +1 th class requirement knowledge phrase relational network Map1 to obtain a third g class requirement knowledge phrase relational network Map3, including: and processing and characteristic derivation are carried out on the g + 1-th class requirement knowledge phrase relation network Map1 through a third characteristic filtering unit and a characteristic derivation unit, so as to determine a third g-th class requirement knowledge phrase relation network Map3, wherein the size of a characteristic filtering node of the third characteristic filtering unit is S x S, and a sliding variable is a set value.

In a possible embodiment, the performing relationship network dimension adjustment on the class 1 requirement knowledge phrase relationship network Map1 to obtain a first class 1 requirement knowledge phrase relationship network Map3 includes: processing the 1 st type requirement knowledge phrase relationship network Map1 through a second feature filtering unit to determine a first 1 st type requirement knowledge phrase relationship network Map3; the size of a characteristic filtering node of the second characteristic filtering unit is S, a sliding variable is a set value, S is greater than 1, and S belongs to Z;

performing relation network dimension derivation on the class-2 requirement knowledge phrase relation network Map1 to obtain a second class-1 requirement knowledge phrase relation network Map3, including: processing and characteristic derivation are carried out on the class 2 requirement knowledge phrase relationship network Map1 through a third characteristic filtering unit and a characteristic derivation unit, and a second class 1 requirement knowledge phrase relationship network Map3 is obtained; and the size of the feature filtering node of the third feature filtering unit is S, and the sliding variable is a set value.

In a possible embodiment, the performing the relational network dimension compression on the X-1 th class requirement knowledge phrase relational network Map1 to obtain a first X-1 th class requirement knowledge phrase relational network Map3 includes: processing the X-1 type requirement knowledge phrase relationship net Map1 through a first feature filtering unit to determine a first X type requirement knowledge phrase relationship net Map3, wherein the feature filtering node size of the first feature filtering unit is S, the sliding variable is S, S and S are greater than 1 and belongs to Z, and the relationship net dimension of the g-1 type requirement knowledge phrase relationship net Map1 is S times of the relationship net dimension of the g type requirement knowledge phrase relationship net Map1;

performing relationship network dimension adjustment on the class X requirement knowledge phrase relationship network Map1 to obtain a second class X requirement knowledge phrase relationship network Map3, including: and processing the class X requirement knowledge phrase relationship network Map1 through a second feature filtering unit to determine a second class X requirement knowledge phrase relationship network Map3, wherein the size of a feature filtering node of the second feature filtering unit is S, and a sliding variable is a set value.

In a possible embodiment, the second and third eigen filter units comprise adjustable eigenfilter units or expansion eigenfilter units.

In a possible embodiment, the method is realized through a GCN algorithm, the GCN algorithm comprises cascaded Q-class sorting model layers and is used for carrying out Q-time relation network dimension change and sorting on the X-class requirement knowledge phrase relation network Map1, each level of sorting model layers comprises a plurality of first characteristic filtering units, a plurality of second characteristic filtering units and a plurality of third characteristic filtering units, Q is greater than 0, and Q is equal to Z; the method for respectively changing and sorting the relation network dimensions of the demand knowledge phrase relation network sets corresponding to the various demand knowledge phrase relation networks Map1 to obtain the X-type demand knowledge phrase relation network Map2 comprises the following steps:

transmitting the X-type requirement knowledge phrase relationship network Map1 into a 1 st type sorting model layer to obtain a 1 st sorted X-type requirement knowledge phrase relationship network Map4;

transmitting the X-type demand knowledge phrase relationship networks Map4 sorted in the d-1 th round into a d-type sorting model layer to obtain the X-type demand knowledge phrase relationship networks Map4 sorted in the d-th round, wherein 1 & ltd & gt Q & ltd & gt belongs to Z.

In a possible embodiment, each level of finishing model layer further includes a normalization layer, and the importing the class-X requirement knowledge phrase relationship network Map4 sorted in the d-1 th round into the class-d finishing model layer to obtain the class-X requirement knowledge phrase relationship network Map4 sorted in the d-th round includes:

respectively carrying out relation network dimension change and arrangement on a requirement knowledge phrase relation network set corresponding to the D-1 th round of arranged X-type requirement knowledge phrase relation network Map4 through a first characteristic filtering unit, a second characteristic filtering unit and a third characteristic filtering unit of the d-th round of arrangement model layer to determine an X-type to-be-processed knowledge phrase relation network of the d-th round of arrangement;

and standardizing the X-type knowledge phrase relationship networks to be processed in the d-th round of arrangement through the normaize layer, and determining the X-type requirement knowledge phrase relationship network Map4 in the d-th round of arrangement.

In a possible embodiment, the method is implemented by a GCN algorithm, the GCN algorithm further includes a linear capture algorithm and a naive bayesian algorithm, and the cloud service preference analysis is performed on the class-X requirement knowledge phrase relationship network Map2 to determine the user preference information of the selected smart service session record, including:

transmitting the X-class requirement knowledge phrase relation network Map2 into the linear capture algorithm, and determining session content corresponding to user preference in the selected intelligent service session record;

and transmitting the X-class requirement knowledge phrase relationship network Map2 into the naive Bayesian algorithm to determine a distinguishing label of the user preference in the selected intelligent service session record, wherein the user preference information comprises session content corresponding to the user preference and the distinguishing label of the user preference.

A cloud computing services processing system, comprising: a memory for storing an executable computer program, a processor for implementing the above method when executing the executable computer program stored in the memory.

A computer-readable storage medium, on which a computer program is stored which, when executed, performs the above-mentioned method.

According to an embodiment of the application, in response to a service demand identification instruction, X-type user service demand identification can be carried out on a selected intelligent service session record to obtain an X-type demand knowledge phrase relationship network Map1, and then each demand knowledge phrase relationship network Map1 and a demand knowledge phrase relationship network in contact with the demand knowledge phrase relationship network Map1 are sorted to obtain an X-type demand knowledge phrase relationship network Map2; the cloud service preference analysis is carried out on the X-type requirement knowledge phrase relation network Map2 to obtain user preference information, then the multi-type requirement knowledge phrases of the X-type requirement knowledge phrase relation network Map1 can be sorted and combined to obtain the comprehensive and accurate X-type requirement knowledge phrase relation network Map2, when the cloud service preference analysis is carried out on the basis of the requirement knowledge phrase relation network Map2, the relation among different requirements can be considered, the user preference is deeply mined, and the richness and accuracy of the obtained user preference information are guaranteed.

Drawings

Fig. 1 is a schematic diagram illustrating one communication configuration of a cloud computing service processing system in which an embodiment of the present application can be implemented.

Fig. 2 is a flowchart illustrating a smart city-based cloud computing service information processing method in which an embodiment of the present application may be implemented.

Fig. 3 is an architecture diagram illustrating an application environment in which the smart city-based cloud computing service information processing method according to the embodiment of the present application can be implemented.

Detailed Description

The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be described in further detail with reference to the accompanying drawings, the described embodiments should not be considered as limiting the present application, and all other embodiments obtained by a person of ordinary skill in the art without making creative efforts fall within the protection scope of the present application. In the following description, reference is made to "some embodiments" which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of the present application only and is not intended to be limiting of the application.

Fig. 1 is a block diagram illustrating a communication configuration of a cloud computing service processing system 100 that can implement an embodiment of the present application, where the cloud computing service processing system 100 includes a memory 101 for storing an executable computer program, and a processor 102 for implementing a smart city-based cloud computing service information processing method in the embodiment of the present application when executing the executable computer program stored in the memory 101.

Fig. 2 is a schematic flowchart illustrating a smart city-based cloud computing service information processing method that can implement an embodiment of the present application, where the smart city-based cloud computing service information processing method can be implemented by the cloud computing service processing system 100 shown in fig. 1, and further may include the technical solutions described in the following related steps.

Step11, responding to the service requirement identification instruction, carrying out X-type user service requirement identification on the selected intelligent service session record, and determining the X-type requirement knowledge phrase relation Map1 of the selected intelligent service session record.

The relation network dimensionality of each type of requirement knowledge phrase relation network Map1 in the X type of requirement knowledge phrase relation network Map1 is different, X is greater than 1, and X belongs to Z. Further, the relationship network dimensions of the various demand knowledge phrase relationship networks Map1 are different, which means that the relationship network dimensions/the relationship network dimensions in the various demand knowledge phrase relationship networks Map1 are different.

In this embodiment of the application, after receiving a service requirement identification instruction uploaded by another system, the cloud computing service processing system may identify a selected smart service session record according to the service requirement identification instruction before performing class-X user service requirement identification on the selected smart service session record, and further, the cloud computing service processing system performs multiple classes of user service requirement identification/extraction on the selected smart service session record according to the service requirement identification instruction, so as to obtain multiple classes of first requirement knowledge characteristic information distribution (i.e., requirement knowledge phrase relationship network Map 1), and in addition, the multiple classes of first requirement knowledge characteristic information in the selected smart service session record may be recorded in a form of a characteristic Map or a characteristic vector.

For example, the selected intelligent service session record may be understood as a pending intelligent service session record, such as a government-enterprise service session record, a digital office service session record, an information security service session record, a digital financial service session record, and the like, without limitation.

And Step12, respectively carrying out relation network dimension change and arrangement on the demand knowledge phrase relation network sets corresponding to the various demand knowledge phrase relation networks Map1 to obtain an X-type demand knowledge phrase relation network Map2.

Each requirement knowledge phrase relationship network set comprises the requirement knowledge phrase relationship network Map1 and a requirement knowledge phrase relationship network Map1' which is in contact with the requirement knowledge phrase relationship network Map1.

In the embodiment of the application, the relationship network dimension change and arrangement of the demand knowledge phrase relationship network sets corresponding to the various demand knowledge phrase relationship networks Map1 can be understood as that demand knowledge phrase relationship network groups corresponding to the various demand knowledge phrase relationship networks Map1 are adjusted and fused one by one, and then second demand knowledge characteristic information distributions (namely, demand knowledge phrase relationship networks Map 2) of multiple categories are obtained.

And Step13, carrying out cloud service preference analysis on the X-type demand knowledge phrase relation network Map2, and determining the user preference information of the selected intelligent service session record.

In the embodiment of the application, the cloud service preference analysis on the X-type demand knowledge phrase relationship network Map2 may be cloud service tendency/habit mining on multiple categories of demand knowledge phrase relationship networks Map2.

For example, the selected smart service session record may be a service session record including cloud service preferences (e.g., visual interface optimization, virtual digital space role skin update, government and enterprise business authority activation, etc.), and the selected smart service session record may be obtained in connection with a service session capture thread (e.g., a service session capture module configured in a cloud computing service processing system) or in connection with other ideas.

In an exemplary embodiment, in Step11, for example, the relation network classification model may be combined to perform multi-class user service requirement identification on the selected smart service session record, and a requirement knowledge phrase relation network is extracted from the category units with differences in the relation network classification model, so as to obtain an X-class requirement knowledge phrase relation network Map1 (which may be understood as a relation network classification model) of the selected smart service session record, where X >1 and X ∈ Z. The relation network dimensionality of each type of requirement knowledge phrase relation network Map1 in the X type of requirement knowledge phrase relation network Map1 is different. The relational network classification model can comprise at least X characteristic filtering units, a down-sampling unit and the like, and the embodiment of the application does not limit the specific model architecture of the relational network classification model. By adopting the service session record with single dimension for analysis, the caching amount of the cache space and the resource consumption can be reduced.

In the subsequent operation flow, if the X-class requirement knowledge phrase relationship network Map1 is immediately sorted, the label information between different layers can be sorted out, but the association degree of the relationship network between the unit layers with the association cannot be further determined. Based on this, sorting/weighting/fusion between various demand knowledge phrase relationship networks Map1 and demand knowledge phrase relationship networks Map1' with relation can be realized through Step 12.

In an illustrative embodiment, in Step12, relationship network dimensions can be changed and sorted respectively for requirement knowledge phrase relationship network sets corresponding to various requirement knowledge phrase relationship networks Map1 to obtain X-type requirement knowledge phrase relationship networks Map2, and each requirement knowledge phrase relationship network set includes the requirement knowledge phrase relationship network Map1 and a requirement knowledge phrase relationship network Map1' linked with the requirement knowledge phrase relationship network Map1. Such as: for any requirement knowledge phrase relationship network Map1, the relationship network dimension of 2e requirement knowledge phrase relationship networks (for example, the upper requirement knowledge phrase relationship network and the lower requirement knowledge phrase relationship network are respectively selected from e) with connection is changed to be the same as the relationship network dimension of the requirement knowledge phrase relationship network Map1, then the changed 2e requirement knowledge phrase relationship networks and the requirement knowledge phrase relationship network Map1 are weighted, so that a requirement knowledge phrase relationship network Map2 corresponding to the requirement knowledge phrase relationship network Map1 is obtained, and e is more than or equal to 1.

In an illustrative embodiment, the relationship network dimensions of the requirement knowledge phrase relationship network set of the requirement knowledge phrase relationship network Map1 (including the requirement knowledge phrase relationship network Map1 and the 2e requirement knowledge phrase relationship networks with links) may also be unified to the specified relationship network dimensions, such as: expanding the requirement knowledge phrase relationship networks in the requirement knowledge phrase relationship network set to multiple of the relationship network dimension of the requirement knowledge phrase relationship network Map1, or compressing the requirement knowledge phrase relationship networks to the set proportion of the relationship network dimension of the requirement knowledge phrase relationship network Map1. And then weighting each changed requirement knowledge phrase relationship network to obtain a requirement knowledge phrase relationship network Map2 corresponding to the requirement knowledge phrase relationship network Map1.

Therefore, the association characteristics of the required knowledge phrase relationship network level and the association characteristics of the session scene level can be captured, and the accuracy and the reliability of the required knowledge phrase relationship network obtained through sorting are improved.

In an exemplary embodiment, cloud service preference analysis may be performed on the class X requirement knowledge phrase relationship network Map2 at Step13 to obtain user preference information of the selected wisdom service session record. Such as: and respectively carrying out prediction and clustering operations on the X-type demand knowledge phrase relation network Map2. After the prediction operation is carried out, a service session record list (such as a windowing analysis result) corresponding to the cloud service preference in the selected intelligent service session record can be determined; after the clustering operation, a distinguishing label of user preferences in the selected smart service session record can be determined. The user preference information of the selected intelligent service session record may include a list of service session records corresponding to the user preferences in the selected intelligent service session record, a distinguishing label of the user preferences, and the like.

By implementing the technical scheme recorded by the steps 11-13, the service requirement identification instruction can be responded, and the X-type user service requirement identification can be carried out on the selected intelligent service session record to obtain the X-type requirement knowledge phrase relation Map1; sorting each demand knowledge phrase relationship network Map1 and the demand knowledge phrase relationship network in contact with the demand knowledge phrase relationship network to obtain an X-type demand knowledge phrase relationship network Map2; the cloud service preference analysis is carried out on the X-type requirement knowledge phrase relation network Map2 to obtain user preference information, then the multi-type requirement knowledge phrases of the X-type requirement knowledge phrase relation network Map1 can be sorted and combined to obtain the comprehensive and accurate X-type requirement knowledge phrase relation network Map2, when the cloud service preference analysis is carried out on the basis of the requirement knowledge phrase relation network Map2, the relation among different requirements can be considered, the user preference is deeply mined, and the richness and accuracy of the obtained user preference information are guaranteed.

In an exemplary embodiment, the dimension of the relationship network of the requirement knowledge phrase relationship network Map1 in the X type requirement knowledge phrase relationship network Map1 obtained in Step11 may be reduced as required, for example: the relation network dimension of the 1 st type requirement knowledge phrase relation network Map1 is W1, the relation network dimension of the 2 nd type requirement knowledge phrase relation network Map1 is W2, the relation network dimension of the 3 rd type requirement knowledge phrase relation network Map1 is a set value W4W 3, and the like. The dimension value of the actual relation network of the X-class requirement knowledge phrase relation network Map1 is not limited in the embodiment of the application.

In one exemplary embodiment, for the g-th type requirement knowledge phrase relationship net Map1 in the X-type requirement knowledge phrase relationship net Map1, (g is Z and 1< g < X), the requirement knowledge phrase relationship net set corresponding to the g-th type requirement knowledge phrase relationship net Map1 comprises the g-1-th type requirement knowledge phrase relationship net Map1, the g-th type requirement knowledge phrase relationship net Map1 and the g + 1-th type requirement knowledge phrase relationship net Map1. Based on this, the technical solution recorded in Step12 may include the following steps 121 to 124.

And Step121, performing relation network dimension compression on the g-1 th class requirement knowledge phrase relation network Map1 to obtain a first g-th class requirement knowledge phrase relation network Map3.

In the embodiment of the application, the dimension compression of the relation network on the g-1-th class requirement knowledge phrase relation network Map1 can be understood as that the dimension reduction of the relation network on the g-1-th class requirement knowledge phrase relation network Map1 is carried out to obtain the first g-th class third requirement knowledge characteristic information (namely, requirement knowledge phrase relation network Map 3)

And Step122, carrying out relation network dimension adjustment on the g-th class requirement knowledge phrase relation network Map1 to obtain a second g-th class requirement knowledge phrase relation network Map3.

In the embodiment of the present application, the relationship network dimension adjustment performed on the g-th requirement knowledge phrase relationship network Map1 may be, for example, performing relationship network scale-invariant transformation on the g-th requirement knowledge phrase relationship network Map1.

And Step123, carrying out relation network dimension derivation on the g +1 th class requirement knowledge phrase relation network Map1 to obtain a third g class requirement knowledge phrase relation network Map3.

In the embodiment of the application, the derivation of the relation network dimension on the g +1 th class requirement knowledge phrase relation network Map1 may be, for example, the enlargement of the relation network dimension on the g +1 th class requirement knowledge phrase relation network Map1.

Step124, the first class-g requirement knowledge phrase relationship network Map3, the second class-g requirement knowledge phrase relationship network Map3 and the third class-g requirement knowledge phrase relationship network Map3 are sorted to obtain a class-g requirement knowledge phrase relationship network Map2.

The relationship network dimensions of the first g-th class requirement knowledge phrase relationship network Map3, the second g-th class requirement knowledge phrase relationship network Map3 and the third g-th class requirement knowledge phrase relationship network Map3 are consistent.

For example, for a requirement knowledge phrase relationship network set corresponding to the g-th class requirement knowledge phrase relationship network Map1, the g-1-th class requirement knowledge phrase relationship network Map1 with a larger relationship network dimension may be compressed to be consistent with the relationship network dimension of the g-th class requirement knowledge phrase relationship network Map1; deriving the g + 1-th requirement knowledge phrase relationship network Map1 with smaller relationship network dimension to be consistent with the relationship network dimension of the g-th requirement knowledge phrase relationship network Map1, so as to conveniently unify the relationship network dimension of each requirement knowledge phrase relationship network in the requirement knowledge phrase relationship network set.

It can be understood that the first g-1 type requirement knowledge phrase relationship network Map3 can be obtained by performing relationship network dimension compression on the g-1 type requirement knowledge phrase relationship network Map1; carrying out relation network dimension adjustment on the g-th class requirement knowledge phrase relation network Map1 to obtain a second g-th class requirement knowledge phrase relation network Map3; and carrying out relation network dimension derivation on the g +1 th class requirement knowledge phrase relation network Map1 to obtain a third g class requirement knowledge phrase relation network Map3. The relationship network dimensions of the first, second and third g-th class requirement knowledge phrase relationship network Map3 are consistent.

In an illustrative embodiment, the relationship network dimension compression can be realized through ideas such as feature filtering operation, feature reduction operation and the like; the method comprises the steps of realizing relation network dimension derivation through ideas such as reverse characteristic filtering operation (deconvolution), characteristic derivation (upsampling), characteristic filtering with a sliding variable (step length) smaller than 1 and the like; the dimension adjustment of the relationship network is realized through characteristic filtering or other processing ideas with the sliding variable as a set value, and the dimension adjustment is not further limited in the embodiment of the application.

In an exemplary embodiment, the first, second and third g-th class requirement knowledge phrase relationship networks Map3 may be immediately weighted or weighted according to a set index, and the g-th class requirement knowledge phrase relationship network Map2 is obtained by sorting, where the relationship network dimension of the g-th class requirement knowledge phrase relationship network Map2 is the same as the relationship network dimension of the g-th class requirement knowledge phrase relationship network Map1. Therefore, the arrangement of the relation network of the demand knowledge phrases with the relation can be realized, and the identification precision of the demand knowledge phrases can be obviously improved.

For an independently implementable technical solution, the performing the relationship network dimension compression on the g-1 th class requirement knowledge phrase relationship network Map1 to obtain the first g-th class requirement knowledge phrase relationship network Map3 may include the following contents: and processing the g-1 type requirement knowledge phrase relationship network Map1 through a first characteristic filtering unit to determine a first g type requirement knowledge phrase relationship network Map3.

The feature filtering node size of the first feature filtering unit is S x S, the sliding variable is S, S and S >1 and S belongs to Z, and the relation network dimension of the g-1 type requirement knowledge phrase relation network Map1 is S times of the relation network dimension of the g type requirement knowledge phrase relation network Map1.

For an independently implementable technical solution, the performing relationship network dimension adjustment on the g-th class requirement knowledge phrase relationship network Map1 to obtain a second g-th class requirement knowledge phrase relationship network Map3 may include the following contents: and processing the g-th class requirement knowledge phrase relationship network Map1 through a second characteristic filtering unit to determine a second g-th class requirement knowledge phrase relationship network Map3.

And the size of the characteristic filtering node of the second characteristic filtering unit is S x S, and the sliding variable is a set value.

For an independently implementable technical solution, performing relationship network dimension derivation on the g + 1-th class requirement knowledge phrase relationship network Map1 to obtain a third g-th class requirement knowledge phrase relationship network Map3 may include the following contents: and processing and characteristic derivation are performed on the g + 1-th class requirement knowledge phrase relationship network Map1 through a third characteristic filtering unit and a characteristic derivation unit, and the third g-th class requirement knowledge phrase relationship network Map3 is determined, wherein the size of a characteristic filtering node (convolution kernel) of the third characteristic filtering unit is S, and a sliding variable is a set value.

For example, processing of each requirement knowledge phrase relationship network in the requirement knowledge phrase relationship network set corresponding to the g-th requirement knowledge phrase relationship network Map1 may be implemented based on configuring different feature filtering units.

In an exemplary embodiment, the first g-1 type requirement knowledge phrase relationship network Map3 may be obtained by processing the g-1 type requirement knowledge phrase relationship network Map1 through the first feature filtering unit. The feature filtering node size of the first feature filtering unit is S × S, the sliding variable is S, S >1 and S ∈ Z, the relation network dimension of the g-1-th class requirement knowledge phrase relation network Map1 is S times of the relation network dimension of the g-th class requirement knowledge phrase relation network Map1, for example, the relation network dimension compression is realized through the feature filtering unit. Such as: and the relation network dimension of the g-1 type requirement knowledge phrase relation network Map1 is W2, the relation network dimension of the g-1 type requirement knowledge phrase relation network Map1 is a set value W4W 3, and s =2, namely the first relation network dimension constraint value and the second relation network dimension constraint value of the g-1 type requirement knowledge phrase relation network Map1 are both 2 times of the first relation network dimension constraint value and the second relation network dimension constraint value of the g-1 type requirement knowledge phrase relation network Map1. After the characteristic filtering processing, the relation network dimension of the first g-th class requirement knowledge phrase relation network Map3 is a set value W4W 3. Wherein S may be, but is not limited to, 3.

In an exemplary embodiment, the g-th class requirement knowledge phrase relationship network Map1 may be processed by a second feature filtering unit to obtain a second g-th class requirement knowledge phrase relationship network Map3, where the size of a feature filtering node (which may be understood as the size of a convolution kernel) of the second feature filtering unit is S × S, and a sliding variable is a set value, it can be understood that the relationship network dimension adjustment is realized through feature filtering. Such as: the relation network dimension of the g-th class requirement knowledge phrase relation network Map1 is a set value W4W 3, and after feature filtering processing is carried out, the relation network dimension of the second g-th class requirement knowledge phrase relation network Map3 is a set value W4W 3.

In an exemplary embodiment, a third g + 1-th class requirement knowledge phrase relationship network Map1 may be processed and subjected to S-fold feature derivation by a third feature filtering unit and a feature derivation unit to obtain a third g-th class requirement knowledge phrase relationship network Map3, where a feature filtering node of the third feature filtering unit has a size S × S and a sliding variable is a set value, for example, the feature filtering and the feature derivation are used to implement the relationship network dimension derivation. Such as: and the relation network dimension of the g + 1-th class requirement knowledge phrase relation network Map1 is W5, and the relation network dimension of the g-th class requirement knowledge phrase relation network Map1 is a set value W4W 3, so that s =2. And after feature filtering and 2-time feature derivation, the relationship network dimension of the third g-class requirement knowledge phrase relationship network Map3 is a set value W4 x W3. Other ways to implement relational network dimension derivation can also be employed, such as: an inverse eigen filter operation or an eigen filter with a sliding variable of 1/s, etc., which is not further limited in the embodiments of the present application.

Therefore, the relation network dimensionality of the required knowledge phrase relation network can be unified and centralized, and later sorting processing is facilitated.

In an exemplary embodiment, the first, second and third g-th class requirement knowledge phrase relationship nets Map3 may be weighted to obtain g-th class requirement knowledge phrase relationship nets Map2. The whole processing process can be realized through pyramid convolution, so that the required knowledge phrase relation network Map2 can be completely and comprehensively obtained, and the accuracy and the reliability of subsequent cloud service preference analysis can be effectively improved.

For an independently implementable technical solution, for the class 1 requirement knowledge phrase relationship net Map1 in the class X requirement knowledge phrase relationship net Map1, the requirement knowledge phrase relationship net set corresponding to the class 1 requirement knowledge phrase relationship net Map1 includes the class 1 requirement knowledge phrase relationship net Map1 and the class 2 requirement knowledge phrase relationship net Map1. In view of this, step12 may further include Step125-Step127 as follows.

And Step125, carrying out relation network dimension adjustment on the class 1 requirement knowledge phrase relation network Map1 to obtain a first class 1 requirement knowledge phrase relation network Map3.

And Step126, carrying out relation network dimension derivation on the class 2 requirement knowledge phrase relation network Map1 to obtain a second class 1 requirement knowledge phrase relation network Map3.

Step127, sorting the first class 1 requirement knowledge phrase relationship network Map3 and the second class 1 requirement knowledge phrase relationship network Map3 to obtain a class 1 requirement knowledge phrase relationship network Map2.

The relationship network dimensions of the first class 1 requirement knowledge phrase relationship network Map3 and the second class 1 requirement knowledge phrase relationship network Map3 are the same.

For example, for the type 1 requirement knowledge phrase relationship network Map1, which has no previous type requirement knowledge phrase relationship network, only the type 1 requirement knowledge phrase relationship network Map1 itself and the type 2 requirement knowledge phrase relationship network Map1 with contact may be processed.

In an illustrative embodiment, the relation network dimension adjustment can be performed on the class 1 requirement knowledge phrase relation network Map1 to obtain a first class 1 requirement knowledge phrase relation network Map3; and carrying out relation network dimension derivation on the class 2 requirement knowledge phrase relation network Map1 to obtain a second class g requirement knowledge phrase relation network Map3. The dimension of the relationship network of the first and second class 1 requirement knowledge phrase relationship networks Map3 is the same.

In an exemplary embodiment, the first and second class 1 requirement knowledge phrase relationship networks Map3 may be weighted to obtain a class 1 requirement knowledge phrase relationship network Map2. In this way, the sorting of the requirement knowledge phrase relationship network of the existence association of the type 1 can be realized.

For an independently implementable technical solution, performing relationship network dimension adjustment on the class 1 requirement knowledge phrase relationship network Map1 in Step125 to obtain a first class 1 requirement knowledge phrase relationship network Map3, may include the following contents: and processing the class-1 requirement knowledge phrase relationship network Map1 through a second feature filtering unit to determine the first class-1 requirement knowledge phrase relationship network Map3, wherein the size of a feature filtering node of the second feature filtering unit is S & ltS & gt, a sliding variable is a set value, S & gt 1 and S & ltE & gt is Z.

For an independently implementable technical solution, performing relationship network dimension derivation on the class 2 requirement knowledge phrase relationship network Map1 in Step126 to obtain a second class 1 requirement knowledge phrase relationship network Map3 may include the following contents: and processing and characteristic derivation are carried out on the class 2 requirement knowledge phrase relationship network Map1 through a third characteristic filtering unit and a characteristic derivation unit to obtain a second class 1 requirement knowledge phrase relationship network Map3, wherein the size of a characteristic filtering node of the third characteristic filtering unit is S x S, and a sliding variable is a set value.

In the embodiment of the application, processing of each requirement knowledge phrase relationship network in a requirement knowledge phrase relationship network set corresponding to the requirement knowledge phrase relationship network Map1 of the 1 st class can be realized based on configuring different feature filtering units. The class-1 requirement knowledge phrase relationship network Map1 can be processed through the second feature filtering unit to obtain a first class-1 requirement knowledge phrase relationship network Map3, for example, the relationship network dimension adjustment is realized through feature filtering; the second class-1 requirement knowledge phrase relationship network Map3 can be obtained by processing the class-2 requirement knowledge phrase relationship network Map1 and performing s-fold feature derivation through the third feature filtering unit and the feature derivation unit, for example, the relationship network dimension derivation is realized through feature filtering processing and feature derivation. Exemplary processing concepts are similar to those described above and embodiments of the present application are not described here in greater detail.

Therefore, the relation network dimensionality of the required knowledge phrase relation network can be unified, and later arrangement is facilitated.

For an independently implementable technical solution, for the class X demand knowledge phrase relationship network Map1 in the class X demand knowledge phrase relationship network Map1, the demand knowledge phrase relationship network set corresponding to the class X demand knowledge phrase relationship network Map1 includes a class X-1 demand knowledge phrase relationship network Map1 and the class X demand knowledge phrase relationship network Map1.

In view of this, step12 may also include the following: carrying out relation network dimension compression on the X-1 type requirement knowledge phrase relation network Map1 to obtain a first X type requirement knowledge phrase relation network Map3; carrying out relation network dimension adjustment on the X-th class requirement knowledge phrase relation network Map1 to obtain a second X-th class requirement knowledge phrase relation network Map3; and sorting the first and second class-X requirement knowledge phrase relationship networks Map3 and Map3 to obtain a class-X requirement knowledge phrase relationship network Map2, wherein the first and second class-X requirement knowledge phrase relationship networks Map3 and Map3 have the same relationship network dimension.

In the embodiment of the application, for the X-th type requirement knowledge phrase relationship network Map1, there is no latter type requirement knowledge phrase relationship network, and only the X-th type requirement knowledge phrase relationship network Map1 itself and the X-1-th type requirement knowledge phrase relationship network Map1 having a relationship may be processed.

In an illustrative embodiment, the relation network dimension compression can be carried out on the X-1 type requirement knowledge phrase relation network Map1 to obtain a first X type requirement knowledge phrase relation network Map3; the relation network dimension adjustment can be carried out on the Xth-class requirement knowledge phrase relation network Map1 to obtain a second Xth-class requirement knowledge phrase relation network Map3. Wherein the first and second class X requirement knowledge phrase relationship networks Map3 have the same relationship network dimension.

In an exemplary embodiment, the first and second class X requirement knowledge phrase relationship networks Map3 may be weighted to obtain a class X requirement knowledge phrase relationship network Map2. Therefore, the arrangement of the requirement knowledge phrase relation network of the existence contact of the X-th class can be realized.

For an independently implementable technical solution, the performing the relationship network dimension compression on the X-1 th class requirement knowledge phrase relationship network Map1 to obtain the first X-1 th class requirement knowledge phrase relationship network Map3 may include the following contents: processing the X-1 type requirement knowledge phrase relationship net Map1 through a first feature filtering unit to determine a first X type requirement knowledge phrase relationship net Map3, wherein the feature filtering node size of the first feature filtering unit is S, the sliding variable is S, S and S >1 and S belongs to Z, and the relationship net dimension of the g-1 type requirement knowledge phrase relationship net Map1 is S times of the relationship net dimension of the g type requirement knowledge phrase relationship net Map1.

For an independently implementable technical solution, the performing relationship network dimension adjustment on the xth requirement knowledge phrase relationship network Map1 to obtain a second xth requirement knowledge phrase relationship network Map3 may include the following contents: and processing the class X requirement knowledge phrase relationship network Map1 through a second feature filtering unit to determine a second class X requirement knowledge phrase relationship network Map3, wherein the size of a feature filtering node of the second feature filtering unit is S, and a sliding variable is a set value.

In the embodiment of the application, processing of each requirement knowledge phrase relationship network in a requirement knowledge phrase relationship network set corresponding to the X-th requirement knowledge phrase relationship network Map1 can be realized based on configuring different feature filtering units. The first characteristic filtering unit can be used for processing the X-1 type requirement knowledge phrase relationship network Map1 to obtain a first X type requirement knowledge phrase relationship network Map3, in other words, the relationship network dimension compression is realized through characteristic filtering; the second feature filtering unit processes the xth-type requirement knowledge phrase relationship network Map1 to obtain a second xth-type requirement knowledge phrase relationship network Map3, in other words, the feature filtering unit realizes the relationship network dimension adjustment. Exemplary processing is similar to that described above, and the embodiments of the present application are not described herein in excess. Therefore, the relation network dimensionality of the requirement knowledge phrase relation network can be unified, and later arrangement is facilitated.

In an exemplary embodiment, the second and third eigen filter units comprise adjustable eigenfilter units or expansion eigenfilter units.

It can be understood that, for the requirement knowledge phrase relationship network set corresponding to the g-th class requirement knowledge phrase relationship network Map1, the first feature filtering unit corresponding to the g-1-th class requirement knowledge phrase relationship network Map1 is a conventional feature filtering unit; the second characteristic filtering unit corresponding to the g-th requirement knowledge phrase relation network Map1 and the third characteristic filtering unit corresponding to the g + 1-th requirement knowledge phrase relation network Map1 are adjustable characteristic filtering units or expansion characteristic filtering units.

In an exemplary embodiment, if the second and third eigen filter units are adjustable eigen filter units, a new eigen filter unit may be preset to analyze errors, and then the raw material relationship network and the errors are used as the input of the adjustable eigen filter unit, and are processed through some series of adjustment processes, and then processed.

In an exemplary embodiment, the second eigen filter unit and the third eigen filter unit are expansion eigen filter units, and the filter range may be adjusted in advance. Therefore, the filtering range can be adaptively changed, and the reliability of the requirement knowledge phrase relation network arrangement is further improved.

In an exemplary embodiment, the cloud computing service information processing method based on the smart city according to the embodiment of the present application may be implemented by a GCN algorithm, and the GCN algorithm may include a relational network classification model for performing multi-category user service requirement identification on a selected smart service session record.

In an exemplary embodiment, the GCN algorithm may include cascaded Q-class sorting model layers (which may be understood as a fusion module) for performing Q-time relationship network dimension change and sorting on the X-class requirement knowledge phrase relationship network Map1, where each level of sorting model layer includes a plurality of first feature filtering units, a plurality of second feature filtering units, and a plurality of third feature filtering units, Q >0 and Q ∈ Z.

In an exemplary embodiment, the relationship network dimension changing and sorting process can be performed in several rounds, the process can be realized through Q-type sorting model layers, each type of sorting model layer comprises a plurality of first feature filtering units, a plurality of second feature filtering units and a plurality of third feature filtering units, and the first feature filtering units, the second feature filtering units and the third feature filtering units are respectively used for processing each requirement knowledge phrase relationship network set formed by requirement knowledge phrase relationship networks with relations. Q takes on the value of 4, for example.

In an exemplary embodiment, each sort of sorting model layer may process a plurality of requirement knowledge phrase relationship network sets, and each requirement knowledge phrase relationship network set corresponds to one group of feature filtering units, and is used to process each requirement knowledge phrase relationship network in the requirement knowledge phrase relationship network set. Such as: for a requirement knowledge phrase relation network set comprising a g-1 type requirement knowledge phrase relation network Map1, a g type requirement knowledge phrase relation network Map1 and a g +1 type requirement knowledge phrase relation network Map1, a group of feature filtering units corresponding to the requirement knowledge phrase relation network set comprises a first feature filtering unit, a second feature filtering unit, a third feature filtering unit and a feature derivation unit, and the feature filtering units are used for respectively processing the g-1 type requirement knowledge phrase relation network Map1, the g type requirement knowledge phrase relation network Map1 and the g +1 type requirement knowledge phrase relation network Map1.

For an independently implementable technical solution, step12 may further include the following: transmitting the X-type requirement knowledge phrase relationship network Map1 into a 1 st type sorting model layer to obtain a 1 st sorted X-type requirement knowledge phrase relationship network Map4; transmitting the X-type requirement knowledge phrase relation network Map4 sorted in the d-1 round into a d-type sorting model layer to obtain the X-type requirement knowledge phrase relation network Map4 sorted in the d-th round, wherein d is an integer and is more than 1 and less than d and less than Q; and transmitting the X-type requirement knowledge phrase relationship network Map4 sorted in the Q-1 round into a Q-type sorting model layer, and outputting the X-type requirement knowledge phrase relationship network Map2.

In the embodiment of the application, the X-type requirement knowledge phrase relationship network Map1 can be transmitted into the 1 st type arrangement model layer, and the 1 st time of relationship network dimension change and arrangement is carried out to obtain the 1 st time of arranged X-type requirement knowledge phrase relationship network Map4; and inputting the X-type requirement knowledge phrase relation network Map4 sorted for the 1 st time into a next type sorting model layer. The X-type requirement knowledge phrase relation network Map4 sorted in the d-1 round can be transmitted into a d-type sorting model layer to carry out dimension change and sorting on the d-type relation network, and the X-type requirement knowledge phrase relation network Map4 sorted in the d-type round is obtained, wherein d is an integer and is more than 1 and less than d and less than Q, so that the completeness of sorting fusion can be further improved.

For an independently implementable technical scheme, each sort of sorting model layer further comprises a normalization layer which is used for carrying out standardization processing on the demand knowledge phrase relationship network after the round of sorting. The step of transferring the d-1 th round of sorted X-type requirement knowledge phrase relationship network Map4 into the d-type sorting model layer to obtain the d-th round of sorted X-type requirement knowledge phrase relationship network Map4 may include the following steps: respectively carrying out relation network dimension change and arrangement on a requirement knowledge phrase relation network set corresponding to the D-1 th round of arranged X-type requirement knowledge phrase relation network Map4 through a first characteristic filtering unit, a second characteristic filtering unit and a third characteristic filtering unit of the d-type arrangement model layer to obtain a D-1 th round of arranged X-type to-be-processed knowledge phrase relation network; and standardizing the X-class knowledge phrase relationship networks to be processed in the d-1 round of arrangement through the normaize layer, and determining the X-class requirement knowledge phrase relationship network Map4 in the d-1 round of arrangement.

For example, for the d-th round of relationship network dimension change and arrangement, the relationship network dimension change and arrangement can be respectively performed on the requirement knowledge phrase relationship network set corresponding to the X-class requirement knowledge phrase relationship network Map4 in the d-1 round of arrangement through the first feature filtering unit, the second feature filtering unit and the third feature filtering unit of the d-th class arrangement model layer, so as to obtain the X-class to-be-processed knowledge phrase relationship network in the d-1 round of arrangement.

In an exemplary embodiment, the d-th sorting model layer may process a plurality of requirement knowledge phrase relationship network sets corresponding to the X-th requirement knowledge phrase relationship network Map4 sorted in the d-1 th round, where each requirement knowledge phrase relationship network set corresponds to one group of feature filtering units and is used to process each requirement knowledge phrase relationship network in the requirement knowledge phrase relationship network set. Such as: for a requirement knowledge phrase relationship network set comprising a g-1 type requirement knowledge phrase relationship network Map1, a g type requirement knowledge phrase relationship network Map1 and a g +1 type requirement knowledge phrase relationship network Map1, a group of feature filtering units corresponding to the requirement knowledge phrase relationship network set comprises a first feature filtering unit, a second feature filtering unit, a third feature filtering unit and a feature derivation unit, and are used for respectively processing the g-1 type requirement knowledge phrase relationship network Map1, the g type requirement knowledge phrase relationship network Map1 and the g +1 type requirement knowledge phrase relationship network Map1.

In an illustrative embodiment, the accumulated quantity (such as an average operation result and a fluctuation evaluation result) of the class X knowledge phrase relationship network to be processed in the d-1 round of arrangement is counted through a normaize layer, the class X knowledge phrase relationship network to be processed in the d-1 round of arrangement is normalized, and the result of the normalization processing is determined as the class X requirement knowledge phrase relationship network Map4 in the d round of arrangement.

In an exemplary embodiment, the GCN algorithm may further include a linear capture algorithm and a naive bayes algorithm, respectively, for implementing a capture task and a clustering task in the cloud service preference analysis. The linear capture algorithm and the naive Bayes algorithm can comprise a feature filtering unit, a triggering unit, a classifying unit and the like, and the specific model architecture of the linear capture algorithm and the naive Bayes algorithm is not limited in the application.

For a solution that can be implemented independently, step13 can include the following steps 131 and 132.

Step131, transmitting the X-class requirement knowledge phrase relation net Map2 into the linear capture algorithm, and determining session contents corresponding to user preferences in the selected intelligent service session records.

Step132, transmitting the X-class requirement knowledge phrase relationship net Map2 into the naive Bayesian algorithm, and determining a distinguishing label of the user preference in the selected intelligent service session record, wherein the user preference information comprises session content corresponding to the user preference and the distinguishing label of the user preference.

For example, the capture task and the clustering task in the cloud service preference analysis can be realized according to the X-class requirement knowledge phrase relation network Map2. The X-type requirement knowledge phrase relation network Map2 can be input into a linear capture algorithm for processing, and session contents corresponding to user preferences in the selected intelligent service session records are captured and obtained; the class X requirement knowledge phrase relationship network Map2 can be input into a naive Bayes algorithm for processing, and the distinguishing label of the user preference in the selected intelligent service session record is determined. The user preference information of the selected intelligent service session record can comprise session content corresponding to the user preference and a distinguishing label of the user preference.

Thus, according to the GCN algorithm of the embodiment of the application, the processing amount can be greatly reduced, and meanwhile, the algorithm processing efficiency can be improved.

In an exemplary embodiment, the GCN algorithm may be configured prior to applying the GCN algorithm according to embodiments of the present application. In other words, the reference service session records in the configuration information set are imported into the GCN algorithm, and the reference user preference information of the reference service session records is obtained through the processing of a relational network classification model, a Q-type sorting model layer, a linear capture algorithm and a naive Bayesian algorithm; determining algorithm evaluation information according to the comparison result of the reference user preference information and the annotation information recorded by the plurality of reference service sessions; changing the variable of the GCN algorithm according to the algorithm evaluation information; and obtaining the GCN algorithm for completing the configuration on the premise of meeting the configuration index (such as meeting the user requirement). The exemplary configuration flow is not limited in this application.

In some independent design considerations, after determining the user preference information of the selected wisdom service session record, the method may further include: determining a smart city product expectation of a target user based on the user preference information; and pushing big data to the smart city client of the target user through the smart city product expectation.

For example, the user preference information may be business preferences related to cross-border e-commerce, such as various physical goods consultation, e-commerce platform security concerns, and the like. Based on the method, user preference information can be further mined and analyzed, so that the wisdom city product expectation of the target user can be obtained, the wisdom city product expectation can visually reflect various service requirements of the target user, and targeted big data pushing can be performed based on the wisdom city product expectation, such as entity commodity e-commerce link pushing, safety protection software pushing and the like. In addition, the smart city client is not limited to a mobile phone, a PC, and the like.

Under some independent design ideas, the smart city product expectation of the target user is determined based on the user preference information, and the method can be realized by the following technical scheme: obtaining a first interest preference vector and a second interest preference vector corresponding to the user preference information, wherein the first interest preference vector comprises a preference vector which does not contain individual interests in the user preference information, and the second interest preference vector comprises a preference vector which contains the individual interests in the user preference information; performing convolution operation on the first interest preference vector to obtain a group interest description field corresponding to the first interest preference vector; performing convolution operation on the second interest preference vector to obtain an individual interest description field corresponding to the second interest preference vector; weighting the individual interest description fields and the group interest description fields to obtain global interest preference corresponding to the user preference information; performing product expectation matching on the global interest preference to obtain a matching index corresponding to the user preference information; and determining the product expectation of the smart city of the user preference information according to the numerical value interval in which the matching index falls.

For example, the numerical range in which the matching index falls corresponds to the quantization range of "entity commodity e-commerce link push", it can be determined that the smart city product is expected to be "entity commodity desired to be purchased". Based on this, different ranges of interests may be considered, thereby guaranteeing the integrity of global interest preferences.

Under some design ideas which can be independent, the obtaining of the first interest preference vector and the second interest preference vector corresponding to the user preference information includes: performing interest prediction on the user preference information to obtain a first preference vector which does not contain individual interests in the user preference information, and performing disassembly processing on the first preference vector in the user preference information to be used as the first interest preference vector; and acquiring a second feature vector containing individual interests in the user preference information according to the first preference vector, and performing dismantling processing on the second feature vector in the user preference information to serve as the second interest preference vector.

Fig. 3 is a schematic architecture diagram illustrating an application environment in which the smart city based cloud computing service information processing method according to the embodiment of the present application may be implemented, where the smart city based cloud computing service information processing method may include a cloud computing service processing system 100 and a service interaction terminal 200 that communicate with each other. Based on this, the cloud computing service processing system 100 and the service interaction terminal 200 implement or partially implement the cloud computing service information processing method based on the smart city according to the embodiment of the present application during operation.

The embodiments of the present application have been described above with reference to the accompanying drawings, and have at least the following beneficial effects: the method comprises the steps that in response to a service demand identification instruction, X-type user service demand identification is carried out on selected intelligent service session records to obtain X-type demand knowledge phrase relationship networks Map1, and then each demand knowledge phrase relationship network Map1 and the demand knowledge phrase relationship network in contact with the demand knowledge phrase relationship network Map1 are sorted to obtain X-type demand knowledge phrase relationship networks Map2; the X-type requirement knowledge phrase relation network Map2 is subjected to cloud service preference analysis to obtain user preference information, then multiple types of requirement knowledge phrases of the X-type requirement knowledge phrase relation network Map1 can be sorted and combined to obtain the comprehensive and accurate X-type requirement knowledge phrase relation network Map2, when cloud service preference analysis is carried out on the basis of the requirement knowledge phrase relation network Map2, the relation among different requirements can be considered, so that user preference is deeply mined, and the richness and accuracy of the obtained user preference information are guaranteed.

The above description is only a preferred embodiment of the present application, and is not intended to limit the scope of the present application.

Claims

1. A cloud computing service information processing method based on a smart city is applied to a cloud computing service processing system, and the method comprises the following steps:

responding to a service requirement identification instruction, carrying out X-type user service requirement identification on the selected intelligent service session record, and determining an X-type requirement knowledge phrase relation network Map1 of the selected intelligent service session record; the relation network dimensionality of each type of demand knowledge phrase relation network Map1 in the X type of demand knowledge phrase relation network Map1 is different, X is greater than 1 and belongs to Z;

respectively carrying out relation network dimension change and arrangement on the demand knowledge phrase relation network sets corresponding to the various demand knowledge phrase relation networks Map1 to obtain an X-type demand knowledge phrase relation network Map2; each demand knowledge phrase relationship network set comprises a demand knowledge phrase relationship network Map1 and a demand knowledge phrase relationship network Map1' which is in contact with the demand knowledge phrase relationship network Map1;

2. The method of claim 1, wherein the requirement knowledge phrase relationship network set corresponding to the g-th class requirement knowledge phrase relationship network Map1 comprises a g-1-th class requirement knowledge phrase relationship network Map1, a g-th class requirement knowledge phrase relationship network Map1 and a g + 1-th class requirement knowledge phrase relationship network Map1, 1-t g-t and g ∈ X; the method for respectively changing and sorting the relation network dimensions of the demand knowledge phrase relation network sets corresponding to the various demand knowledge phrase relation networks Map1 to obtain the X-type demand knowledge phrase relation network Map2 comprises the following steps:

carrying out relation network dimension derivation on the g + 1-th class requirement knowledge phrase relation network Map1 to obtain a third g-th class requirement knowledge phrase relation network Map3;

sorting the first g-th class requirement knowledge phrase relationship network Map3, the second g-th class requirement knowledge phrase relationship network Map3 and the third g-th class requirement knowledge phrase relationship network Map3 to obtain a g-th class requirement knowledge phrase relationship network Map2; the dimensionality of the relationship network of the first g-th class requirement knowledge phrase relationship network Map3, the second g-th class requirement knowledge phrase relationship network Map3 and the third g-th class requirement knowledge phrase relationship network Map3 is the same.

3. The method as claimed in claim 1, wherein the requirement knowledge phrase relationship network set corresponding to the class 1 requirement knowledge phrase relationship network Map1 includes the class 1 requirement knowledge phrase relationship network Map1 and the class 2 requirement knowledge phrase relationship network Map1, and the step of performing relationship network dimension change and sorting on the requirement knowledge phrase relationship network set corresponding to the various classes of requirement knowledge phrase relationship networks Map1 to obtain the class X requirement knowledge phrase relationship network Map2 includes:

carrying out relation network dimension adjustment on the 1 st type requirement knowledge phrase relation network Map1 to obtain a first 1 st type requirement knowledge phrase relation network Map3;

4. The method as claimed in claim 2, wherein the requirement knowledge phrase relationship network set corresponding to the X-th class requirement knowledge phrase relationship network Map1 includes an X-1-th class requirement knowledge phrase relationship network Map1 and the X-th class requirement knowledge phrase relationship network Map1, and the modifying and sorting of the relationship network dimension of the requirement knowledge phrase relationship network set corresponding to the various classes of requirement knowledge phrase relationship networks Map1 to obtain an X-class requirement knowledge phrase relationship network Map2 includes:

sorting the first class X requirement knowledge phrase relationship network Map3 and the second class X requirement knowledge phrase relationship network Map3 to obtain a class X requirement knowledge phrase relationship network Map2; the first class-X requirement knowledge phrase relationship network Map3 and the second class-X requirement knowledge phrase relationship network Map3 have the same relationship network dimension.

5. The method as claimed in claim 2, wherein the performing the relationship network dimension compression on the g-1 th class requirement knowledge phrase relationship network Map1 to obtain a first g-th class requirement knowledge phrase relationship network Map3 comprises: processing the g-1 type requirement knowledge phrase relationship net Map1 through a first feature filtering unit to determine a first g type requirement knowledge phrase relationship net Map3, wherein the feature filtering node size of the first feature filtering unit is S, the sliding variable is S, S and S are greater than 1 and belongs to Z, and the relationship net dimension of the g-1 type requirement knowledge phrase relationship net Map1 is S times of the relationship net dimension of the g type requirement knowledge phrase relationship net Map1;

performing relationship network dimension adjustment on the g-th class requirement knowledge phrase relationship network Map1 to obtain a second g-th class requirement knowledge phrase relationship network Map3, including: processing the g-th class requirement knowledge phrase relationship network Map1 through a second feature filtering unit to determine a second g-th class requirement knowledge phrase relationship network Map3, wherein the feature filtering node size of the second feature filtering unit is S x S, and the sliding variable is a set value;

performing relational network dimension derivation on the g +1 th class requirement knowledge phrase relational network Map1 to obtain a third g class requirement knowledge phrase relational network Map3, including: processing and characteristic derivation are carried out on the g + 1-th class requirement knowledge phrase relation network Map1 through a third characteristic filtering unit and a characteristic derivation unit, and a third g-th class requirement knowledge phrase relation network Map3 is determined, wherein the size of a characteristic filtering node of the third characteristic filtering unit is S x S, and a sliding variable is a set value;

wherein the second and third feature filter units comprise adjustable or expandable feature filter units.

6. The method of claim 3, wherein the performing a relationship network dimension adjustment on the class 1 requirement knowledge phrase relationship network Map1 to obtain a first class 1 requirement knowledge phrase relationship network Map3 comprises: processing the class 1 requirement knowledge phrase relationship network Map1 through a second feature filtering unit to determine a first class 1 requirement knowledge phrase relationship network Map3; the size of a characteristic filtering node of the second characteristic filtering unit is S, a sliding variable is a set value, S is greater than 1, and S belongs to Z;

7. The method as claimed in claim 4, wherein the performing the relationship network dimension compression on the X-1 th class requirement knowledge phrase relationship network Map1 to obtain a first X-th class requirement knowledge phrase relationship network Map3 comprises: processing the X-1 type requirement knowledge phrase relationship net Map1 through a first feature filtering unit to determine a first X type requirement knowledge phrase relationship net Map3, wherein the feature filtering node size of the first feature filtering unit is S, the sliding variable is S, S and S are greater than 1 and belongs to Z, and the relationship net dimension of the g-1 type requirement knowledge phrase relationship net Map1 is S times of the relationship net dimension of the g type requirement knowledge phrase relationship net Map1;

8. The method according to claim 5, wherein the method is implemented by a GCN algorithm, the GCN algorithm comprises cascaded Q-class sorting model layers for performing Q-time relation network dimension change and sorting on the X-class requirement knowledge phrase relation network Map1, each level of sorting model layer comprises a plurality of first feature filtering units, a plurality of second feature filtering units and a plurality of third feature filtering units, Q >0 and Q e Z; the step of respectively carrying out relationship network dimension change and arrangement on the requirement knowledge phrase relationship network sets corresponding to the various requirement knowledge phrase relationship networks Map1 to obtain an X-type requirement knowledge phrase relationship network Map2 comprises the following steps:

transmitting the D-1 th round of sorted X-type requirement knowledge phrase relationship network Map4 into a d-type sorting model layer to obtain the D-th round of sorted X-type requirement knowledge phrase relationship network Map4, wherein 1-d-Q and d ∈ Z;

wherein each level of the sorting model layer further comprises a normaize layer, and the step of transmitting the class-X requirement knowledge phrase relationship network Map4 sorted in the (d-1) th round into the class-d sorting model layer to obtain the class-X requirement knowledge phrase relationship network Map4 sorted in the (d) th round comprises the following steps:

and standardizing the X-class knowledge phrase relationship networks to be processed in the d-th round of arrangement through the normalize layer, and determining an X-class requirement knowledge phrase relationship network Map4 in the d-th round of arrangement.

9. The method of claim 1, wherein the method is implemented by a GCN algorithm, the GCN algorithm further comprising a linear capture algorithm and a naive bayes algorithm, the cloud service preference analysis of the class X requirement knowledge phrase relationship network Map2 to determine user preference information for the selected smart service session record, comprising:

10. A cloud computing service processing system, comprising:

a memory for storing an executable computer program, a processor for implementing the method of any one of claims 1-9 when executing the executable computer program stored in the memory.