Method, device, equipment and computer-readable storage medium for automatically arranging message application functions

Abstract

Method, device, equipment and computer-readable storage medium for automatic arrangement of message application functions. The embodiment of the present invention relates to the field of communication technology, and discloses a method for automatic arrangement of message application functions, including: obtaining target message application requirement parameters and function parameter sets; the function parameter set includes function parameters of each target function node; converting the target message application requirement parameters and function parameters into target requirement feature vectors and target function feature vectors; inputting the target requirement feature vectors and target function feature vectors into the message composite capability arrangement model to obtain the target arrangement relationship between the target function nodes corresponding to the target message application requirement parameters; the message composite capability arrangement model is trained based on historical message application requirement parameters, function parameters of historical function nodes and corresponding historical arrangement relationships between historical function nodes; and outputting the target arrangement relationship between each target function node corresponding to the target message application requirement parameters. In the above manner, the embodiment of the present invention realizes the automatic arrangement of the required function nodes.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of mobile communication technology, and in particular to a method for automatically arranging message application functions, an apparatus for automatically arranging message application functions, an automatic arrangement device for automatically arranging message application functions, and a computer-readable storage medium. Background Art

[0002] At present, with the development of technology, the development of application functions is becoming more and more intelligent, but there is still a lack of an intelligent development tool that can automatically form an application according to demand. For example, for 5G message applications, the diversity of functions and information of 5G message applications has increased compared to traditional SMS. However, at present, in the existing technology, the composite capability orchestration required by 5G message applications mainly relies on expert experience to manually develop and implement typical 5G message industry applications. However, due to the endless emergence of industry needs and the trend of becoming more and more diversified, the manual orchestration method is inefficient, time-consuming, labor-intensive, and prone to errors. Summary of the invention

[0003] In view of the above problems, the embodiments of the present invention provide a method for automatically arranging functions of a message application, an apparatus for automatically arranging functions of a message application, an equipment for automatically arranging functions of a message application, and a computer-readable storage medium, which are used to solve the technical problem in the prior art that message applications cannot automatically arrange functions according to needs.

[0004] According to one aspect of an embodiment of the present invention, a method for automatically arranging information application functions is provided, the method comprising:

[0005] Obtaining target message application requirement parameters and a function parameter set; the function parameter set includes function parameters of each target function node;

[0006] Convert the target message application requirement parameters and the function parameters of each target function node into a target requirement feature vector and a target function feature vector respectively;

[0007] The target requirement feature vector and the target function feature vector are input into the message composite capability orchestration model to obtain the target orchestration relationship between each target function node corresponding to the target message application requirement parameter; wherein the message composite capability orchestration model is trained based on the historical message application requirement parameters, the function parameters of the historical function nodes and the historical orchestration relationship between the corresponding historical function nodes;

[0008] Output the target orchestration relationship between each target function node corresponding to the target message application requirement parameter.

[0009] In an optional manner, the message composite capability orchestration model includes a network capability topology encoder, a message application requirement encoder, and a capability orchestration topology generator;

[0010] The network capability topology encoder is used to extract features from the target function feature vector to obtain a potential feature representation of the function;

[0011] The message application requirement encoder is used to extract features from the requirement feature vector to obtain a requirement potential feature representation;

[0012] The capability orchestration topology generator is used to determine the target orchestration relationship between target function nodes according to the function potential feature representation and the requirement potential feature representation.

[0013] In an optional manner, the network capability topology encoder includes a first relationship graph convolution layer, a first fully connected layer, a first random discard layer, a second relationship graph convolution layer, a second fully connected layer and a second random discard layer; the network capability topology encoder is used to extract features from the target function feature vector to obtain a functional potential feature representation;

[0014] The message application requirement encoder comprises a first word embedding layer, a first convolution layer, a first pooling layer, a second convolution layer, a second pooling layer, and a first flattening layer; the message application requirement encoder is used to extract features from the requirement feature vector to obtain a potential feature representation of the requirement;

[0015] The capability orchestration topology generator comprises a first merging layer and a tensor factor decomposition layer; the capability orchestration topology generator is used to determine a target orchestration relationship between target function nodes according to the function potential feature representation and the requirement potential feature representation.

[0016] In an optional manner, the message composite capability orchestration model is trained according to historical message application requirement parameters, function parameters of historical function nodes and historical orchestration relationships between corresponding historical function nodes, and further includes:

[0017] Obtain historical message application requirement parameters and historical function parameters of each function node;

[0018] Convert historical message application demand parameters into historical demand feature vectors;

[0019] Converting the function parameters of the historical function nodes into historical function feature vectors corresponding to the historical function nodes;

[0020] Marking the arrangement relationship of each historical function feature vector according to the arrangement relationship between each historical function node to obtain a historical feature vector set, wherein the historical feature vector set includes the historical function feature vectors and the historical arrangement relationship of each historical function node;

[0021] The historical demand feature vector and the historical feature vector set are input into a preset neural network model for training to obtain a message composite capability orchestration model.

[0022] In an optional manner, the target message application requirement parameters include a target message application requirement description text, and the function parameters of each target function node include a function description text of each target function node; converting the target message application requirement parameters and the function parameters of each target function node into a target requirement feature vector and a target function feature vector, respectively, includes:

[0023] Performing text cleaning on the message application requirement description text and the function description text of each function node to obtain a target requirement text and a target function text;

[0024] Serializing the target requirement text and the target function text to obtain a target requirement text sequence and a target function text sequence;

[0025] The target requirement text sequence and the target function text sequence are respectively converted into vector representation to obtain a target requirement feature vector and a target function feature vector.

[0026] In an optional manner, a target message application requirement parameter and a function parameter set are obtained; the function parameter set includes function parameters of each target function node, including:

[0027] A target message application requirement request is obtained, wherein the target message application requirement request carries the target message application requirement parameter.

[0028] According to another aspect of an embodiment of the present invention, there is provided a device for automatically arranging message application functions, comprising:

[0029] An acquisition module, used to acquire target message application requirement parameters and a function parameter set; the function parameter set includes function parameters of each target function node;

[0030] A conversion module, used to convert the target message application requirement parameters and the function parameters of each target function node into a target requirement feature vector and a target function feature vector respectively;

[0031] A prediction module is used to input the target demand feature vector and the target function feature vector into a message composite capability orchestration model to obtain a target orchestration relationship between target function nodes corresponding to the target message application demand parameter; wherein the message composite capability orchestration model is trained based on historical message application demand parameters, function parameters of historical function nodes, and historical orchestration relationships between corresponding historical function nodes;

[0032] The output module is used to output the target orchestration relationship between each target function node corresponding to the target message application requirement parameter.

[0033] In an optional manner, the message composite capability orchestration model includes a network capability topology encoder, a message application requirement encoder, and a capability orchestration topology generator;

[0034] The network capability topology encoder is used to extract features from the target function feature vector to obtain a potential feature representation of the function;

[0035] The message application requirement encoder is used to extract features from the requirement feature vector to obtain a requirement potential feature representation;

[0036] The capability orchestration topology generator is used to determine the target orchestration relationship between target function nodes according to the function potential feature representation and the requirement potential feature representation.

[0037] According to another aspect of an embodiment of the present invention, there is provided a device for automatically arranging message application functions, comprising:

[0038] A processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus;

[0039] The memory is used to store at least one executable instruction, and the executable instruction enables the processor to execute the operation of the above-mentioned message application function automatic arrangement method.

[0040] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is provided, wherein the storage medium stores at least one executable instruction. When the executable instruction is executed on a message application function automatic arrangement device, the message application function automatic arrangement device performs the operations of the above-mentioned message application function automatic arrangement method.

[0041] The embodiment of the present invention uses a neural network to predict the orchestration relationship of capability nodes corresponding to message application requirements, and can implement multi-scenario message applications according to message application requirements. Enterprises can quickly complete the deployment of message applications through the platform without complex code development, and can automatically, simply and conveniently create message applications.

[0042] In addition, by setting the specific structures of the network capability topology encoder, the message application requirement encoder and the capability orchestration topology generator, the neural network can make the prediction of the orchestration relationship more accurate.

[0043] The above description is only an overview of the technical solution of the embodiment of the present invention. In order to more clearly understand the technical means of the embodiment of the present invention, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the embodiment of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The accompanying drawings are only used to illustrate the embodiments and are not to be considered as limiting the present invention. In addition, the same reference symbols are used to represent the same components throughout the accompanying drawings. In the accompanying drawings:

[0045] Figure 1 A schematic diagram showing a flow chart of a method for automatically arranging message application functions provided by an embodiment of the present invention;

[0046] Figure 2 A schematic diagram showing the structure of a message composite capability arrangement model provided by an embodiment of the present invention is shown;

[0047] Figure 3 A flow chart of a training method for a message composite capability orchestration model provided by an embodiment of the present invention is shown;

[0048] Figure 4 A schematic diagram showing the structure of a device for automatically arranging message application functions provided by an embodiment of the present invention is shown;

[0049] Figure 5 A schematic diagram of the structure of a device for automatically arranging message application functions provided by an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0050] Exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0051] 5G message: It is an upgrade of SMS service and a basic telecommunications service of operators. It realizes a leap in service experience based on IP technology, supports more media formats and richer forms of expression. Compared with traditional SMS with single functions, 5G message not only broadens the breadth of information sending and receiving, supports users to use multimedia content such as text, audio and video, cards, and location, but also extends the depth of interactive experience. Users can complete service search, discovery, interaction, payment and other services in the message window, and build a one-stop service information window. 5G message has the characteristics of both 2C and 2B applications. It can not only facilitate the transmission of rich media information such as voice, pictures, videos, cards, and files between users, but also support enterprises to provide interactive services in the form of chatbots on the 5G message platform. 5G message is not only a comprehensive upgrade of traditional SMS, but also provides an entry for a series of rich 5G applications that are coming soon. People can use 5G messages to shop, order meals, book tickets, and book hotels, watch movies, hold video conferences, experience VR and online education. 5G message will become a bridge connecting people and society.

[0052] In the embodiment of the present invention, the functional node refers to the smallest granularity capability in the network capability, that is, the node that realizes the smallest function, such as location capability, call capability, big data capability, SMS capability, etc.

[0053] Figure 1 The flowchart of the method for automatically arranging message application functions provided by an embodiment of the present invention is shown. The method is executed by a device for automatically arranging message application functions. The device for automatically arranging message application functions can be a computer device or a terminal device. Figure 1 As shown, the method comprises the following steps:

[0054] Step 110: Obtain target message application requirement parameters and a function parameter set; the function parameter set includes function parameters of each target function node.

[0055] The embodiment of the present invention takes the 5G message application as an example, which is not limited to the 5G message application. Among them, industry users can initiate a new target message application demand request to the 5G message application open platform, and the target message application demand request carries the target message application demand parameters. Since a message application may have multiple functions, the target message application demand parameters include multiple sub-message application demand parameters, and each sub-message application demand parameter is implemented through at least one target function node orchestration. The requirement parameter set of the 5G message application includes the functions that the message application needs to implement. For example, the demand description of an educational message application is: to collect and analyze the daily health status reporting information of students and teachers, and to issue epidemic prevention and health guidance to students and teachers.

[0056] Step 120: Convert the target message application requirement parameters and the function parameters of each target function node into a target requirement feature vector and a target function feature vector respectively.

[0057] The target message application requirement parameters include target message application requirement description text, and the function parameters of each target function node include function description text of each target function node.

[0058] In the embodiment of the present invention, converting the target message application requirement parameters and the function parameters of each target function node into a target requirement feature vector and a target function feature vector respectively specifically includes the following steps:

[0059] The message application requirement description text and the function description text of each function node are cleaned to obtain a target requirement text and a target function text; the message application requirement description text includes multiple sub-requirement texts, and the function description text includes multiple sub-function description texts. Text cleaning includes word segmentation processing, special symbol removal processing, etc.

[0060] The target requirement text and the target function text are serialized to obtain a target requirement text sequence and a target function text sequence. Serialization may refer to converting the text of an entity into an abstract numerical representation, so that the text information can be converted into a data stream for easy reading and transmission. Since the message application requirement description text includes multiple sub-requirement texts, the target requirement text sequence corresponds to multiple sub-target requirement text sequences; accordingly, the function description text sequence corresponds to multiple sub-function description text sequences.

[0061] The target requirement text sequence and the target function text sequence are respectively converted into vector representation to obtain a target requirement feature vector and a target function feature vector, wherein the target function feature vector includes feature attributes of the corresponding target function node.

[0062] Among them, the target requirement text and the target function text are serialized to obtain a target requirement text sequence and a target function text sequence. Specifically, the target requirement text and the target function text are converted into an integer sequence through a preset dictionary, and the sequence length of the longest sub-target requirement text in the target requirement text is taken as the demand coding sequence length demand_length. The sub-target requirement text sequence whose sequence length is less than the demand coding sequence length is completed to the demand coding sequence length. Similarly, the function coding sequence length is determined according to the length of the longest sub-target function text sequence in the target function text sequence, and any sub-target function text sequence in the target function text sequence is completed according to the function coding sequence length. The embodiment of the present invention does not specifically limit the content of the preset dictionary, and those skilled in the art can construct a dictionary according to the existing text dictionary conversion rules. The dictionary size is demand_vocab_size. The target function feature vector of the target function node is: {x 1 、x 2 、x 3 , …, x N}, containing the function text sequence of each target function node.

[0063] In an embodiment of the present invention, for 5G message application scenarios, a historical 5G message industry application requirement description set and a corresponding historical function node feature set are obtained from the 5G message open platform, and the missing orchestration relationships between the function nodes of each 5G message application requirement in the data set are annotated. The list of historical existing function nodes is represented as a directed, unweighted topological graph, in which the function feature vector contains the characteristic attributes of each function node. The function node topological graph can be represented as a directed, unweighted graph G = (V, E, R), where E is a set of edges, where an edge represents the orchestration relationship between two function nodes. Since the calling relationships between the target function nodes are unknown, the input E is empty. V is a set of atomic capability nodes V = {V 1 ,V 2 ,V 3 ,…,V N}, represented by the target function feature vector set of the function node.

[0064] Step 130: Input the target demand feature vector and the target function feature vector into the message composite capability orchestration model to obtain the target orchestration relationship between the target function nodes corresponding to the target message application demand parameters; wherein the message composite capability orchestration model is trained based on historical message application demand parameters, function parameters of historical function nodes and corresponding historical orchestration relationships between historical function nodes.

[0065] Among them, Figure 2As shown, the specific structure of the message composite capability orchestration model of the embodiment of the present invention is shown. The present invention is implemented on the basis of the relational graph convolutional neural network GCN to optimize the message composite capability orchestration model. The essence of GCN is to extract the spatial features of the topological graph, and R-GCN can process the multi-relational data features in the knowledge base. The graph in the embodiment of the present invention refers to the capability orchestration topological graph, each functional node in the topological graph represents a network capability, each edge represents the call relationship between the functional nodes, and the feature of each functional node is the characteristic attribute of the functional node. The message composite capability orchestration model can learn a mapping of signals or features on the capability orchestration topological graph G = (V, E, R), input the feature vector of each functional node of the model, use the relational graph convolutional network to generate the functional potential feature representation of each functional node, and then use these functional potential feature representations to predict the tensor factor decomposition model of the labeled edge, thereby predicting the call relationship between the functional nodes.

[0066] In the embodiment of the present invention, the message composite capability orchestration model includes a network capability topology encoder, a message application requirement encoder and a capability orchestration topology generator.

[0067] The network capability topology encoder includes a first relationship graph convolution layer, a first fully connected layer, a first random discard layer, a second relationship graph convolution layer, a second fully connected layer and a second random discard layer; the network capability topology encoder is used to extract features from the target function feature vector to obtain a functional latent feature representation; the network capability topology encoder is used to extract features from the target function feature vector to obtain a functional latent feature representation.

[0068] The message application requirement encoder includes a first word embedding layer, a first convolutional layer, a first pooling layer, a second convolutional layer, a second pooling layer and a first flattening layer; the message application requirement encoder is used to extract features from the requirement feature vector to obtain a requirement potential feature representation. The message application requirement encoder is used to extract features from the requirement feature vector to obtain a requirement potential feature representation.

[0069] The capability orchestration topology generator includes a first merging layer and a tensor factor decomposition layer; the capability orchestration topology generator is used to determine the target orchestration relationship between each target function node according to the function potential feature representation and the requirement potential feature representation. The capability orchestration topology generator is used to determine the target orchestration relationship between each target function node according to the function potential feature representation and the requirement potential feature representation.

[0070] That is, the target function feature vector is input into the network capability topology encoder, and the network capability topology encoder maps each target function node according to the target requirement feature vector to obtain the function potential feature representation of each network capability node. The target requirement feature vector is input into the message application requirement encoder to obtain the requirement potential feature representation corresponding to the target message application requirement parameter according to the target requirement feature vector. After the capability orchestration topology generator merges the function potential feature representation and the requirement potential feature representation, it scores the potential orchestration relationship between each target function node through tensor factor decomposition operation, thereby obtaining the target orchestration relationship between each target function node corresponding to the target message application requirement parameter.

[0071] Step 140: Output the target orchestration relationship between each target function node corresponding to the target message application requirement parameter.

[0072] After obtaining the target orchestration relationship between the target functional nodes corresponding to the target message application requirement parameters in the above manner, a functional node orchestration plan is formed, and the functional node orchestration plan is fed back to the 5G message open platform. The 5G message open platform performs orchestration and implementation according to the functional node orchestration plan, thereby realizing the functional realization of industry users' message application requirements.

[0073] The embodiment of the present invention uses a neural network to predict the orchestration relationship of capability nodes corresponding to message application requirements, and can implement multi-scenario message applications according to message application requirements. Enterprises can quickly complete the deployment of message applications through the platform without complex code development, and can automatically, simply and conveniently create message applications.

[0074] In addition, by setting the specific structures of the network capability topology encoder, the message application requirement encoder and the capability orchestration topology generator, the neural network can make the prediction of the orchestration relationship more accurate.

[0075] Please refer to Figure 2 and Figure 3 , Figure 3 The training flow chart of the message composite capability arrangement model of the embodiment of the present invention is shown. Before executing steps 110-140, it is also necessary to perform model training on the message composite capability arrangement model. Specifically, the following steps are included:

[0076] Step 210: Acquire historical data, wherein the historical data includes historical message application requirement parameters, corresponding function parameters of each historical function node, and the orchestration relationship between the historical function nodes.

[0077] Step 220: Convert the historical message application requirement parameters and the function parameters of each historical function node into a historical requirement feature vector and a historical function feature vector respectively.

[0078] Among them, the conversion process of the historical demand feature vector and the historical function feature vector is the same as the conversion process of the target demand feature vector and the target function feature vector, which will not be repeated here.

[0079] After obtaining the historical function feature vectors corresponding to each of the historical function nodes, the historical orchestration relationship between each of the historical function feature vectors is annotated according to the historical orchestration relationship between the historical function nodes. Use 1 to represent that function node i needs to call function node j, output -1 represents that function node i needs to be called by function node j, and output 0 represents that there is no calling relationship between function node i and function node j. After the annotation is completed, a training set is obtained. The training set includes the historical demand feature vector and the corresponding set of historical function feature vectors carrying the historical orchestration relationship. After obtaining the training set, the training sample set can be divided into a training sample set and a test sample set for use in training the message composite capability orchestration model.

[0080] For 5G message application scenarios, the historical 5G message industry application requirement description set and the corresponding historical function node feature set are obtained from the 5G message open platform, and the missing orchestration relationship between the function nodes of each 5G message application requirement in the data set is annotated. The list of historical existing function nodes is represented as a directed, unweighted topological graph, in which the function feature vector contains the characteristic attributes of each function node.

[0081] Step 230: Construct a message composite capability orchestration model.

[0082] As described above, in a specific embodiment of the present invention, the message composite capability orchestration model includes a network capability topology encoder, a message application requirement encoder and a capability orchestration topology generator.

[0083] First, the network capability topology encoder includes a first relationship graph convolution layer, a first fully connected layer, a first random discard layer, a second relationship graph convolution layer, a second fully connected layer and a second random discard layer connected in sequence; the network capability topology encoder is used to perform feature extraction on the target function feature vector to obtain a functional latent feature representation; the network capability topology encoder is used to perform feature extraction on the target function feature vector to obtain a functional latent feature representation.

[0084] Specifically, the number of convolution kernels of the first relationship graph convolution layer is set to 128, that is, the output dimension, and the activation function is set to "relu". The expression of each layer of the first relationship graph convolution layer is:

[0085]

[0086] Among them, h il is the node v in the lth layer of the neural network i The hidden state of , r is the relationship type, is the parameter matrix of the lth neural network layer specific to the relation type. The activation function is ReLU(·)=max(0,·), where relu is a rectified linear unit. The input of the first layer is the feature vector x of the functional node. i =h i 0 , if L layers are stacked, the final output of the encoder is z i =h i L In ordinary GCN, D' -1 / 2 A'D' -1 / 2 It is the symmetric normalization of the adjacency matrix A, A'=A+I, D' is the node degree diagonal matrix of A', and for a single functional node in the embodiment of the present invention, normalization is to divide it by the degree of its functional node, so that the value of each adjacent edge information transmission is standardized, and the influence of the former will not be greater than that of the latter due to the fact that one functional node has more edges and the other node has fewer edges. Therefore, 1 / c i,r This is equivalent to the normalization of the adjacency matrix in GCN, c i,r is a regularization constant, choose c i,r =|N i r |, N i r Represents the neighbor set of function node i under relationship r.

[0087] The number of neurons in the first fully connected layer is 128, and the activation function is set to "relu"; the first random discard layer sets the discard probability to 0.2, and randomly disconnects the input neurons according to the first preset probability each time the parameters are updated during the training process. The first random discard layer is used to prevent overfitting.

[0088] The number of convolution kernels of the second relationship graph convolution layer is 64, and the activation function is set to "lamda". The expression of each layer of the second relationship graph convolution layer is the same as the expression of the first relationship graph convolution graph above, which will not be repeated here.

[0089] The number of neurons in the second fully connected layer is 64, and the activation function is set to "relu".

[0090] The dropout probability of the second random dropout layer is set to 0.2.

[0091] By utilizing the network capability topology encoder, the functional latent feature representation Z of each functional node is output.

[0092] Furthermore, the message application requirement encoder includes a first word embedding layer, a first convolution layer, a first pooling layer, a second convolution layer, a second pooling layer and a first flattening layer connected in sequence; the message application requirement encoder is used to extract features from the requirement feature vector to obtain a requirement potential feature representation. The message application requirement encoder is used to extract features from the requirement feature vector to obtain a requirement potential feature representation.

[0093] Specifically, the input data dimensions of the first word embedding layer (embedding) are set to scene_length and template_length respectively, and the output is set to the size of 128 dimensions required to convert the word into the vector space. The function of this layer is to perform vector mapping (word embeddings) on each word in the input text, that is, to convert the integer sequence of each word in the text into a vector of fixed shape 128 dimensions. In the embodiment of the present invention, the demand feature vector is converted into a vector of fixed shape 128 dimensions.

[0094] The number of convolution kernels in the first convolution layer is 128 (i.e., the dimension of the output), the spatial window length of the convolution kernel is set to 2 (i.e., the convolution kernel reads 2 words continuously each time), and the activation function is set to "relu". The convolution layer is used to extract text features. The embodiment of the present invention extracts text features from the data processed by the first word embedding layer.

[0095] The first pooling layer is the maximum pooling layer, and the pooling window size is set to 2. The maximum pooling layer retains the maximum value of the eigenvalues ​​extracted by the convolution kernel of the first convolutional layer and discards all other eigenvalues.

[0096] The number of convolution kernels in the second convolution layer is 128 (i.e., the dimension of the output), the spatial window length of the convolution kernel is set to 2 (i.e., the convolution kernel reads 2 words continuously each time), and the activation function is set to "relu". The convolution layer is used to extract text features. The embodiment of the present invention extracts text features from the data processed by the first pooling layer.

[0097] The second pooling layer is the maximum pooling layer, and the pooling window size is set to 2. The maximum pooling layer retains the maximum value of the eigenvalues ​​extracted by the convolution kernel of the second convolutional layer and discards all other eigenvalues.

[0098] The first flattening layer is used to “flatten” the input, converting the three-dimensional input into two dimensions.

[0099] Through the network capability topology encoder and the message application requirement encoder, the target function feature vector and the target requirement feature vector are mapped to the function latent feature representation Z and the requirement latent feature representation U with dimension d respectively.

[0100] Finally, the capability orchestration topology generator includes a first merging layer and a tensor factor decomposition layer connected in sequence; the capability orchestration topology generator is used to determine the target orchestration relationship between each target function node according to the function potential feature representation and the requirement potential feature representation. The capability orchestration topology generator is used to determine the target orchestration relationship between each target function node according to the function potential feature representation and the requirement potential feature representation.

[0101] Specifically, the first merging layer is used to merge the function potential feature representation Z and the requirement potential feature representation U into a potential feature vector Z'. For example, for function node i and function node j, the potential feature vectors are Z i ' and Z j '.

[0102] The tensor factorization layer (DistMult factorization) is used to predict the candidate edges (v i ,r,v j ), the activation function is set to "sigmoid".

[0103] By the function g(v i ,r,v j ) to the possible edges (v i ,r,v j ) to determine the possibility that these edges belong to the set E. The score g(v i ,r,v j ) represents the function node v i and function node v j The possibility of association through relationship r. Relation r represents the function node v i and function node v j The arrangement relationship between can realize the relationship between any sub-goal demand feature vector in the target demand feature vector. Using the potential feature vector z of the merged function node i and function node j i ' and z j '. Predict candidate edges (v) by using DistMult decomposition as the scoring function i ,r,v j ), the specific formula is:

[0104]

[0105] Among them, R r is a diagonal matrix of shape d*d, which represents z i 'The importance of each dimension to the association relationship r. Finally, sigmoid(g(v i ,r,v j )) represents the edge (vi ,r,v j ) belongs to a certain arrangement relationship. Output 1 represents that capability node i needs to call capability node j, output -1 represents that capability node i needs to be called by capability node j, and output 0 represents that there is no calling relationship between capability node i and capability node j.

[0106] Step 240: Input the training sample set into the constructed message compound capability orchestration model for training, thereby obtaining a trained message compound capability orchestration model.

[0107] Specifically, the historical demand feature vector and the corresponding historical function feature vector set carrying the historical orchestration relationship are respectively input into the network capability topology encoder, the message application demand encoder and the capability orchestration topology generator, and the error between the predicted orchestration relationship and the real function orchestration relationship (i.e., the orchestration relationship label) is calculated. The training goal is to minimize the error. The objective function selects the 'categorical_crossentropy' multi-class logarithmic loss function. The number of training rounds is set to 2000 (epochs = 2000), and the gradient descent optimization algorithm selects the adam optimizer to improve the learning speed of the traditional gradient descent (optimizer = 'adam'). Through gradient descent, the optimal weight value of the parameter of the message composite capability orchestration model that minimizes the objective function can be found. By using the training set for training, the smaller the objective function, the better, and after each round of training, the test set is used to evaluate and verify the model, and the weight value is adjusted until the objective function converges, and the current weight value is determined as the parameter weight of the message composite capability orchestration model, thereby obtaining a trained message composite capability orchestration model.

[0108] Figure 4 FIG. 1 is a schematic diagram showing the structure of a device for automatically arranging message application functions provided by an embodiment of the present invention. Figure 4 As shown, the device 300 includes: an acquisition module 310, a conversion module 320, a prediction module 330 and an output module 340.

[0109] The acquisition module 310 is used to acquire the target message application requirement parameters and the function parameter set; the function parameter set includes the function parameters of each target function node;

[0110] A conversion module 320, configured to convert the target message application requirement parameters and the function parameters of each target function node into a target requirement feature vector and a target function feature vector respectively;

[0111] The prediction module 330 is used to input the target demand feature vector and the target function feature vector into the message composite capability orchestration model to obtain the target orchestration relationship between each target function node corresponding to the target message application demand parameter; wherein the message composite capability orchestration model is trained based on the historical message application demand parameters, the function parameters of the historical function nodes and the historical orchestration relationship between the corresponding historical function nodes;

[0112] The output module 340 is used to output the target orchestration relationship between each target function node corresponding to the target message application requirement parameter.

[0113] In an optional manner, the message composite capability orchestration model includes a network capability topology encoder, a message application requirement encoder, and a capability orchestration topology generator;

[0114] The network capability topology encoder is used to extract features from the target function feature vector to obtain a potential feature representation of the function;

[0115] The message application requirement encoder is used to extract features from the requirement feature vector to obtain a requirement potential feature representation;

[0116] The capability orchestration topology generator is used to determine the target orchestration relationship between target function nodes according to the function potential feature representation and the requirement potential feature representation.

[0117] In an optional manner, the network capability topology encoder includes a first relationship graph convolution layer, a first fully connected layer, a first random discard layer, a second relationship graph convolution layer, a second fully connected layer and a second random discard layer; the network capability topology encoder is used to extract features from the target function feature vector to obtain a functional potential feature representation;

[0118] The message application requirement encoder comprises a first word embedding layer, a first convolution layer, a first pooling layer, a second convolution layer, a second pooling layer, and a first flattening layer; the message application requirement encoder is used to extract features from the requirement feature vector to obtain a potential feature representation of the requirement;

[0119] The capability orchestration topology generator comprises a first merging layer and a tensor factor decomposition layer; the capability orchestration topology generator is used to determine a target orchestration relationship between target function nodes according to the function potential feature representation and the requirement potential feature representation.

[0120] In an optional manner, the message composite capability orchestration model is trained according to historical message application requirement parameters, function parameters of historical function nodes and historical orchestration relationships between corresponding historical function nodes, and further includes:

[0121] Obtain historical message application requirement parameters and historical function parameters of each function node;

[0122] Convert historical message application demand parameters into historical demand feature vectors;

[0123] Converting the function parameters of the historical function nodes into historical function feature vectors corresponding to the historical function nodes;

[0124] Marking the arrangement relationship of each historical function feature vector according to the arrangement relationship between each historical function node to obtain a historical feature vector set, wherein the historical feature vector set includes the historical function feature vectors and the historical arrangement relationship of each historical function node;

[0125] The historical demand feature vector and the historical feature vector set are input into a preset neural network model for training to obtain a message composite capability orchestration model.

[0126] In an optional manner, the target message application requirement parameters include a target message application requirement description text, and the function parameters of each target function node include a function description text of each target function node; converting the target message application requirement parameters and the function parameters of each target function node into a target requirement feature vector and a target function feature vector, respectively, includes:

[0127] Performing text cleaning on the message application requirement description text and the function description text of each function node to obtain a target requirement text and a target function text;

[0128] Serializing the target requirement text and the target function text to obtain a target requirement text sequence and a target function text sequence;

[0129] The target requirement text sequence and the target function text sequence are respectively converted into vector representation to obtain a target requirement feature vector and a target function feature vector.

[0130] In an optional manner, a target message application requirement parameter and a function parameter set are obtained; the function parameter set includes function parameters of each target function node, including:

[0131] A target message application requirement request is obtained, wherein the target message application requirement request carries the target message application requirement parameter.

[0132] The embodiment of the present invention uses a neural network to predict the orchestration relationship of capability nodes corresponding to message application requirements, and can implement multi-scenario message applications according to message application requirements. Enterprises can quickly complete the deployment of message applications through the platform without complex code development, and can automatically, simply and conveniently create message applications.

[0133] In addition, by setting the specific structures of the network capability topology encoder, the message application requirement encoder and the capability orchestration topology generator, the neural network can make the prediction of the orchestration relationship more accurate.

[0134] Figure 5 The schematic diagram of the structure of the automatic arrangement device for message application functions provided by the embodiment of the present invention is shown. The specific embodiment of the present invention does not limit the specific implementation of the automatic arrangement device for message application functions.

[0135] like Figure 5 As shown, the message application function automatic arrangement device may include: a processor (processor) 402, a communication interface (Communications Interface) 404, a memory (memory) 406, and a communication bus 408.

[0136] The processor 402, the communication interface 404, and the memory 406 communicate with each other via the communication bus 408. The communication interface 404 is used to communicate with other devices such as a client or other server network elements. The processor 402 is used to execute the program 410, which can specifically execute the relevant steps in the above-mentioned embodiment of the method for automatic arrangement of message application functions.

[0137] Specifically, the program 410 may include program code including computer executable instructions.

[0138] The processor 402 may be a central processing unit (CPU), or an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiment of the present invention. The one or more processors included in the automatic arrangement device for message application functions may be processors of the same type, such as one or more CPUs; or processors of different types, such as one or more CPUs and one or more ASICs.

[0139] The memory 406 is used to store the program 410. The memory 406 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0140] Program 410 can be specifically called by processor 402 to enable the message application function automatic arrangement device to perform the following operations:

[0141] Obtaining target message application requirement parameters and a function parameter set; the function parameter set includes function parameters of each target function node;

[0142] Convert the target message application requirement parameters and the function parameters of each target function node into a target requirement feature vector and a target function feature vector respectively;

[0143] The target requirement feature vector and the target function feature vector are input into the message composite capability orchestration model to obtain the target orchestration relationship between each target function node corresponding to the target message application requirement parameter; wherein the message composite capability orchestration model is trained based on the historical message application requirement parameters, the function parameters of the historical function nodes and the historical orchestration relationship between the corresponding historical function nodes;

[0144] Output the target orchestration relationship between each target function node corresponding to the target message application requirement parameter.

[0145] In an optional manner, the message composite capability orchestration model includes a network capability topology encoder, a message application requirement encoder, and a capability orchestration topology generator;

[0146] The network capability topology encoder is used to extract features from the target function feature vector to obtain a potential feature representation of the function;

[0147] The message application requirement encoder is used to extract features from the requirement feature vector to obtain a requirement potential feature representation;

[0148] The capability orchestration topology generator is used to determine the target orchestration relationship between target function nodes according to the function potential feature representation and the requirement potential feature representation.

[0149] In an optional manner, the network capability topology encoder includes a first relationship graph convolution layer, a first fully connected layer, a first random discard layer, a second relationship graph convolution layer, a second fully connected layer and a second random discard layer; the network capability topology encoder is used to extract features from the target function feature vector to obtain a functional potential feature representation;

[0150] The message application requirement encoder comprises a first word embedding layer, a first convolution layer, a first pooling layer, a second convolution layer, a second pooling layer, and a first flattening layer; the message application requirement encoder is used to extract features from the requirement feature vector to obtain a potential feature representation of the requirement;

[0151] The capability orchestration topology generator comprises a first merging layer and a tensor factor decomposition layer; the capability orchestration topology generator is used to determine a target orchestration relationship between target function nodes according to the function potential feature representation and the requirement potential feature representation.

[0152] In an optional manner, the message composite capability orchestration model is trained according to historical message application requirement parameters, function parameters of historical function nodes and historical orchestration relationships between corresponding historical function nodes, and further includes:

[0153] Obtain historical message application requirement parameters and historical function parameters of each function node;

[0154] Convert historical message application demand parameters into historical demand feature vectors;

[0155] Converting the function parameters of the historical function nodes into historical function feature vectors corresponding to the historical function nodes;

[0156] Marking the arrangement relationship of each historical function feature vector according to the arrangement relationship between each historical function node to obtain a historical feature vector set, wherein the historical feature vector set includes the historical function feature vectors and the historical arrangement relationship of each historical function node;

[0157] The historical demand feature vector and the historical feature vector set are input into a preset neural network model for training to obtain a message composite capability orchestration model.

[0158] In an optional manner, the target message application requirement parameters include a target message application requirement description text, and the function parameters of each target function node include a function description text of each target function node; converting the target message application requirement parameters and the function parameters of each target function node into a target requirement feature vector and a target function feature vector, respectively, includes:

[0159] Performing text cleaning on the message application requirement description text and the function description text of each function node to obtain a target requirement text and a target function text;

[0160] Serializing the target requirement text and the target function text to obtain a target requirement text sequence and a target function text sequence;

[0161] The target requirement text sequence and the target function text sequence are respectively converted into vector representation to obtain a target requirement feature vector and a target function feature vector.

[0162] In an optional manner, a target message application requirement parameter and a function parameter set are obtained; the function parameter set includes function parameters of each target function node, including:

[0163] A target message application requirement request is obtained, wherein the target message application requirement request carries the target message application requirement parameter.

[0164] The embodiment of the present invention uses a neural network to predict the orchestration relationship of capability nodes corresponding to message application requirements, and can implement multi-scenario message applications according to message application requirements. Enterprises can quickly complete the deployment of message applications through the platform without complex code development, and can automatically, simply and conveniently create message applications.

[0165] In addition, by setting the specific structures of the network capability topology encoder, the message application requirement encoder and the capability orchestration topology generator, the neural network can make the prediction of the orchestration relationship more accurate.

[0166] An embodiment of the present invention provides a computer-readable storage medium storing at least one executable instruction. When the executable instruction runs on a message application function automatic arrangement device, the message application function automatic arrangement device executes the message application function automatic arrangement method in any of the above method embodiments.

[0167] The executable instructions may be specifically used to enable the message application function automatic arrangement device to perform the following operations:

[0168] Obtaining target message application requirement parameters and a function parameter set; the function parameter set includes function parameters of each target function node;

[0169] Convert the target message application requirement parameters and the function parameters of each target function node into a target requirement feature vector and a target function feature vector respectively;

[0170] The target requirement feature vector and the target function feature vector are input into the message composite capability orchestration model to obtain the target orchestration relationship between each target function node corresponding to the target message application requirement parameter; wherein the message composite capability orchestration model is trained based on the historical message application requirement parameters, the function parameters of the historical function nodes and the historical orchestration relationship between the corresponding historical function nodes;

[0171] Output the target orchestration relationship between each target function node corresponding to the target message application requirement parameter.

[0172] In an optional manner, the message composite capability orchestration model includes a network capability topology encoder, a message application requirement encoder, and a capability orchestration topology generator;

[0173] The network capability topology encoder is used to extract features from the target function feature vector to obtain a potential feature representation of the function;

[0174] The message application requirement encoder is used to extract features from the requirement feature vector to obtain a requirement potential feature representation;

[0175] The capability orchestration topology generator capability orchestration topology generator is used to determine the target orchestration relationship between each target function node according to the function potential feature representation and the requirement potential feature representation.

[0176] In an optional manner, the network capability topology encoder includes a first relationship graph convolution layer, a first fully connected layer, a first random discard layer, a second relationship graph convolution layer, a second fully connected layer and a second random discard layer; the network capability topology encoder is used to extract features from the target function feature vector to obtain a functional potential feature representation;

[0177] The message application requirement encoder comprises a first word embedding layer, a first convolution layer, a first pooling layer, a second convolution layer, a second pooling layer, and a first flattening layer; the message application requirement encoder is used to extract features from the requirement feature vector to obtain a potential feature representation of the requirement;

[0178] The capability orchestration topology generator capability orchestration topology generator includes a first merging layer and a tensor factor decomposition layer; the capability orchestration topology generator capability orchestration topology generator is used to determine the target orchestration relationship between each target function node according to the function potential feature representation and the requirement potential feature representation.

[0179] In an optional manner, the message composite capability orchestration model is trained according to historical message application requirement parameters, function parameters of historical function nodes and historical orchestration relationships between corresponding historical function nodes, and further includes:

[0180] Obtain historical message application requirement parameters and historical function parameters of each function node;

[0181] Convert historical message application demand parameters into historical demand feature vectors;

[0182] Converting the function parameters of the historical function nodes into historical function feature vectors corresponding to the historical function nodes;

[0183] Marking the arrangement relationship of each historical function feature vector according to the arrangement relationship between each historical function node to obtain a historical feature vector set, wherein the historical feature vector set includes the historical function feature vectors and the historical arrangement relationship of each historical function node;

[0184] The historical demand feature vector and the historical feature vector set are input into a preset neural network model for training to obtain a message composite capability orchestration model.

[0185] In an optional manner, the target message application requirement parameters include a target message application requirement description text, and the function parameters of each target function node include a function description text of each target function node; converting the target message application requirement parameters and the function parameters of each target function node into a target requirement feature vector and a target function feature vector, respectively, includes:

[0186] Performing text cleaning on the message application requirement description text and the function description text of each function node to obtain a target requirement text and a target function text;

[0187] Serializing the target requirement text and the target function text to obtain a target requirement text sequence and a target function text sequence;

[0188] The target requirement text sequence and the target function text sequence are respectively converted into vector representation to obtain a target requirement feature vector and a target function feature vector.

[0189] In an optional manner, a target message application requirement parameter and a function parameter set are obtained; the function parameter set includes function parameters of each target function node, including:

[0190] A target message application requirement request is obtained, wherein the target message application requirement request carries the target message application requirement parameter.

[0191] The embodiment of the present invention uses a neural network to predict the orchestration relationship of capability nodes corresponding to message application requirements, and can implement multi-scenario message applications according to message application requirements. Enterprises can quickly complete the deployment of message applications through the platform without complex code development, and can automatically, simply and conveniently create message applications.

[0192] In addition, by setting the specific structures of the network capability topology encoder, the message application requirement encoder and the capability orchestration topology generator, the neural network can make the prediction of the orchestration relationship more accurate.

[0193] The embodiment of the present invention provides a device for automatically arranging message application functions, which is used to execute the above-mentioned method for automatically arranging message application functions.

[0194] An embodiment of the present invention provides a computer program, which can be called by a processor to enable a message application function automatic arrangement device to execute the message application function automatic arrangement method in any of the above method embodiments.

[0195] An embodiment of the present invention provides a computer program product, which includes a computer program stored on a computer-readable storage medium, and the computer program includes program instructions. When the program instructions are executed on a computer, the computer executes the method for automatically arranging message application functions in any of the above method embodiments.

[0196] The algorithm or display provided herein is not inherently related to any particular computer, virtual system or other equipment. Various general purpose systems can also be used together with the teachings based on this. According to the above description, it is obvious to construct the structure required for this type of system. In addition, the embodiment of the present invention is not directed to any specific programming language yet. It should be understood that various programming languages ​​can be utilized to realize the content of the present invention described herein, and the description made to specific languages ​​above is for disclosing the best mode of the present invention.

[0197] In the description provided herein, a large number of specific details are described. However, it is understood that embodiments of the present invention can be practiced without these specific details. In some instances, well-known methods, structures and techniques are not shown in detail so as not to obscure the understanding of this description.

[0198] Similarly, it should be understood that in order to streamline the present invention and aid in understanding one or more of the various inventive aspects, in the above description of exemplary embodiments of the present invention, various features of the embodiments of the present invention are sometimes grouped together into a single embodiment, figure, or description thereof. However, this method of disclosure should not be interpreted as reflecting an intention that the claimed invention requires more features than those expressly recited in each claim.

[0199] Those skilled in the art will appreciate that the modules in the devices in the embodiments may be adaptively changed and arranged in one or more devices different from the embodiments. The modules or units or components in the embodiments may be combined into one module or unit or component, and may be divided into a plurality of submodules or subunits or subcomponents. Except that at least some of such features and / or processes or units are mutually exclusive, all features disclosed in this specification (including the accompanying claims, abstracts and drawings) and all processes or units of any method or device disclosed in this manner may be combined in any combination. Unless otherwise expressly stated, each feature disclosed in this specification (including the accompanying claims, abstracts and drawings) may be replaced by an alternative feature providing the same, equivalent or similar purpose.

[0200] It should be noted that the above embodiments illustrate the present invention rather than limit it, and that those skilled in the art may devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference symbol between brackets shall not be construed as a limitation on the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "one" or "an" preceding an element does not exclude the presence of a plurality of such elements. The present invention may be implemented by means of hardware comprising a number of different elements and by means of a suitably programmed computer. In a unit claim enumerating a number of devices, several of these devices may be embodied by the same hardware item. The use of the words first, second, and third, etc. does not indicate any order. These words may be interpreted as names. The steps in the above embodiments, unless otherwise specified, should not be understood as limitations on the order of execution.

Claims

1. A method for automatically arranging message application functions, It is characterized in that The method comprises: Obtaining target message application requirement parameters and a function parameter set; the function parameter set includes function parameters of each target function node; Convert the target message application requirement parameters and the function parameters of each target function node into a target requirement feature vector and a target function feature vector respectively; The target requirement feature vector and the target function feature vector are input into the message composite capability orchestration model to obtain the target orchestration relationship between each target function node corresponding to the target message application requirement parameter; wherein the message composite capability orchestration model is trained based on the historical message application requirement parameters, the function parameters of the historical function nodes and the historical orchestration relationship between the corresponding historical function nodes; Outputting the target orchestration relationship between each target function node corresponding to the target message application requirement parameter; The message composite capability orchestration model includes a network capability topology encoder, a message application requirement encoder and a capability orchestration topology generator; The network capability topology encoder is used to extract features from the target function feature vector to obtain a potential feature representation of the function; The message application requirement encoder is used to extract features from the requirement feature vector to obtain a requirement potential feature representation; The capability orchestration topology generator is used to determine the target orchestration relationship between each target function node according to the function potential feature representation and the demand potential feature representation; The network capability topology encoder comprises a first relationship graph convolution layer, a first fully connected layer, a first random discard layer, a second relationship graph convolution layer, a second fully connected layer and a second random discard layer; the network capability topology encoder is used to extract features from the target function feature vector to obtain a potential feature representation of the function; The message application requirement encoder comprises a first word embedding layer, a first convolution layer, a first pooling layer, a second convolution layer, a second pooling layer, and a first flattening layer; the message application requirement encoder is used to extract features from the requirement feature vector to obtain a potential feature representation of the requirement; The capability orchestration topology generator comprises a first merging layer and a tensor factor decomposition layer; the capability orchestration topology generator is used to determine a target orchestration relationship between target function nodes according to the function potential feature representation and the requirement potential feature representation.

2. The method according to claim 1, It is characterized in that The message composite capability arrangement model is trained based on historical message application requirement parameters, function parameters of historical function nodes, and historical arrangement relationships between corresponding historical function nodes, and further includes: Obtain historical message application requirement parameters and historical function parameters of each function node; Convert historical message application demand parameters into historical demand feature vectors; Converting the function parameters of the historical function nodes into historical function feature vectors corresponding to the historical function nodes; Marking the arrangement relationship of each historical function feature vector according to the arrangement relationship between each historical function node to obtain a historical feature vector set, wherein the historical feature vector set includes the historical function feature vectors and the historical arrangement relationship of each historical function node; The historical demand feature vector and the historical feature vector set are input into a preset neural network model for training to obtain a message composite capability orchestration model.

3. The method according to any one of claims 1 to 2, It is characterized in that The target message application requirement parameters include a target message application requirement description text, and the function parameters of each target function node include a function description text of each target function node; converting the target message application requirement parameters and the function parameters of each target function node into a target requirement feature vector and a target function feature vector, respectively, includes: Performing text cleaning on the message application requirement description text and the function description text of each function node to obtain a target requirement text and a target function text; Serializing the target requirement text and the target function text to obtain a target requirement text sequence and a target function text sequence; The target requirement text sequence and the target function text sequence are respectively converted into vector representation to obtain a target requirement feature vector and a target function feature vector.

4. The method according to claim 1, It is characterized in that Obtain target message application requirement parameters and function parameter set; the function parameter set includes function parameters of each target function node, including: A target message application requirement request is obtained, wherein the target message application requirement request carries the target message application requirement parameter.

5. A device for automatically arranging message application functions, It is characterized in that The device comprises: An acquisition module, used to acquire target message application requirement parameters and a function parameter set; the function parameter set includes function parameters of each target function node; A conversion module, used to convert the target message application requirement parameters and the function parameters of each target function node into a target requirement feature vector and a target function feature vector respectively; A prediction module is used to input the target demand feature vector and the target function feature vector into a message composite capability orchestration model to obtain a target orchestration relationship between target function nodes corresponding to the target message application demand parameter; wherein the message composite capability orchestration model is trained based on historical message application demand parameters, function parameters of historical function nodes, and historical orchestration relationships between corresponding historical function nodes; An output module, used to output the target orchestration relationship between each target function node corresponding to the target message application requirement parameter; The message composite capability orchestration model includes a network capability topology encoder, a message application requirement encoder and a capability orchestration topology generator; The network capability topology encoder is used to extract features from the target function feature vector to obtain a potential feature representation of the function; The message application requirement encoder is used to extract features from the requirement feature vector to obtain a requirement potential feature representation; The capability orchestration topology generator is used to determine the target orchestration relationship between each target function node according to the function potential feature representation and the demand potential feature representation; The network capability topology encoder comprises a first relationship graph convolution layer, a first fully connected layer, a first random discard layer, a second relationship graph convolution layer, a second fully connected layer and a second random discard layer; the network capability topology encoder is used to extract features from the target function feature vector to obtain a potential feature representation of the function; The message application requirement encoder comprises a first word embedding layer, a first convolution layer, a first pooling layer, a second convolution layer, a second pooling layer, and a first flattening layer; the message application requirement encoder is used to extract features from the requirement feature vector to obtain a potential feature representation of the requirement; The capability orchestration topology generator comprises a first merging layer and a tensor factor decomposition layer; the capability orchestration topology generator is used to determine a target orchestration relationship between target function nodes according to the function potential feature representation and the requirement potential feature representation.

6. A device for automatically arranging message application functions, It is characterized in that include: A processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform the operation of the message application function automatic arrangement method as described in any one of claims 1-4.

7. A computer-readable storage medium, It is characterized in that The storage medium stores at least one executable instruction. When the executable instruction runs on the message application function automatic arrangement device, the message application function automatic arrangement device performs the operation of the message application function automatic arrangement method as described in any one of claims 1 to 4.

Images