CN115526474A

CN115526474A - Dispatching method, dispatching device, computer equipment and storage medium

Info

Publication number: CN115526474A
Application number: CN202211146236.7A
Authority: CN
Inventors: 严杨扬
Original assignee: Ping An Property and Casualty Insurance Company of China Ltd
Current assignee: Ping An Property and Casualty Insurance Company of China Ltd
Priority date: 2022-09-20
Filing date: 2022-09-20
Publication date: 2022-12-27

Abstract

The embodiment of the application belongs to the field of artificial intelligence and relates to a dispatching method, a dispatching device, computer equipment and a storage medium, wherein the dispatching method comprises the following steps: constructing a historical bipartite graph according to historical transaction information, wherein the historical bipartite graph comprises an operator node, a client node and node characteristics; based on node characteristics, calculating a recommended transaction probability between the operator node and the client node through an initial prediction model based on a GraphSAGE algorithm; calculating model loss according to the connection relation of the operator node and the client node in the bipartite graph and the recommended deal probability to adjust an initial prediction model to obtain a prediction model; updating the historical bipartite graph according to graph network updating information to obtain a current bipartite graph; and calculating the recommended deal probability between the operator node to be predicted and the client node to be predicted in the current bipartite graph through a prediction model so as to dispatch operators corresponding to the operator node. The application also relates to blockchain technology, and historical transaction information can be stored in blockchains. The method and the device improve the dispatching accuracy.

Description

Dispatching method, dispatching device, computer equipment and storage medium

Technical Field

The present application relates to the field of artificial intelligence technologies, and in particular, to a method and an apparatus for dispatching a job, a computer device, and a storage medium.

Background

With the development of computer technology, more and more service platforms realize dispatching through computers. In the field of financial insurance, dispatching can refer to a business platform dispatching a salesman to a customer, and the dispatched salesman can provide services such as product recommendation to the customer. The dispatch is related to the experience of the customer and the conversion rate of the product recommendation, so that the dispatch is very important for the service platform.

However, current dispatching techniques typically assign the clerks to the customers randomly, or based on their workload, with less-workload clerks being tasked more. However, these dispatching techniques do not consider whether the salesman is suitable for the customer, so that the dispatching accuracy is low.

Disclosure of Invention

An object of the embodiments of the present application is to provide a method, an apparatus, a computer device and a storage medium for dispatching, so as to solve the problem of low accuracy of dispatching.

In order to solve the foregoing technical problem, an embodiment of the present application provides a dispatching method, which adopts the following technical solutions:

acquiring historical transaction information;

constructing a history bipartite graph according to the history transaction information; the service person and the client in the historical transaction information respectively correspond to a service person node and a client node in the historical bipartite graph, and the service person node and the client node have node characteristics;

calculating a recommended deal probability between the operator node and the client node through an initial prediction model based on the node characteristics; the initial prediction model is constructed based on GraphSAGE algorithm;

calculating model loss according to the connection relation of the operator node and the client node in the bipartite graph and the recommended deal probability, and adjusting the initial prediction model based on the model loss until the initial prediction model converges to obtain a prediction model;

updating the historical bipartite graph according to graph network update information to obtain a current bipartite graph;

and calculating the recommended transaction probability between the operator node to be predicted and the client node to be predicted in the current bipartite graph through the prediction model, and dispatching the operator corresponding to the operator node according to the recommended transaction probability.

In order to solve the above technical problem, an embodiment of the present application further provides a dispatching device, which adopts the following technical solution:

the information acquisition module is used for acquiring historical transaction information;

the bipartite graph building module is used for building a historical bipartite graph according to the historical transaction information; the service person and the client in the historical transaction information respectively correspond to a service person node and a client node in the historical bipartite graph, and the service person node and the client node have node characteristics;

the probability prediction module is used for calculating the recommended deal probability between the operator node and the client node through an initial prediction model based on the node characteristics; the initial prediction model is constructed based on a GraphSAGE algorithm;

the model adjusting module is used for calculating model loss according to the connection relation of the operator nodes and the client nodes in the bipartite graph and the recommended intersection probability, and adjusting the initial prediction model based on the model loss until the initial prediction model converges to obtain a prediction model;

the bipartite graph updating module is used for updating the historical bipartite graph according to graph network updating information to obtain a current bipartite graph;

and the deal prediction module is used for calculating the recommended deal probability between the operator node to be predicted and the client node to be predicted in the current bipartite graph through the prediction model and dispatching the operator corresponding to the operator node according to the recommended deal probability.

In order to solve the above technical problem, an embodiment of the present application further provides a computer device, which adopts the following technical solutions:

acquiring historical transaction information;

constructing a history bipartite graph according to the history transaction information; the business member and the client in the historical transaction information respectively correspond to a business member node and a client node in the historical bipartite graph, and the business member node and the client node have node characteristics;

calculating a recommended deal probability between the operator node and the client node through an initial prediction model based on the node characteristics; the initial prediction model is constructed based on a GraphSAGE algorithm;

calculating model loss according to the connection relation of the operator node and the client node in the bipartite graph and the recommended intersection probability, and adjusting the initial prediction model based on the model loss until the initial prediction model converges to obtain a prediction model;

In order to solve the foregoing technical problem, an embodiment of the present application further provides a computer-readable storage medium, which adopts the following technical solutions:

acquiring historical transaction information;

calculating a recommended transaction probability between the operator node and the client node through an initial prediction model based on the node characteristics; the initial prediction model is constructed based on a GraphSAGE algorithm;

Compared with the prior art, the embodiment of the application mainly has the following beneficial effects: acquiring historical transaction information and constructing a historical bipartite graph, wherein an operator and a client in the historical transaction information respectively correspond to an operator node and a client node in the historical bipartite graph, and the operator node and the client node have node characteristics; based on node characteristics, calculating a recommendation transaction probability between an operator node and a client node through an initial prediction model constructed according to a GraphSAGE algorithm, wherein the size of the probability value represents the success probability of recommending products to the client by a predicted operator, and the connection relation between the operator node and the client node can be used as a label, so that model loss can be calculated, and the initial prediction model is adjusted until the prediction model is obtained; when the method is applied, the historical bipartite graph is updated according to graph network updating information to obtain a current bipartite graph, the recommended deal probability between a to-be-predicted salesman node and a to-be-predicted client node in the current bipartite graph can be accurately calculated through a prediction model, and the salesman corresponding to the salesman node is dispatched according to the recommended deal probability, so that the conversion rate of product recommendation after the dispatch is considered, the adaptability between the salesman and the client is improved, and the accuracy of dispatching is improved.

Drawings

In order to more clearly illustrate the solution of the present application, the drawings used in the description of the embodiments of the present application will be briefly described below, and it is obvious that the drawings in the description below are some embodiments of the present application, and that other drawings may be obtained by those skilled in the art without inventive effort.

FIG. 1 is an exemplary system architecture diagram to which the present application may be applied;

FIG. 2 is a flow diagram of one embodiment of a method of dispatch according to the present application;

FIG. 3 is a schematic structural diagram of an embodiment of a dispatching device according to the present application;

FIG. 4 is a schematic block diagram of one embodiment of a computer device according to the present application.

Detailed Description

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 in the description of the application herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the application; the terms "including" and "having," and any variations thereof, in the description and claims of this application and the description of the above figures are intended to cover non-exclusive inclusions. The terms "first," "second," and the like in the description and claims of this application or in the foregoing drawings are used for distinguishing between different objects and not for describing a particular sequential order.

Reference herein to "an embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the application. The appearances of the phrase in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. It is explicitly and implicitly understood by one skilled in the art that the embodiments described herein can be combined with other embodiments.

In order to make the technical solutions better understood by those skilled in the art, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

As shown in fig. 1, the system architecture 100 may include

terminal devices

101, 102, 103, a network 104, and a server 105. The network 104 serves as a medium for providing communication links between the

terminal devices

101, 102, 103 and the server 105. Network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables, to name a few.

The user may use the

terminal devices

101, 102, 103 to interact with the server 105 via the network 104 to receive or send messages or the like. Various communication client applications, such as a web browser application, a shopping application, a search application, an instant messaging tool, a mailbox client, social platform software, etc., may be installed on the

terminal devices

101, 102, 103.

The

terminal devices

101, 102, 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to a smart phone, a tablet computer, an e-book reader, an MP3 player (Moving Picture experts Group Audio Layer III, motion Picture experts compression standard Audio Layer 3), an MP4 player (Moving Picture experts Group Audio Layer IV, motion Picture experts compression standard Audio Layer 4), a laptop portable computer, a desktop computer, and the like.

The server 105 may be a server providing various services, such as a background server providing support for pages displayed on the

terminal devices

101, 102, 103.

It should be noted that the dispatching method provided in the embodiment of the present application is generally executed by a server, and accordingly, the dispatching device is generally disposed in the server.

It should be understood that the number of terminal devices, networks, and servers in fig. 1 is merely illustrative. There may be any number of terminal devices, networks, and servers, as desired for an implementation.

With continued reference to FIG. 2, a flow diagram of one embodiment of a method of dispatch in accordance with the subject application is shown. The dispatching method comprises the following steps:

in step S201, historical transaction information is acquired.

In this embodiment, the electronic device (for example, the server shown in fig. 1) on which the dispatch method operates may communicate with the terminal through a wired connection manner or a wireless connection manner. It should be noted that the wireless connection means may include, but is not limited to, a 3G/4G/5G connection, a WiFi connection, a bluetooth connection, a WiMAX connection, a Zigbee connection, an UWB (ultra wideband) connection, and other wireless connection means now known or developed in the future.

The historical transaction information can record product recommendation behaviors occurring between a salesperson and a client historically and product transaction behaviors generated when the client purchases products.

Specifically, the historical transaction information is recorded information of past clerk and customer behaviors. Information interaction can be carried out between the salesman and the client, such as chatting and communication behaviors generated between the salesman and the client, the client can consult the salesman, and the salesman can recommend products to the client; after a consultation or product recommendation, the customer may purchase the product; all the above information is recorded to generate historical transaction information.

The historical transaction information may also include clerk characteristics as well as customer characteristics. The characteristics of the service personnel are related characteristics of the service personnel, such as personal basic information, historical performance information, portrayal and the like of the service personnel. The client characteristics are client-related characteristics, including consultation characteristics (such as historical price inquiry times of a client for an insurance product), claim characteristics (such as claim state, claim amount, claim mode, whether to take out insurance from other places and the like of the insurance product purchased by the client), basic information characteristics (such as sex, age, region and the like of the client), association mechanism characteristics (data of other subsidiaries of the group of the client, such as transaction amount), target information characteristics (brand of a target object, delivery time and the like), client social characteristics (such as policyholder, social software activity and the like) and client intention characteristics (such as intention of the client on preference in consultation of the product and intention of the client on product details).

It is emphasized that, in order to further ensure the privacy and security of the historical transaction information, the historical transaction information may also be stored in a node of a block chain.

The block chain referred by the application is a novel application mode of computer technologies such as distributed data storage, point-to-point transmission, a consensus mechanism, an encryption algorithm and the like. A block chain (Blockchain), which is essentially a decentralized database, is a series of data blocks associated by using a cryptographic method, and each data block contains information of a batch of network transactions, so as to verify the validity (anti-counterfeiting) of the information and generate a next block. The blockchain may include a blockchain underlying platform, a platform product service layer, an application service layer, and the like.

Step S202, a history bipartite graph is constructed according to history transaction information; the salesman and the client in the historical transaction information respectively correspond to a salesman node and a client node in the historical bipartite graph, and the salesman node and the client node have node characteristics.

Specifically, a bipartite graph is constructed according to historical transaction information, and the bipartite graph is also called a bipartite graph and is a model in a graph theory. The bipartite graph comprises nodes and connecting edges, wherein the nodes comprise an operator node and a client node, the operator node corresponds to an operator in the historical transaction information, and the client node corresponds to a client in the historical transaction information. If the customer purchases the product after being recommended by the salesman, a connecting edge exists between the salesman node corresponding to the salesman and the customer node corresponding to the customer. In addition, the attendant nodes and client nodes in the bipartite graph have node characteristics. The node characteristics of the attendant nodes are from attendant characteristics of the attendant, and the node characteristics of the client nodes are from client characteristics.

Step S203, calculating the recommended deal probability between the operator node and the client node through an initial prediction model based on the node characteristics; the initial prediction model is constructed based on the GraphSAGE algorithm.

Wherein the initial prediction model may be a prediction model for which model training has not been completed.

Specifically, the recommendation deal probability between the salesman and the client, that is, the possibility that the client purchases the product after the salesman recommends the product to the client, is predicted through the prediction model. The initial prediction model/prediction model is constructed based on a GraphSAGE algorithm, the GraphSAGE algorithm is a graph neural network algorithm, the core of the algorithm is to optimize the sampling of the whole graph to the sampling of the neighbor nodes of the target node, the K neighbor nodes are used for calculating the attribute of the target node, and the model is an inductive learning model.

In training, if the recommendation transaction probability between the operator node and the client node needs to be predicted, the client node is required to be used as a collected neighbor node, and then the recommendation transaction probability between the operator node and the client node is calculated through an initial prediction model based on a GraphSAGE algorithm; and the node characteristics of the related nodes are used as the input of the GraphSAGE algorithm to complete the calculation of the recommended deal probability.

And step S204, calculating model loss according to the connection relation and the recommended transaction probability of the operator node and the client node in the bipartite graph, and adjusting the initial prediction model based on the model loss until the initial prediction model converges to obtain the prediction model.

Specifically, the bipartite graph carries labels required for model training. If a connecting edge exists between the operator node and the client node, the condition that the operator successfully recommends a product to the client once is indicated, and the label is marked as 1; if there is no connecting edge between the operator node and the customer node, it indicates that the operator has not successfully recommended the product to the customer, and the label is marked as 0.

The loss function in the training may be a cross-entropy loss function. Model loss can be calculated according to the labels and the recommended deal probability; and adjusting model parameters of the initial prediction model by taking the minimized model loss as a training target, performing iterative training on the initial prediction model after parameter adjustment until the initial prediction model is converged, and stopping training to obtain the prediction model.

Step S205, the historical bipartite graph is updated according to the graph network updating information, and the current bipartite graph is obtained.

Specifically, the historical bipartite graph is constructed according to historical data, new salesmen and new customers may be added in the application, and at this time, graph network update information needs to be acquired to update the historical bipartite graph to obtain the current bipartite graph.

And step S206, calculating the recommended transaction probability between the operator node to be predicted and the client node to be predicted in the current bipartite graph through a prediction model, and dispatching the operator corresponding to the operator node according to the recommended transaction probability.

Specifically, the nodes in the current bipartite graph are the waiter node to be predicted and the client node to be predicted. The prediction model can respectively predict the recommended transaction probability between each operator node to be predicted and each client node to be predicted.

It is understood that the larger the numerical value of the recommended deal probability, the higher the possibility that the customer purchases the product after the representative dealer recommends the product to the customer. Therefore, the staff corresponding to the staff can be dispatched according to the recommended transaction probability, that is, the higher the recommended transaction probability between the staff and the client is, the better the staff is dispatched to the client.

In the embodiment, historical transaction information is obtained and a historical bipartite graph is constructed, wherein an operator and a client in the historical transaction information respectively correspond to an operator node and a client node in the historical bipartite graph, and the operator node and the client node have node characteristics; based on node characteristics, calculating a recommendation interaction probability between an operator node and a client node through an initial prediction model constructed according to a GraphSAGE algorithm, wherein the size of the probability value represents the possibility of success of recommending products to a client by a predicted operator, and the connection relation between the operator node and the client node can be used as a label, so that model loss can be calculated, and the initial prediction model is adjusted until the prediction model is obtained; when the method is applied, the historical bipartite graph is updated according to graph network updating information to obtain a current bipartite graph, the recommended deal probability between a to-be-predicted salesman node and a to-be-predicted client node in the current bipartite graph can be accurately calculated through a prediction model, and the salesman corresponding to the salesman node is dispatched according to the recommended deal probability, so that the conversion rate of product recommendation after the dispatch is considered, the adaptability between the salesman and the client is improved, and the accuracy of dispatching is improved.

Further, the step S202 may include: identifying an operator and a client in the historical transaction information and a recommended transaction relationship between the operator and the client; extracting the characteristics of the salesman and the characteristics of the customer from the historical transaction information; the method comprises the following steps of respectively taking a salesman and a client as a salesman node and a client node, and connecting the salesman node and the client node with a recommended transaction relationship; and determining the characteristics of the operator as the node characteristics of the corresponding operator node, and determining the characteristics of the client as the node characteristics of the corresponding client node to obtain a historical bipartite graph.

Specifically, the data stored in the historical transaction information is provided with an identifier, and the salesman and the client recorded in the data can be identified according to the identifier, and the recommended transaction relationship between the salesman and the client can be identified. Wherein, if the salesman successfully recommends the product to the customer, the salesman and the customer have a recommendation interaction relationship. Meanwhile, the characteristics of the salesman and the characteristics of the customer can be extracted from the historical transaction information.

When the history bipartite graph is constructed, an operator is used as an operator node, and a client is used as a client node. And if the recommended interaction relationship exists between the service person and the client, connecting the service person node corresponding to the service person and the client node corresponding to the client by using the connecting edge. The connecting edge can be provided with an identifier and used for displaying the type of the connecting edge, and the connecting edge can be established based on the recommendation intersection relationship; in one embodiment, if it is determined through the historical transaction information that there has been a chat or consultation between the operator and the client, a connection edge may also be established between the operator node and the client node, and the connection edge is distinguished from the connection edge established based on the recommended transaction relationship by the connection edge identifier.

And then, determining the operator characteristics of the operator corresponding to the operator nodes as the node characteristics of the operator nodes, determining the client characteristics of the client corresponding to the client nodes as the node characteristics of the client nodes, and completing the construction of the history bipartite graph.

In this embodiment, the salesman and the customer in the historical transaction information are respectively used as a salesman node and a customer node, the salesman node and the customer node with the recommended transaction relationship are connected, and node features are respectively added to the salesman node and the customer node, so that the construction of the historical bipartite graph is completed, and the model training is ready.

Further, after the step S202, the method may further include: calculating the density of the historical bipartite graph; and when the consistency is smaller than a preset consistency threshold value, adjusting the historical bipartite graph according to a preset consistency increasing strategy.

The density reflects the density degree of the historical bipartite graph in a numerical form, and the higher the density is, the more available information in the historical bipartite graph is represented.

Specifically, the consistency has a plurality of calculation modes, and in one embodiment, the total number of nodes and the number of non-isolated points in the history bipartite graph are obtained, and the ratio of the number of non-isolated points to the total number of nodes is used as the consistency. The outlier may be a node in the history bipartite graph that has no connecting edge, that is, the node does not connect any node through a connecting edge, and for a non-outlier, the node is connected to other nodes through at least one connecting edge.

In one embodiment, the maximum number of connection edges which can be established between the server node and the client node is calculated according to the number of the server nodes and the number of the client nodes in the history bipartite graph, then the number of the existing connection edges between the server nodes and the client nodes in the history bipartite graph is counted, and the ratio of the number of the existing connection edges to the maximum number is used as the density.

And acquiring a preset consistency threshold, comparing the calculated consistency with the consistency threshold, and if the consistency is less than the consistency threshold, indicating that the available information in the historical bipartite graph is less. At this time, the historical bipartite graph may be adjusted according to a preset consistency increasing strategy to increase the consistency of the historical bipartite graph. The adjustment of the historical bipartite graph is mainly to adjust the connection edges in the historical bipartite graph, for example, to increase the number of the connection edges.

In this embodiment, the consistency of the historical bipartite graph is calculated, when the consistency is smaller than a consistency threshold, it is indicated that the available information of the historical bipartite graph is less, and the historical bipartite graph is adjusted according to a preset consistency increasing strategy, so that the information in the historical bipartite graph is enriched, and the accuracy of model training is ensured.

Further, the step of adjusting the historical bipartite graph according to a preset consistency increase strategy may include: based on the node characteristics, calculating the similarity of the same type of nodes of the historical bipartite graph; when the similarity of the same type of nodes larger than the similarity threshold exists, connecting two nodes corresponding to the similarity of the same type of nodes through a connecting edge; or, according to the interactive information in the historical transaction information, weak connection edges are established between the operator nodes and the client nodes of the historical bipartite graph; the edge weight of the weak connection edge is lower than the edge weight of the connection edge established based on the recommended intersection relationship; the edge weight participates in calculating the recommended deal probability through the node characteristics.

Specifically, there are two types of nodes in the history bipartite graph, namely, an operator node and a client node, where the two operator nodes are similar nodes and the two client nodes are also similar nodes.

For two nodes belonging to the same type of node, the similarity of the same type of node between the two nodes can be calculated based on the node characteristics, and the larger the similarity of the same type of node is, the more similar the two nodes are. During calculation, respective node features of the two nodes can be arranged and then converted into feature vectors, and the similarity between the two nodes is measured through the similarity between the feature vectors of the two nodes.

The similarity of the same type nodes of any group of the same type nodes can be calculated, and if the similarity of the same type nodes is larger than a preset similarity threshold, two nodes in the group of the same type nodes are connected through a connecting edge.

In another embodiment, interactive information in the historical transaction information is obtained, the interactive information can record chatting and consulting behaviors between an operator and a client, and if the chatting and consulting behaviors between the operator and the client are determined through the interactive information and the product recommendation does not successfully occur, a weak connection edge can be established between an operator node corresponding to the operator and a client node corresponding to the client. The creation of weakly connected edges introduces the concept of edge weights for the connected edges. If the connection edge is established based on the recommended intersection relationship, the edge weight of the connection edge is greater than the edge weight of the weak connection edge. The edge weight may be applied to the node characteristics of the nodes connected by the connection edge, and in the calculation, the edge weight may be given to the node characteristics of the nodes connected by the connection edge, and the lower the edge weight is, the lower the role exerted by the node weight of the nodes connected by the connection edge in the calculation is.

In one embodiment, the connecting edges established according to the similarity of the same type of nodes may also have edge weights, and such edge weights are lower than the edge weights of the connecting edges established based on the recommended intersection relationship.

In this embodiment, the number of the connection edges in the history bipartite graph may be increased, or the edge weight may be added to the connection edges, so as to enrich the information contained in the history bipartite graph.

Further, the step S203 may include: randomly acquiring a one-hop neighbor node and a two-hop neighbor node of a server node through an initial prediction model; generating node embedding representation of the operator node based on the node characteristics of the one-hop neighbor node and the two-hop neighbor node; embedding the nodes into a full-connection layer representing and inputting the initial prediction model to obtain the recommended transaction probability between the operator node and the client node; the client node is a node in a one-hop neighbor node or a two-hop neighbor node.

Specifically, the initial prediction model is constructed based on the GraphSAGE algorithm. When prediction is carried out, aiming at a certain operator node, a one-hop neighbor node and a two-hop neighbor node of the operator node are randomly collected. And generating node embedding representation of the operator node based on the node characteristics of the one-hop neighbor node and the two-hop neighbor node.

In one embodiment, the node characteristics of the two-hop neighbor nodes are aggregated to generate the node embedded representation of the one-hop neighbor node. Node embedding tokens that characterize the node embedding tokens of the compute attendant nodes are then aggregated for one-hop neighbor nodes.

In one embodiment, information may be aggregated for all neighbor nodes of the server node, which is information-lossless.

After the node embedding representation is generated, the node embedding representation is input into a full-connection layer in the initial prediction model, and the full-connection layer outputs recommended deal probabilities between the operator nodes and the client nodes. The client node is a node in a one-hop neighbor node or a two-hop neighbor node when sampling is required, that is, the client node needs to be collected in the sampling process.

In the embodiment, the neighbor nodes of the operator node are collected, the node embedding expression of the operator node is calculated according to the node characteristics, and the node embedding expression is input into the full-connection layer of the initial prediction model to obtain the recommendation deal probability between the operator node and the client node, so that the product recommendation deal probability between the operator and the client is predicted, and the matching degree between the operator and the client is also measured.

Further, the step S205 may include: acquiring bipartite graph updating information; and adjusting the operator node and the corresponding node characteristic thereof, and the client node and the corresponding node characteristic thereof in the historical bipartite graph according to the node updating information and the node characteristic updating information in the bipartite graph updating information to obtain the current bipartite graph.

Specifically, update information of the bipartite graph is acquired, and the update information of the bipartite graph comprises node update information and node feature update information, so that the operator nodes and the node features thereof in the historical bipartite graph can be adjusted to obtain the current bipartite graph. The adjustment to the service provider node may include adding and deleting, and the adjustment to the node characteristic includes adding, deleting and modifying.

When adjusting, a plurality of new operator nodes and customer nodes can be added at one time so as to predict new operators and new customers through the prediction model.

In this embodiment, the nodes and the node characteristics in the historical bipartite graph are adjusted according to the bipartite graph update information, so that the prediction model can predict according to the current bipartite graph.

Further, the step S206 may include: calculating the ranking value of each salesman according to historical transaction information; sequencing all the salesmen according to the sequencing value to obtain a salesmen queue; and for each salesman in the salesman queue, carrying out dispatching processing on the salesman according to the recommended transaction probability of the salesman node corresponding to the salesman.

Specifically, the ranking value of each salesman is calculated according to the historical transaction information, and the ranking value can be the number of times that the salesman successfully recommends the product in the past preset time or the total value of the successfully recommended product. The ranking value may also be calculated based on node characteristics of the attendant nodes.

It is understood that the larger the ranking value, the more times the recommendation of a product by the representative clerk is successful, or the higher the total value of the products that are successfully recommended. And sorting the salesmen in a descending order according to the sorting value to obtain a salesmen queue. And in dispatching working hours, the salesmen in the salesmen queue is selected from front to back, and the salesmen are dispatched according to each recommended transaction probability of the salesmen node corresponding to the salesmen. That is, the higher the ranking value, the higher the priority of the order assignment. When dispatching the order, the service person can autonomously select the client according to the recommended transaction probability or delegate the service person to the client of which the recommended transaction probability is greater than the preset probability value.

In the embodiment, the sequencing value of the salesman is calculated, the salesman is sequenced and dispatched according to the sequencing value, the salesman with a higher sequencing value can delegate the customer preferentially, and the accuracy and the success rate of product recommendation after the dispatching process are improved.

The embodiment of the application can acquire and process related data based on an artificial intelligence technology. Among them, artificial Intelligence (AI) is a theory, method, technique and application system that simulates, extends and expands human Intelligence using a digital computer or a machine controlled by a digital computer, senses the environment, acquires knowledge and uses the knowledge to obtain the best result.

The artificial intelligence infrastructure generally includes technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technologies, operation/interaction systems, mechatronics, and the like. The artificial intelligence software technology mainly comprises a computer vision technology, a robot technology, a biological recognition technology, a voice processing technology, a natural language processing technology, machine learning/deep learning and the like.

It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above may be implemented by hardware that is configured to be instructed by computer-readable instructions, which can be stored in a computer-readable storage medium, and when executed, the programs may include the processes of the embodiments of the methods described above. The storage medium may be a non-volatile storage medium such as a magnetic disk, an optical disk, a Read-Only Memory (ROM), or a Random Access Memory (RAM).

It should be understood that, although the steps in the flowcharts of the figures are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and may be performed in other orders unless explicitly stated herein. Moreover, at least a portion of the steps in the flow chart of the figure may include multiple sub-steps or multiple stages, which are not necessarily performed at the same time, but may be performed at different times, which are not necessarily performed in sequence, but may be performed alternately or alternately with other steps or at least a portion of the sub-steps or stages of other steps.

With further reference to fig. 3, as an implementation of the method shown in fig. 2, the present application provides an embodiment of a dispatching device, which corresponds to the embodiment of the method shown in fig. 2 and can be applied to various electronic devices.

As shown in fig. 3, the dispatching device 300 of the present embodiment includes: an information acquisition module 301, a bipartite graph construction module 302, a probability prediction module 303, a model adjustment module 304, a bipartite graph update module 305, and a deal prediction module 306, wherein:

the information obtaining module 301 is configured to obtain historical transaction information.

A bipartite graph construction module 302, configured to construct a historical bipartite graph according to historical transaction information; the salesman and the client in the historical transaction information respectively correspond to a salesman node and a client node in the historical bipartite graph, and the salesman node and the client node have node characteristics.

A probability prediction module 303, configured to calculate, based on the node characteristics, a recommended deal probability between the operator node and the client node through an initial prediction model; the initial prediction model is constructed based on the GraphSAGE algorithm.

And the model adjusting module 304 is used for calculating model loss according to the connection relation and the recommended transaction probability of the operator node and the client node in the bipartite graph, and adjusting the initial prediction model based on the model loss until the initial prediction model converges to obtain the prediction model.

And a bipartite graph updating module 305, configured to update the historical bipartite graph according to the graph network update information, so as to obtain a current bipartite graph.

And the deal prediction module 306 is used for calculating the recommended deal probability between the operator node to be predicted and the client node to be predicted in the current bipartite graph through a prediction model, and dispatching the operator corresponding to the operator node according to the recommended deal probability.

In the embodiment, historical transaction information is obtained and a historical bipartite graph is constructed, wherein an operator and a client in the historical transaction information respectively correspond to an operator node and a client node in the historical bipartite graph, and the operator node and the client node have node characteristics; based on node characteristics, calculating a recommendation interaction probability between an operator node and a client node through an initial prediction model constructed according to a GraphSAGE algorithm, wherein the size of the probability value represents the possibility of success of recommending products to a client by a predicted operator, and the connection relation between the operator node and the client node can be used as a label, so that model loss can be calculated, and the initial prediction model is adjusted until the prediction model is obtained; when the method is applied, the historical bipartite graph is updated according to graph network updating information to obtain a current bipartite graph, the recommendation deal probability between a to-be-predicted operator node and a to-be-predicted client node in the current bipartite graph can be accurately calculated through a prediction model, and an operator corresponding to the operator node is dispatched according to the recommendation deal probability, so that the conversion rate of product recommendation after the dispatching is considered, the adaptability between the operator and the client is improved, and the accuracy of the dispatching is improved.

In some optional implementations of this embodiment, the bipartite graph building module 302 may include: the system comprises an information identification submodule, a feature extraction submodule, a node determination submodule and a feature determination submodule, wherein:

and the information identification submodule is used for identifying the salesman and the client in the historical transaction information and the recommended transaction relationship between the salesman and the client.

And the characteristic extraction submodule is used for extracting the characteristics of the salesman and the characteristics of the customer from the historical transaction information.

And the node determination submodule is used for respectively taking the operator and the client as an operator node and a client node and connecting the operator node and the client node with a recommended interaction relationship.

And the characteristic determining submodule is used for determining the characteristics of the operator as the node characteristics of the corresponding operator node and determining the characteristics of the client as the node characteristics of the corresponding client node to obtain the history bipartite graph.

In this embodiment, the operator and the client in the historical transaction information are respectively used as an operator node and a client node, the operator node and the client node with the recommended transaction relationship are connected, and node characteristics are respectively added to the operator node and the client node, so that the construction of the historical bipartite graph is completed, and the model training is prepared.

In some optional implementation manners of this embodiment, the dispatching device 300 may further include: the density calculation module and the bipartite graph adjustment module are provided, wherein:

and the density calculating module is used for calculating the density of the history bipartite graph.

And the bipartite graph adjusting module is used for adjusting the historical bipartite graph according to a preset density increasing strategy when the density is smaller than a preset density threshold value.

In this embodiment, the consistency of the historical bipartite graph is calculated, when the consistency is smaller than the consistency threshold, it is indicated that the available information of the historical bipartite graph is less, and the historical bipartite graph is adjusted according to a preset consistency increasing strategy, so that the information in the historical bipartite graph is enriched, and the accuracy of model training is ensured.

In some optional implementations of this embodiment, the bipartite graph adjustment module may include: similarity operator module, node connection submodule or set up the submodule piece, wherein:

and the similarity operator module is used for calculating the similarity of the same type of nodes of the historical bipartite graph based on the node characteristics.

And the node connection submodule is used for connecting two nodes corresponding to the similarity of the same type of nodes through the connection edge when the similarity of the same type of nodes larger than the similarity threshold exists.

The establishing submodule is used for establishing a weak connection edge between an operator node and a client node of the historical bipartite graph according to interaction information in the historical transaction information; the edge weight of the weak connection edge is lower than the edge weight of the connection edge established based on the recommended intersection relationship; the edge weight participates in calculating the recommended deal probability through the node characteristics.

In this embodiment, the number of the connecting edges in the history bipartite graph may be increased, or the edge weight may be added to the connecting edges, so as to enrich the information contained in the history bipartite graph.

In some optional implementations of this embodiment, the probability prediction module 303 may include: the node acquisition submodule, the characterization generation submodule and the probability prediction submodule, wherein:

and the node acquisition submodule is used for randomly acquiring the one-hop neighbor node and the two-hop neighbor node of the operator node through the initial prediction model.

And the characterization generation submodule is used for generating node embedded characterization of the operator node based on the node characteristics of the one-hop neighbor node and the two-hop neighbor node.

The probability prediction submodule is used for embedding the nodes into a full-connection layer representing and inputting the nodes into an initial prediction model to obtain the recommended transaction probability between the operator node and the client node; the client node is a node in a one-hop neighbor node or a two-hop neighbor node.

In some optional implementations of this embodiment, the bipartite graph updating module 305 may include: an update acquisition submodule and a node adjustment submodule, wherein:

and the update acquisition submodule is used for acquiring bipartite graph update information.

And the node adjusting submodule is used for adjusting the operator node and the corresponding node characteristic thereof and the client node and the corresponding node characteristic thereof in the historical bipartite graph according to the node updating information and the node characteristic updating information in the bipartite graph updating information to obtain the current bipartite graph.

In some optional implementations of this embodiment, the deal prediction module 306 may include: the system comprises a sequencing calculation sub-module, a queue generation sub-module and a dispatching processing sub-module, wherein:

and the sequencing calculation submodule is used for calculating the sequencing value of each salesman according to the historical transaction information.

And the queue generation submodule is used for sequencing all the salesmen according to the sequencing value to obtain a salesmen queue.

And the dispatching processing submodule is used for dispatching each salesman in the salesman queue according to the recommended transaction probability of the salesman node corresponding to the salesman.

In the embodiment, the ranking value of the salesman is calculated, the salesman is ranked and dispatched according to the ranking value, the salesman with the higher ranking value can assign customers preferentially, and the accuracy and the success rate of product recommendation after the dispatching process are improved.

In order to solve the technical problem, the embodiment of the application further provides computer equipment. Referring to fig. 4 in particular, fig. 4 is a block diagram of a basic structure of a computer device according to the embodiment.

The computer device 4 comprises a memory 41, a processor 42, a network interface 43 communicatively connected to each other via a system bus. It is noted that only computer device 4 having components 41-43 is shown, but it is understood that not all of the shown components are required to be implemented, and that more or fewer components may be implemented instead. As will be understood by those skilled in the art, the computer device is a device capable of automatically performing numerical calculation and/or information processing according to instructions set or stored in advance, and the hardware thereof includes but is not limited to a microprocessor, an Application Specific Integrated Circuit (ASIC), a Programmable Gate Array (FPGA), a Digital Signal Processor (DSP), an embedded device, and the like.

The computer device can be a desktop computer, a notebook, a palm computer, a cloud server and other computing devices. The computer equipment can carry out man-machine interaction with a user through a keyboard, a mouse, a remote controller, a touch panel or voice control equipment and the like.

The memory 41 includes at least one type of readable storage medium including a flash memory, a hard disk, a multimedia card, a card type memory (e.g., SD or DX memory, etc.), a Random Access Memory (RAM), a Static Random Access Memory (SRAM), a Read Only Memory (ROM), an Electrically Erasable Programmable Read Only Memory (EEPROM), a Programmable Read Only Memory (PROM), a magnetic memory, a magnetic disk, an optical disk, etc. In some embodiments, the memory 41 may be an internal storage unit of the computer device 4, such as a hard disk or a memory of the computer device 4. In other embodiments, the memory 41 may also be an external storage device of the computer device 4, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), and the like, provided on the computer device 4. Of course, the memory 41 may also include both internal and external storage devices of the computer device 4. In this embodiment, the memory 41 is generally used for storing an operating system and various application software installed in the computer device 4, such as computer readable instructions of a dispatching method. Further, the memory 41 may also be used to temporarily store various types of data that have been output or are to be output.

The processor 42 may be a Central Processing Unit (CPU), controller, microcontroller, microprocessor, or other data Processing chip in some embodiments. The processor 42 is typically used to control the overall operation of the computer device 4. In this embodiment, the processor 42 is configured to execute computer readable instructions stored in the memory 41 or process data, for example, execute computer readable instructions of the dispatching method.

The network interface 43 may comprise a wireless network interface or a wired network interface, and the network interface 43 is generally used for establishing communication connection between the computer device 4 and other electronic devices.

The computer device provided in this embodiment may execute the dispatch method. The dispatching method can be the dispatching method of the above embodiments.

In the embodiment, historical transaction information is obtained, a historical bipartite graph is constructed, a salesman and a client in the historical transaction information respectively correspond to a salesman node and a client node in the historical bipartite graph, and the salesman node and the client node have node characteristics; based on node characteristics, calculating a recommendation interaction probability between an operator node and a client node through an initial prediction model constructed according to a GraphSAGE algorithm, wherein the size of the probability value represents the possibility of success of recommending products to a client by a predicted operator, and the connection relation between the operator node and the client node can be used as a label, so that model loss can be calculated, and the initial prediction model is adjusted until the prediction model is obtained; when the method is applied, the historical bipartite graph is updated according to graph network updating information to obtain a current bipartite graph, the recommendation deal probability between a to-be-predicted operator node and a to-be-predicted client node in the current bipartite graph can be accurately calculated through a prediction model, and an operator corresponding to the operator node is dispatched according to the recommendation deal probability, so that the conversion rate of product recommendation after the dispatching is considered, the adaptability between the operator and the client is improved, and the accuracy of the dispatching is improved.

The present application further provides another embodiment, which is to provide a computer-readable storage medium storing computer-readable instructions executable by at least one processor to cause the at least one processor to perform the steps of the dispatch method as described above.

Through the above description of the embodiments, those skilled in the art will clearly understand that the method of the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better implementation manner. Based on such understanding, the technical solutions of the present application may be embodied in the form of a software product, which is stored in a storage medium (such as ROM/RAM, magnetic disk, optical disk) and includes instructions for enabling a terminal device (such as a mobile phone, a computer, a server, an air conditioner, or a network device) to execute the method according to the embodiments of the present application.

It should be understood that the above-described embodiments are merely exemplary of some, and not all, embodiments of the present application, and that the drawings illustrate preferred embodiments of the present application without limiting the scope of the claims appended hereto. This application is capable of embodiments in many different forms and the embodiments are provided so that this disclosure will be thorough and complete. Although the present application has been described in detail with reference to the foregoing embodiments, it will be apparent to one skilled in the art that the present application may be practiced without modification or with equivalents of some of the features described in the foregoing embodiments. All equivalent structures made by using the contents of the specification and the drawings of the present application are directly or indirectly applied to other related technical fields and are within the protection scope of the present application.

Claims

1. A dispatching method is characterized by comprising the following steps:

acquiring historical transaction information;

constructing a historical bipartite graph according to the historical transaction information; the service person and the client in the historical transaction information respectively correspond to a service person node and a client node in the historical bipartite graph, and the service person node and the client node have node characteristics;

calculating a recommended transaction probability between the operator node and the client node through an initial prediction model based on the node characteristics; the initial prediction model is constructed based on GraphSAGE algorithm;

2. The method of dispatching as claimed in claim 1, wherein said step of building a historical bipartite graph from the historical transaction information comprises:

identifying an operator and a customer in the historical transaction information, and a recommended deal relationship between the operator and the customer;

extracting the characteristics of the salesman and the characteristics of the customer from the historical transaction information;

the operator and the client are respectively used as an operator node and a client node, and the operator node and the client node which have a recommended interaction relationship are connected;

and determining the operator characteristics as the node characteristics of the corresponding operator node, and determining the client characteristics as the node characteristics of the corresponding client node to obtain a historical bipartite graph.

3. The method of dispatching as claimed in claim 1, wherein after said step of constructing a historical bipartite graph from the historical transaction information, further comprising:

calculating the density of the historical bipartite graph;

and when the consistency is smaller than a preset consistency threshold value, adjusting the historical bipartite graph according to a preset consistency increasing strategy.

4. The dispatching method according to claim 3, wherein the step of adjusting the historical bipartite graph according to a preset consistency addition strategy comprises:

based on node characteristics, calculating the similarity of the same type of nodes of the historical bipartite graph;

when the similarity of the same type of nodes larger than the similarity threshold exists, connecting two nodes corresponding to the similarity of the same type of nodes through a connecting edge;

alternatively, the first and second electrodes may be,

according to the interaction information in the historical transaction information, weak connection edges are established between the operator nodes and the client nodes of the historical bipartite graph; the edge weight of the weak connection edge is lower than the edge weight of the connection edge established based on the recommended intersection relationship; the edge weight participates in calculating the recommended deal probability through the node characteristics.

5. The method of dispatching as claimed in claim 1, wherein said step of calculating a probability of recommended deal between the server node and the client node through an initial predictive model based on said node characteristics comprises:

randomly collecting a one-hop neighbor node and a two-hop neighbor node of the operator node through the initial prediction model;

generating a node embedding representation of the operator node based on the node characteristics of the one-hop neighbor node and the two-hop neighbor node;

embedding the node into a full-connection layer representing and inputting the node into the initial prediction model to obtain a recommended transaction probability between the operator node and the client node; the client node is a node in the one-hop neighbor node or the two-hop neighbor node.

6. The method of dispatching according to claim 1, wherein said updating said historical bipartite graph according to graph network update information, and obtaining a current bipartite graph comprises:

acquiring bipartite graph updating information;

and adjusting the operator node and the corresponding node characteristic thereof, and the client node and the corresponding node characteristic thereof in the historical bipartite graph according to the node updating information and the node characteristic updating information in the bipartite graph updating information to obtain the current bipartite graph.

7. The dispatching method according to claim 1, wherein the step of dispatching the salesman corresponding to the salesman node according to the recommended deal probability comprises:

calculating the ranking value of each salesman according to the historical transaction information;

sequencing the salesmen according to the sequencing value to obtain a salesmen queue;

and for each salesman in the salesman queue, carrying out dispatching processing on the salesman according to the recommended deal-making probability of the salesman node corresponding to the salesman.

8. A dispatching device is characterized by comprising:

the bipartite graph building module is used for building a historical bipartite graph according to the historical transaction information; the business member and the client in the historical transaction information respectively correspond to a business member node and a client node in the historical bipartite graph, and the business member node and the client node have node characteristics;

the probability prediction module is used for calculating the recommended transaction probability between the operator node and the client node through an initial prediction model based on the node characteristics; the initial prediction model is constructed based on GraphSAGE algorithm;

9. A computer device comprising a memory having computer readable instructions stored therein and a processor which when executed implements the steps of the method of dispatch of any one of claims 1 to 7.

10. A computer readable storage medium having computer readable instructions stored thereon which, when executed by a processor, implement the steps of the method of dispatch of any of claims 1 to 7.