Staff determination method, electronic device and computer program product
By extracting and matching the characteristics of customers and staff, the problem of customer dissatisfaction caused by random allocation is solved, the transaction rate and satisfaction are improved, and the best matching between staff and customers is achieved.
Patent Information
- Application Number
- CN202411201423.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-29
- Publication Date
- 2024-11-26
AI Technical Summary
In the prior art, random allocation of staff causes customers to fail to get a satisfactory experience, staff cannot obtain good performance, and merchants may lose potential customers, so how to determine the staff to connect with customers is crucial.
By extracting the identity information and operational behavior of customers, combining the identity information and results characteristics of staff, the double tower model is used to match the staff to determine the staff most in line with customer preferences.
It improves customer transaction rate and satisfaction, ensures the matching between staff and customers, and improves merchants' performance and customer experience.
Smart Images

Figure CN119026868A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to the field of artificial intelligence technology, and in particular to a staff determination method, electronic equipment, and computer program product. Background Art
[0002] Different staff members may have different working methods and working habits, and different customers may also have different working methods and working habits that are acceptable to different staff members.
[0003] In the related art, the staff members who are assigned to connect with customers are assigned based on the principle of proximity or in a round-robin manner. This will result in the staff members who are assigned to connect with customers being random and may not meet the preferences of customers, which in turn will result in customers not being able to get a good experience, staff members not being able to get good performance, and even merchants being likely to lose potential customers. Therefore, how to determine the staff members who are assigned to connect with customers is crucial. Summary of the invention
[0004] In order to solve one of the above technical problems, the present disclosure provides a worker determination method, an electronic device and a computer program product.
[0005] According to the first aspect of the present disclosure, a staff determination method is provided, including: determining the target characteristics of the customer based on the customer's identity information and the customer's operating behavior on the APP; determining the work result characteristics of each staff member among multiple staff members based on the identity information, work results and operating behavior of successfully connected customers on the APP; and matching the work result characteristics of each staff member with the target characteristics of the customer respectively to determine the staff member connected to the customer.
[0006] According to at least one embodiment of the present disclosure, determining the target characteristics of the customer based on the customer's identity information and the customer's operating behavior on the APP includes: performing feature extraction on the customer's identity information and the customer's operating behavior on the APP to obtain the customer's operating behavior characteristics; and performing feature extraction on the customer's identity information and the customer's operating behavior characteristics to obtain the customer's target characteristics.
[0007] According to at least one embodiment of the present disclosure, the extracting features of the customer's identity information and the customer's operating behavior on the APP to obtain the customer's operating behavior features includes: inputting the customer's identity information and the customer's operating behavior on the APP into the customer tower of the first double-tower model, extracting features of the customer's identity information and the customer's operating behavior on the APP through the customer tower to obtain the customer's operating behavior features, wherein the first double-tower model also includes an article tower, the article tower is a dnn structure, and the customer tower is an mmoe structure; extracting features of the customer's identity information and the customer's operating behavior features to obtain the customer's target features includes: inputting the customer's identity information and the operating behavior features into the customer tower of the second double-tower model, extracting features of the customer's identity information and the operating behavior features through the customer tower to obtain the customer's target features, wherein the second double-tower model also includes a staff tower, the customer tower is a dnn structure, and the staff tower is an mmoe structure.
[0008] According to at least one embodiment of the present disclosure, in response to the fact that the customer has not used the APP and the customer's friend has used the APP, the method further includes: extracting features from the identity information of the customer's friend and the operating behaviors of the customer's friend on the APP to obtain the operating behavior features of the customer's friend; and fusing the operating behavior features of multiple friends of the customer, and using the fused operating behavior features as the operating behavior features of the customer.
[0009] According to at least one embodiment of the present disclosure, in response to the fact that neither the customer nor the customer's friends have used the APP, the method further includes: matching the customer's identity information with the identity information of the person using the APP to obtain the APP user matching the customer; extracting features of the identity information of the APP user matching the customer and his / her operating behavior on the APP to obtain the operating behavior features of the APP user matching the customer; and fusing the operating behavior features of the APP user matching the customer, and using the fused operating behavior features as the operating behavior features of the customer.
[0010] According to at least one embodiment of the present disclosure, the personnel relationship network includes APP users and friends of the APP users, and the friends of the APP users include at least one of persons who have used the APP and persons who have not used the APP, and the method further includes: matching the identity information of the customer with the identity information of the persons in the personnel relationship network; if the identity information of a person who has used the APP in the personnel relationship network is the same as the identity information of the customer, it is determined that the customer has used the APP; if the identity information of a person who has not used the APP in the personnel relationship network is the same as the identity information of the customer, it is determined that the customer has not used the APP and the customer's friends have used the APP; if there is no person in the personnel relationship network whose identity information is the same as the identity information of the customer, it is determined that neither the customer nor the customer's friends have used the APP.
[0011] According to at least one embodiment of the present disclosure, the work result characteristics of each staff member among the multiple staff members are determined based on the identity information, work results and operational behaviors of successfully connected customers on the APP of the staff member, including: extracting features from the identity information of the successfully connected customers and the operational behaviors of the successfully connected customers on the APP to obtain the operational behavior characteristics of the successfully connected customers; and extracting features from the identity information, work results and operational behavior characteristics of the successfully connected customers of the multiple staff members to obtain the work result characteristics of each staff member.
[0012] According to at least one embodiment of the present disclosure, the feature extraction of the identity information of the successfully connected customer and the operation behavior of the successfully connected customer on the APP to obtain the operation behavior characteristics of the successfully connected customer includes: inputting the identity information of the successfully connected customer and the operation behavior of the successfully connected customer on the APP into the customer tower of the first double-tower model, and extracting the identity information of the successfully connected customer and the operation behavior of the successfully connected customer on the APP through the customer tower to obtain the operation behavior characteristics of the successfully connected customer, wherein the first double-tower model also includes an article tower, the article tower is a dnn structure, and the customer tower is an mmoe structure structure; said feature extraction of the identity information, work results and operational behavior characteristics of the successfully connected customers of each of the multiple staff members to obtain the work result characteristics of each staff member includes: inputting the identity information, work results and operational behavior characteristics of the successfully served customers of each of the multiple staff members into the staff tower of the second double-tower model, and extracting the identity information, work results and operational behavior characteristics of the successfully connected customers of each staff member through the staff tower to obtain the work result characteristics of each staff member, wherein the second double-tower model also includes a customer tower, the customer tower is a dnn structure, and the staff tower is a mmoe structure.
[0013] According to at least one embodiment of the present disclosure, the work result characteristics include transaction characteristics, test drive characteristics and satisfaction characteristics, and the work result characteristics of each staff member are matched with the target characteristics of the customer to determine the staff member to be connected with the customer, including: performing dot product of the transaction characteristics of each staff member with the target characteristics of the customer to obtain a first dot product value corresponding to each staff member; performing dot product of the test drive characteristics of each staff member with the target characteristics of the customer to obtain a second dot product value corresponding to each staff member; performing dot product of the satisfaction characteristics of each staff member with the target characteristics of the customer to obtain a third dot product value corresponding to each staff member; performing weighted summation of the first dot product value, the second dot product value and the third dot product value corresponding to each staff member to obtain the similarity between each staff member and the customer; and selecting the staff member with the greatest similarity as the staff member to be connected with the customer.
[0014] According to at least one embodiment of the present disclosure, it also includes: in response to the customer having not used the APP and the customer's friend having used the APP, taking the operating behavior of the customer's friend on the APP as the operating behavior of the customer on the APP; in response to the customer and the customer's friend having not used the APP, matching the customer's identity information with the identity information of the user of the APP to obtain the APP user matching the customer, and taking the operating behavior of the APP user matching the customer on the APP as the operating behavior of the customer on the APP.
[0015] According to a second aspect of the present disclosure, there is provided an electronic device, including a processor and a memory, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the following processes: determining the target characteristics of the customer based on the customer's identity information and the customer's operating behavior on the APP; determining the work result characteristics of each staff member among a plurality of staff members based on the identity information, work results and operating behavior of the successfully connected customer on the APP; and matching the work result characteristics of each staff member with the target characteristics of the customer respectively to determine the staff member to be connected with the customer.
[0016] According to a third aspect of the present disclosure, a computer program product is provided. The computer program product includes a computer program. When the computer program is executed by a processor, it is used to implement at least the method described in the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] The accompanying drawings illustrate exemplary embodiments of the present disclosure and together with the description serve to explain the principles of the present disclosure. These drawings are included to provide a further understanding of the present disclosure and are incorporated in and constitute a part of this specification.
[0018] Figure 1 is a schematic diagram of a staff determination system according to an embodiment of the present disclosure.
[0019] Figure 2 It is an interactive schematic diagram of a staff determination system according to an embodiment of the present disclosure.
[0020] Figure 3 It is a schematic diagram of the process of training the first double-tower model according to an embodiment of the present disclosure.
[0021] Figure 4 The diagram is a visualization example of a personnel relationship network according to an embodiment of the present disclosure.
[0022] Figure 5It is a schematic diagram of the process of training the second double-tower model according to an embodiment of the present disclosure.
[0023] Figure 6 It is a schematic diagram of a process of determining target characteristics of a customer according to an embodiment of the present disclosure.
[0024] Figure 7 is a schematic diagram of a process for determining target characteristics of a customer according to another embodiment of the present disclosure.
[0025] Figure 8 It is a flowchart of a method for determining a staff member according to an embodiment of the present disclosure.
[0026] Fig. 9 It is a flowchart of a staff determination method according to another embodiment of the present disclosure.
[0027] Fig.10 It is a schematic block diagram of the structure of an electronic device using a hardware implementation of a processor according to an embodiment of the present disclosure. DETAILED DESCRIPTION
[0028] The present disclosure is further described in detail below in conjunction with the accompanying drawings and examples. It is understood that the specific examples described herein are only used to explain the relevant content, rather than to limit the present disclosure. It should also be noted that, for ease of description, only the parts related to the present disclosure are shown in the accompanying drawings.
[0029] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in the present disclosure can be combined with each other. The technical solution of the present disclosure will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.
[0030] In the related art, since the staff members are randomly assigned to customers, and the work habits and working methods of the randomly assigned staff members may not meet the customer's preferences, the customer's transaction rate will be uncontrollable, or even lowered. In order to improve the customer's transaction rate, in the present disclosure, the staff members who are connected to the customer are determined according to the matching degree between the mined customer target characteristics and the staff members' work results characteristics, so that the staff members who are connected to the customer are more in line with the customer's preferences and improve the customer's transaction rate. To this end, the following technical solution is proposed in the present disclosure.
[0031] In the present disclosure, terminal devices include but are not limited to mobile phones, tablet computers, laptop computers, personal computers, wearable devices, ATMs, etc. In addition, the server can be a server with a physical form or a cloud server, and the present disclosure does not limit the type of server.
[0032] Figure 1A schematic diagram of a staff determination system according to an embodiment of the present disclosure is shown. In the staff determination system, a first terminal device 100, a second terminal device 200 and a server 300 may be provided, and the server 300 can communicate with the first terminal device 100 and the second terminal device 200, respectively, so as to perform data transmission or instruction transmission. In the present disclosure, the first terminal device 100, the second terminal device 200 and the server 300 each include at least one processor and at least one memory.
[0033] Exemplarily, the first terminal device 100 may be a terminal device held by a customer, and the second terminal device 200 may be a terminal device held by a staff member or a merchant, and the staff member may be a salesperson, etc. The server 300 can collect information or data related to the customer through the first terminal device 100, and collect information or data related to the staff member through the second terminal device 200, and then determine the staff member to be connected to the customer based on the collected information or data, and prompt the determined staff member to be connected to the customer through the corresponding second terminal device 200, so that the staff member can connect with the customer in time and provide services to the customer.
[0034] Figure 2 An interactive schematic diagram of a staff determination system according to an embodiment of the present disclosure is shown. Figure 2 The detailed description is for a clearer understanding of the technical solution of the present disclosure and should not be considered as a limitation of the protection scope of the present disclosure. In the process of implementing the technical idea of the present disclosure, one or some steps may be omitted, or other alternatives may be adopted.
[0035] In step S201, multiple first terminal devices collect APP user data, and the APP user data includes the identity information of the APP user, the operation behavior of the APP user on the APP, the operation object data of the APP user on the APP, etc. The first terminal device may be installed with an APP, and the APP may be an APP with a shopping function, or a social APP, a reading APP, or a search engine without a shopping function, etc., which is not limited here. The identity information of the APP user can be obtained when he registers the APP. The operation behavior and operation object data of the APP user on the APP can be obtained when he uses the APP. The identity information of the APP user includes but is not limited to the personnel ID, age, gender, and the location of the personnel. The operation behavior of the APP user on the APP includes but is not limited to the number of days of registration, the number of active days, the list of historical clicked articles, the total APP browsing time, the average daily APP browsing time, the number of clicks in the APP (like, forward, comment, follow, etc.), and the average number of APP clicks per day (like, forward, comment, follow, etc.). The operation object may include articles, pictures, videos, etc. Taking the operation object as an example, the operation object data includes but is not limited to the article ID, article category, article keywords and article operation behavior. The article operation behavior includes but is not limited to the number of days the article is online, the total number of clicks (likes, forwarding, comments, attention, etc.), the average number of clicks per day (likes, forwarding, comments, attention, etc.), etc. The following describes the staff determination method of the present disclosure by taking the operation object as an article as an example.
[0036] The first terminal device may be configured with a communication device or an I / O interface. In step S202, the first terminal device sends the collected customer data to the server via its own communication device or I / O interface.
[0037] In step S203, the server receives the APP user data sent by the multiple first terminal devices, and trains the first dual-tower model according to the APP user data from the multiple first terminal devices. The generalized APP user operation behavior features (embedding features) output by the trained first dual-tower model can better characterize the APP users.
[0038] Figure 3 FIG. 1 is a schematic diagram of a process of training a first dual-tower model according to an embodiment of the present disclosure. The process of training the first dual-tower model may be executed by a server. Figure 3 As shown, the process of training the first dual-tower model may include step S301, step S302, step S303, step S304, step S305 and step S306.
[0039] S301: Determine a first training set, the first training set includes identity information of APP users, operation behaviors on APP, operation object data, and multiple operation behavior annotation values for operation behaviors. The server can integrate the APP user data received from multiple first terminal devices to obtain the first training set. Exemplarily, multiple operation behavior annotation values may include click annotation values, browsing time annotation values, like annotation values, forwarding annotation values, comment annotation values, and attention annotation values. For an operation object, if the APP user clicks on the operation object, the click annotation value is 1, otherwise it is 0; if the APP user likes the operation object, the like annotation value is 1, otherwise it is 0; if the APP user forwards the operation object, the forwarding annotation value is 1, otherwise it is 0; if the APP user comments on the operation object, the comment annotation value is 1, otherwise it is 0; if the APP user follows the operation object, the attention annotation value is 1, otherwise it is 0.
[0040] S302: Construct a first dual-tower model, the first dual-tower model includes a user tower and an article tower. The user tower and the article tower may be structures of the same type or different types, which are not limited here. For example, considering that the operation behaviors of APP users are more diverse, and as the operation behaviors of APP users increase, the changes are relatively large, and the mmoe (Mixture of Expert) structure can capture these changes more sensitively, so as to better represent the different interests of APP users in terms of clicks, browsing time, likes, forwarding, comments, and attention, and depict the interests and hobbies of APP users from multiple angles. Therefore, the user tower adopts the mmoe structure, and the number of operation behavior features output by the user tower is the same as the number of the first training targets. The first training target, that is, the operation behavior of the APP user, may include one or more of clicks, browsing time, likes, forwarding, comments, and attention. At the same time, considering that the content of the operation object data is relatively small, the dnn (Deep Neural Network) structure can be used directly to extract the relevant features of the operation object. Therefore, the article tower adopts the dnn structure, and the number of article features output by the article tower is one. In this way, the first dual-tower model constructed using the MMoe structure and the DNN structure has the advantages of simplified structure and better training effect. It is understandable that if both the user tower and the article tower adopt the MMoe structure, the structure of the first dual-tower model will be too complicated, reducing the training efficiency; if both the user tower and the article tower adopt the DNN structure, the structure of the first dual-tower model will be too simple, reducing the accuracy of the trained model.
[0041] S303: Input the identity information of the APP user and the operation behavior of the APP user on the APP into the user tower, and use the user tower to extract features of the identity information of the APP user and the operation behavior of the APP user on the APP to obtain multiple operation behavior features. The dense features of the identity information of the APP user and the operation behavior of the APP user on the APP can be directly input into the user tower, and the sparse features of the identity information of the APP user and the operation behavior of the APP user on the APP can be input into the user tower after sparse processing. Dense features refer to features with numerical information such as registration days, active days, and browsing time. Sparse features refer to features other than dense features, such as ID-type features such as personnel ID and article ID, which can also be obtained by sparse processing of dense features. Sparse processing can be achieved through related technologies such as unique hot encoding, category embedding, and hashing techniques. In one example, the dense features of the identity information of an APP user and the operating behavior of the APP user on the APP are spliced together with the sparse features of the identity information of the APP user and the operating behavior of the APP user on the APP after sparse processing to obtain a first spliced feature, and then the first spliced feature is input into the user tower for feature extraction to obtain multiple operating behavior features.
[0042] S304: Input the operation object data of the APP user on the APP into the article tower, and use the article tower to extract features of the operation object data of the APP user on the APP to obtain an article feature. The dense features in the operation object data can be directly input into the article tower, and the sparse features in the operation object data can be input into the article tower after sparse processing. In one example, the dense features in the operation object data and the sparse features in the operation object data after sparse processing are spliced together to obtain a second spliced feature, and then the second spliced feature is input into the article tower for feature extraction to obtain an article feature.
[0043] S305: Multiply each of the multiple operation behavior features with an article feature to obtain multiple operation behavior prediction values. Exemplarily, the multiple operation behavior features include click features, browsing time features, like features, forwarding features, comment features, and attention features. The click feature is multiplied with the article feature to obtain a click prediction value, the browsing time feature is multiplied with the article feature to obtain a browsing time prediction value, the like feature is multiplied with the article feature to obtain a like prediction value, the forwarding feature is multiplied with the article feature to obtain a forwarding prediction value, the comment feature is multiplied with the article feature to obtain a comment prediction value, and the attention feature is multiplied with the article feature to obtain an attention prediction value.
[0044] S306: Adjust the parameters of the first twin-tower model according to multiple operation behavior prediction values and multiple operation behavior annotation values to obtain the trained first twin-tower model. Specifically, multiple operation behavior prediction values and multiple operation behavior annotation values can be substituted into the first cross entropy loss function to calculate the first loss value, and then the parameters of the first twin-tower model can be adjusted in the direction of reducing the first loss value, and when the first stop condition is reached, the training is stopped to obtain the trained first twin-tower model. The first stop condition may include that the number of training times reaches the first number or the first loss value tends to be stable. In this way, the user tower in the trained first twin-tower model can be used to extract features of the operation behavior of APP users on the APP and the identity information of the APP users, and the potential behavior of APP users can be fully explored, and then the operation behavior characteristics of APP users can be obtained. The operation behavior characteristics of APP users can characterize the hidden meanings under a large number of operation behaviors. In some embodiments, the trained first twin-tower model can be updated regularly to improve the accuracy of the first twin-tower model.
[0045] In step S204, the server generates a personnel relationship network based on one or more of the friend list and follow list of each APP user in the APP, the friend list and follow list of each APP user on the social software, and the mobile phone address book of each APP user. Exemplarily, the people in the friend list and follow list of the APP user and the mobile phone address book can be regarded as the friends of the APP user. The personnel relationship network includes multiple APP users and multiple friends of APP users. The friends of the APP user can be people who have used the APP or people who have not used the APP, and the APP user may not have any friends. Figure 4 is a visualization example diagram of a personnel relationship network according to an embodiment of the present disclosure, such as Figure 4As shown, the personnel relationship network can be in the form of an undirected graph, and every two people with a friend relationship can be connected by a line. Two people directly connected by a line are friends, and two people not directly connected by a line are not friends. In this way, it is convenient to determine the APP usage of the customer and his friends according to the personnel relationship network, and select a corresponding method to determine the customer's operation behavior characteristics according to the APP usage of the customer and his friends. Exemplarily, the customer's identity information is matched with the identity information of the personnel in the personnel relationship network; if the identity information of the person who has used the APP in the personnel relationship network is the same as the identity information of the customer, it is determined that the customer has used the APP; if the identity information of the person who has not used the APP in the personnel relationship network is the same as the identity information of the customer, it is determined that the customer has not used the APP and the customer's friend has used the APP; if the identity information of the person who does not use the APP in the personnel relationship network is the same as the identity information of the customer, it is determined that the customer has not used the APP and the customer's friend has used the APP; if the identity information of the person who does not use the APP in the personnel relationship network is the same as the identity information of the customer, it is determined that neither the customer nor the customer's friend has used the APP. In some embodiments, the server automatically updates the personnel relationship network when it detects the emergence of a new APP user or reaches the target update time. In this way, the automatic update of the personnel relationship network can be achieved, which is conducive to accurately judging the APP usage of the customer and his friends.
[0046] In step S205, multiple second terminal devices collect staff data, which includes staff identity information, work results, and identity information of successfully connected customers. The staff identity information includes, but is not limited to, gender, age, etc. Work results include, but are not limited to, time of employment, number of transactions, number of receptions, number of test drives, satisfaction, etc. The identity information of successfully connected customers includes, but is not limited to, the identity information of historical reception customers, the identity information of historical transaction customers, the identity information of historical test drive customers, etc.
[0047] The second terminal device may be configured with a communication device or an I / O interface. In step S206, the second terminal device sends the collected staff data to the server via its own communication device or I / O interface.
[0048] In step S207, the server receives the staff data sent by the plurality of second terminal devices, and trains the second dual-tower model according to the staff data from the plurality of second terminal devices. Figure 5 FIG. 1 is a schematic diagram of a process of training a second dual-tower model according to an embodiment of the present disclosure. The process of training the second dual-tower model may be executed by a server. Figure 5 As shown, the process of training the second dual-tower model may include step S501, step S502, step S503, step S504, step S505 and step S506.
[0049] S501: Determine the second training set, which includes the staff's identity information, work results, operational behavior characteristics of successfully connected customers, identity information of successfully connected customers, and multiple work result annotation values. The operational behavior characteristics of successfully connected customers can be determined based on the APP usage of the successfully connected customers and their friends. In the case that the successfully connected customer has used the APP, the identity information of the successfully connected customer and the customer's operational behavior on the APP can be input into the user tower of the first dual-tower model, and the identity information of the successfully connected customer and the operational behavior of the successfully connected customer on the APP are extracted through the user tower to obtain the operational behavior characteristics of the successfully connected customer; in the case that the successfully connected customer has not used the APP but the customer's friends have used the APP, the identity information of the successfully connected customer's friends and the operational behavior of the successfully connected customer's friends on the APP can be extracted based on the user tower to obtain the operational behavior characteristics of the successfully connected customer's friends, and the successfully connected customer's friends can be extracted. The operation behavior characteristics of multiple friends of the successfully connected customer are integrated, and the integrated operation behavior characteristics are used as the operation behavior characteristics of the successfully connected customer; in the case that the successfully connected customer and his friends have not used the APP, the identity information of the successfully connected customer can be matched with the identity information of the APP user to obtain the APP user matching the successfully connected customer, and based on the user tower, the identity information of the APP user matching the successfully connected customer and their operation behavior on the APP are extracted to obtain the operation behavior characteristics of the APP user matching the successfully connected customer, and the operation behavior characteristics of the APP user matching the successfully connected customer are integrated, and the integrated operation behavior characteristics are used as the operation behavior characteristics of the successfully connected customer. The identity information of the successfully connected customer may include the customer's location, age, gender, etc. Exemplarily, multiple work result marking values may include transaction marking value, test drive marking value and satisfaction marking value. For a customer contacted by a staff member, if the customer finally successfully places an order, the transaction marking value is 1, otherwise it is 0; if the customer participates in a test drive, the test drive marking value is 1, otherwise it is 0; if the customer is satisfied with the staff member's evaluation, the satisfaction marking value is 1, otherwise it is 0.
[0050] S502: Construct a second dual-tower model, the second dual-tower model includes a customer tower and a staff tower. The customer tower and the staff tower may be of the same type of structure or of different types of structures, which are not limited here. Exemplarily, the customer tower adopts a DNN structure, and the number of target features output by the customer tower is one; the staff tower adopts an MMOE structure, and the number of work result features output by the staff tower is the same as the number of the second training target, and the second training target may include one or more of test drive, transaction, and satisfaction. In this way, the second dual-tower model constructed using the MMOE structure and the DNN structure has the advantages of simplified structure and better training effect. It can be understood that if both the customer tower and the staff tower adopt the MMOE structure, the structure of the second dual-tower model will be too complicated, reducing the training efficiency; if both the customer tower and the staff tower adopt the DNN structure, the structure of the second dual-tower model will be too simple, reducing the accuracy of the trained model.
[0051] S503: Input the identity information and operation behavior characteristics of the successfully connected customer into the customer tower, and use the customer tower to extract the identity information and operation behavior characteristics of the successfully connected customer to obtain a target feature. In some embodiments, the identity information of the successfully connected customer is sparsely processed to make a single hot feature, and the single hot feature is spliced with the operation behavior characteristics of the successfully connected customer to obtain a third spliced feature, and then the third spliced feature is input into the customer tower for feature extraction to obtain a target feature.
[0052] S504: Input the staff member's identity information, work results, and successfully connected customer operation behavior characteristics into the staff member tower, and use the staff member tower to extract the staff member's identity information, work results, and successfully connected customer operation behavior characteristics to obtain multiple work result characteristics. In some embodiments, the staff member's identity information, work results, and successfully connected customer operation behavior characteristics are preprocessed, and the preprocessed staff member's identity information, work results, and successfully connected customer operation behavior characteristics are spliced together to obtain a fourth splicing feature, and then the fourth splicing feature is input into the staff member tower for feature extraction to obtain multiple work result characteristics.
[0053] S505: Perform dot multiplication of each of the multiple work result features with a target feature to obtain multiple work result prediction values. For example, the multiple work result features include a transaction feature, a test drive feature, and a satisfaction feature. The transaction feature is dot multiplied with the target feature to obtain a transaction prediction value, the test drive feature is dot multiplied with the target feature to obtain a test drive prediction value, and the satisfaction feature is dot multiplied with the target feature to obtain a satisfaction prediction value.
[0054] S506: According to a plurality of work result prediction values and a plurality of work result annotation values, the parameters of the second twin-tower model are adjusted to obtain the trained second twin-tower model. The plurality of work result prediction values and the plurality of work result annotation values can be substituted into the second cross entropy loss function to calculate the second loss value, and then the parameters of the second twin-tower model are adjusted in the direction of reducing the second loss value, and when the second stop condition is reached, the training is stopped to obtain the trained second twin-tower model. The second stop condition may include that the number of training times reaches the second number or the second loss value tends to be stable. In this way, the customer tower in the trained second twin-tower model can extract the customer's operation behavior characteristics and identity information, and then obtain the customer's target characteristics; the staff tower in the trained second twin-tower model can extract the staff's identity information, work results and successfully connected customer operation behavior characteristics, and then obtain the work result characteristics. In some embodiments, the trained second twin-tower model can be updated regularly to improve the accuracy of the second twin-tower model.
[0055] In step S208, the second terminal device sends a staff confirmation request to the server. Exemplarily, when the second terminal device detects that a customer has entered a target area (such as a sales store), it can prompt the staff or customer in the target area to enter the customer's identity information, or the second terminal device automatically obtains the customer's identity information, and then the second terminal device sends a staff confirmation request to the server, and the staff confirmation request may carry the customer's identity information. The customer's identity information includes but is not limited to the person ID, age, gender, and the person's location.
[0056] In step S209, after receiving the staff confirmation request, the server determines the APP usage of the customer and his friends based on the customer's identity information and the personnel relationship network. The APP can be an APP provided by the enterprise to which the sales store belongs. The personnel relationship network includes APP users and APP users' friends. The APP users' friends include at least one of a person who has used the APP and a person who has not used the APP. Specifically, the server matches the customer's identity information with the identity information of the people in the personnel relationship network; if the identity information of a person who has used the APP in the personnel relationship network is the same as the customer's identity information, it is determined that the customer has used the APP; if the identity information of a person who has not used the APP in the personnel relationship network is the same as the customer's identity information, it is determined that the customer has not used the APP and the customer's friends have ... neither the customer nor the customer's friends have used the APP.
[0057] In step S210, the server selects a corresponding method to determine the customer's target characteristics based on the APP usage of the customer and his friends, thereby ensuring that the customer's target characteristics can be successfully determined for different APP usage situations and ensuring the smooth execution of the staff's determination method.
[0058] Please combine Figure 6 In one example, when a customer has used the APP, the server determines the customer's target features based on the customer's identity information and the customer's operating behaviors on the APP; when the customer has not used the APP and the customer's friends have used the APP, the server uses the customer's friends' operating behaviors on the APP as the customer's operating behaviors on the APP, and then determines the customer's target features based on the customer's identity information and the customer's operating behaviors on the APP; when neither the customer nor the customer's friends have used the APP, the server matches the customer's identity information with the identity information of the APP user to obtain the APP user that matches the customer, and uses the operating behaviors of the APP user that matches the customer as the customer's operating behaviors on the APP, and then determines the customer's target features based on the customer's identity information and the customer's operating behaviors on the APP. In this way, for different APP usage situations, the customer's operating behaviors on the APP can be successfully obtained, thereby ensuring that the customer's target features can be determined later and ensuring the smooth execution of the staff's determination method.
[0059] Please combine Figure 7In another example, when the customer has used the APP, the server extracts features from the customer's identity information and the customer's operating behavior on the APP to obtain the customer's operating behavior features, and then extracts features from the customer's identity information and the customer's operating behavior features to obtain the customer's target features. In this way, the customer's target features can be accurately determined, which can ensure the smooth execution of the staff's determination method. When the customer has not used the APP and the customer's friends have used the APP, the server extracts features from the customer's friends' identity information and the customer's friends' operating behavior on the APP to obtain the customer's friends' operating behavior features, and fuses the operating behavior features of multiple friends of the customer, and uses the fused operating behavior features as the customer's operating behavior features, and then extracts features from the customer's identity information and the customer's operating behavior features to obtain the customer's target features. In this way, the customer's operating behavior features are represented by the customer's friends' operating behavior features, which can subsequently more accurately determine the customer's target features and ensure the smooth execution of the staff's determination method. When the customer and In the case that none of the customer's friends have used the APP, the server matches the customer's identity information with the identity information of the APP user (for example, matching one or more of age, gender, and location of the person), obtains the APP user matching the customer, and extracts features from the identity information of the APP user matching the customer and their operation behavior on the APP, obtains the operation behavior features of the APP user matching the customer, and fuses the operation behavior features of the APP user matching the customer, and uses the fused operation behavior features as the customer's operation behavior features, and then extracts features from the customer's identity information and the customer's operation behavior features, and obtains the customer's target features. In this way, the missing features of the customer are mined by using the downward generalization feature method, and the customer's operation behavior features are represented by the operation behavior features of the APP user matching the customer. The target features of the customer can be determined more accurately in the future, ensuring the smooth execution of the staff determination method, and solving the problem of not being able to match staff due to missing information / data. It can be understood that in the relevant technology, although the relevant features of the customer are portrayed based on the customer portrait, the customer portrait is portrayed based on existing explicit information, such as gender, age, height, weight, income, etc.In the present disclosure, the customer's own or the customer's friends or the APP user's identity information and operation behavior on the APP can be used to comprehensively characterize the customer's operation behavior characteristics from the perspectives of APP operation behavior, personnel relationship network or downward generalized feature dimensions. Characterizing customers through operation behavior characteristics can not only well express customers with rich information, but also well express customers with only a small amount of information. For example, some customers only leave their phone number, address and gender. The customer portraits in the relevant technology cannot characterize these customers, but the method disclosed in the present disclosure can accurately and comprehensively characterize these customers, thereby improving the accuracy of matching staff. It should be pointed out that in this example, the user tower of the first dual-tower model trained in step S203 can extract the identity information of the person and the operation behavior on the APP to obtain the operation behavior characteristics, and the customer tower of the second dual-tower model trained in step S207 can extract the identity information of the person and the operation behavior characteristics to obtain the target characteristics. For the sake of brevity, it will not be repeated here.
[0060] In step S211, the server determines the work result characteristics of each staff member according to the identity information, work results and operation behavior of the successfully connected customers on the APP of each staff member in the target area. Exemplarily, feature extraction is performed on the identity information of the successfully connected customers and the operation behavior of the successfully connected customers on the APP to obtain the operation behavior characteristics of the successfully connected customers; and feature extraction is performed on the identity information, work results and operation behavior characteristics of the successfully connected customers of each staff member in the multiple staff members to obtain the work result characteristics of each staff member. In this way, the operation behavior characteristics of the customers successfully connected by the staff are used as part of the work result characteristics to construct the work result characteristics, so as to achieve the effect of accurate characterization of the work result characteristics, which is conducive to the subsequent accurate matching of the staff members connected to the customers. The identity information of the successfully connected customers and the operation behavior of the successfully connected customers on the APP can be input into the user tower of the first double-tower model trained in step S203, and the identity information of the successfully connected customers and the operation behavior of the successfully connected customers on the APP are extracted through the user tower to obtain the operation behavior characteristics of the successfully connected customers. The number of customers successfully connected by the staff may be multiple, and the operational behavior features of the customers successfully connected by multiple staff can be averaged and pooled to obtain the third average pooled feature, which is used as the operational behavior features of the customers successfully connected by the staff when participating in the subsequent feature extraction. Further, the identity information, work results and operational behavior features of the customers successfully served by each staff member among the multiple staff members can be input into the staff tower of the second double-tower model, and the identity information, work results and operational behavior features of the successfully connected customers of each staff member are extracted through the staff tower to obtain the work result features of each staff member. In addition, considering that the number of staff members is much smaller than the number of customers, and the staff members are usually fixed and have little change, after obtaining the work result features, a mapping relationship between the work result features and the identity information of the staff members can be established, and the mapping relationship can be stored, and then the next time the work result features are needed, the corresponding work result features can be directly found from the pre-stored mapping relationship according to the identity information of the staff members, thereby improving the efficiency of staff identification.
[0061] In step S212, the server matches the work result characteristics of each staff member with the target characteristics of the customer, and determines them as the staff member to be connected with the customer. Exemplarily, the work result characteristics include transaction characteristics, test drive characteristics, and satisfaction characteristics. Accordingly, step S212 may include the following steps: dot-multiplying the transaction characteristics of each staff member with the target characteristics of the customer, and obtaining the first dot-multiply value corresponding to each staff member; dot-multiplying the test drive characteristics of each staff member with the target characteristics of the customer, and obtaining the second dot-multiply value corresponding to each staff member; dot-multiplying the satisfaction characteristics of each staff member with the target characteristics of the customer, and obtaining the third dot-multiply value corresponding to each staff member; weighted summing the first dot-multiply value, the second dot-multiply value, and the third dot-multiply value corresponding to each staff member, and obtaining the similarity between each staff member and the customer; and using the staff member with the greatest similarity as the staff member to be connected with the customer. In this way, the staff member and the customer can be matched more accurately based on the similarity between the work result characteristics and the target characteristics, and the accuracy of the staff member determined to be connected with the customer is improved, thereby taking into account the improvement of satisfaction, test drive rate, and transaction rate. Transaction features, test drive features, and satisfaction features may correspond to different weights, and the weights may be pre-set by relevant personnel and are not limited here. The calculation formula for the similarity score is: score=k1 * (user_emb, sale_emb) + k2 * (user_emb, try_emb) + k3 * (user_emb, satisfy_emb), where user_emb represents the target feature, sale_emb represents the transaction feature, k1 is the weight corresponding to the transaction feature, try_emb represents the test drive feature, k2 is the weight corresponding to the test drive feature, satisfy_emb represents the satisfaction feature, k3 is the weight corresponding to the satisfaction feature, (user_emb, sale_emb) represents the dot product of the target feature and the transaction feature, (user_emb, try_emb) represents the dot product of the target feature and the test drive feature, and (user_emb, satisfy_emb) represents the dot product of the target feature and the satisfaction feature. Exemplarily, in the case where the customer has used the APP, the process of determining the staff member who will connect with the customer is as follows: Figure 8 shown.
[0062] In step S213, the server sends the determined identification of the staff member who is to connect with the customer to the second terminal device, so that the user of the second terminal device can arrange the corresponding staff member to connect with the customer according to the identification of the staff member who is to connect with the customer received by the second terminal device. In other embodiments, the server can also send the determined identification of the staff member who is to connect with the customer to the first terminal device, so that the customer can actively contact the corresponding staff member according to the identification of the staff member who is to connect with the customer received by the first terminal device.
[0063] In the present disclosure, through the coordinated use of the first dual-tower model and the second dual-tower model, the customer's operating behavior on the APP can be introduced into the scenario of matching staff, so as to achieve the effect of quickly and accurately determining the staff to be connected to the customer, and ultimately allow the customer to obtain a better experience and improve the transaction rate.
[0064] According to a further embodiment of the present disclosure, a staff determination method is also provided. Fig. 9 A flow chart of a staff determination method according to the present disclosure is shown. Fig. 9 As shown, the staff determination method may include step S910, step S920 and step S930.
[0065] S910: Determine the target characteristics of the customer based on the customer's identity information and the customer's operation behavior on the APP.
[0066] S920: Determine the work result characteristics of each staff member according to the identity information, work results and operation behaviors of successfully connected customers on the APP of each staff member among the multiple staff members.
[0067] S930: Match the work performance characteristics of each staff member with the target characteristics of the customer, and determine the staff member to be connected with the customer.
[0068] The staff determination method of the disclosed implementation mode can determine the target characteristics of the customer based on the customer's identity information and the customer's operating behavior on the APP, and determine the work result characteristics of each staff member based on the identity information, work results and operating behavior of successfully connected customers on the APP of each staff member among multiple staff members, and then match the work result characteristics of each staff member with the customer's target characteristics respectively, and determine the staff member who connects with the customer, so that the staff who connects with the customer is more in line with the customer's preferences, thereby improving the customer's transaction rate.
[0069] Step S910 may include: S911, extracting features from the customer's identity information and the customer's operation behavior on the APP to obtain the customer's operation behavior features; and S912, extracting features from the customer's identity information and the customer's operation behavior features to obtain the customer's target features. Further, step S911 may include: inputting the customer's identity information and the customer's operation behavior on the APP into the user tower of the first double-tower model, extracting features from the customer's identity information and the customer's operation behavior on the APP through the user tower, and obtaining the customer's operation behavior features, wherein the first double-tower model also includes an article tower, the article tower is a dnn structure, and the user tower is an mmoe structure. Step S911 may include: inputting the customer's identity information and the operation behavior features into the customer tower of the second double-tower model, extracting features from the customer's identity information and the operation behavior features through the customer tower, and obtaining the customer's target features, wherein the second double-tower model also includes a staff tower, the customer tower is a dnn structure, and the staff tower is an mmoe structure.
[0070] The first dual-tower model is obtained by adjusting the parameters of the first dual-tower model according to the differences between multiple operation behavior prediction values and multiple operation behavior annotation values. The multiple operation behavior prediction values are obtained by performing dot product of each operation behavior feature of multiple operation behavior features of APP users in the first training set with an article feature. The multiple operation behavior features of APP users in the first training set are obtained by performing feature extraction of the identity information of APP users and their operation behaviors on the APP through the user tower in the first training set. An article feature is obtained by performing feature extraction of the identification, category, keyword and operated behavior of articles operated by APP users on the APP in the first training set through the article tower. The first training set includes multiple operation behavior annotation values.
[0071] The second double-tower model is obtained by adjusting the parameters of the second double-tower model according to the differences between multiple work result prediction values and multiple work result annotation values. The multiple work result prediction values are obtained by performing dot multiplication of each work result feature of the staff in the second training set with a target feature of the customers in the second training set. The multiple work result features of the staff in the second training set are obtained by extracting the identity information, work results and operational behavior characteristics of the staff in the second training set, through the staff tower, and the operational behavior characteristics of the customers who have been successfully served. The target feature of the customers in the second training set is obtained by extracting the identity information and operational behavior characteristics of the customers in the second training set through the customer tower. The second training set includes multiple work result annotation values.
[0072] In response to the fact that the customer has not used the APP, and the customer's friends have used the APP, the method also includes: extracting features from the identity information of the customer's friends and the operating behaviors of the customer's friends on the APP to obtain operating behavior features of the customer's friends; and fusing the operating behavior features of multiple friends of the customer, and using the fused operating behavior features as the operating behavior features of the customer.
[0073] In response to the fact that neither the customer nor the customer's friends have used the APP, the method also includes: matching the customer's identity information with the identity information of the APP user to obtain the APP user who matches the customer; extracting features of the identity information of the APP user who matches the customer and his / her operating behavior on the APP to obtain the operating behavior features of the APP user who matches the customer; and fusing the operating behavior features of the APP user who matches the customer, and using the fusing operating behavior features as the customer's operating behavior features.
[0074] The personnel relationship network includes APP users and friends of APP users, and the friends of APP users include at least one of persons who have used APP and persons who have not used APP. The method also includes: matching the customer's identity information with the identity information of persons in the personnel relationship network; if the identity information of a person who has used APP in the personnel relationship network is the same as the customer's identity information, it is determined that the customer has used APP; if the identity information of a person who has not used APP in the personnel relationship network is the same as the customer's identity information, it is determined that the customer has not used APP and the customer's friends have used APP; if there is no person in the personnel relationship network whose identity information is the same as the customer's identity information, it is determined that neither the customer nor the customer's friends have used APP.
[0075] Step S920 may include: S921, extracting features from the identity information of the successfully connected customer and the operation behavior of the successfully connected customer on the APP to obtain the operation behavior features of the successfully connected customer; and S922, extracting features from the identity information, work results and operation behavior features of the successfully connected customer of each of the multiple staff members to obtain the work result features of each staff member. Further, step S921 may include: inputting the identity information of the successfully connected customer and the operation behavior of the successfully connected customer on the APP into the user tower of the first dual-tower model, extracting features from the identity information of the successfully connected customer and the operation behavior of the successfully connected customer on the APP through the user tower to obtain the operation behavior features of the successfully connected customer, wherein the first dual-tower model also includes an article tower, the article tower is a dnn structure, and the user tower is an mmoe structure. Step S922 may include: inputting the identity information, work results and operational behavior characteristics of successfully served customers of each staff member among multiple staff members into the staff tower of the second dual-tower model, and extracting features of the identity information, work results and operational behavior characteristics of successfully served customers of each staff member through the staff tower to obtain the work result characteristics of each staff member, wherein the second dual-tower model also includes a customer tower, the customer tower is a dnn structure, and the staff tower is an mmoe structure.
[0076] The work result characteristics include transaction characteristics, test drive characteristics and satisfaction characteristics. Accordingly, step S930 may include: performing dot product of each staff member's transaction characteristics with the customer's target characteristics to obtain a first dot product value corresponding to each staff member; performing dot product of each staff member's test drive characteristics with the customer's target characteristics to obtain a second dot product value corresponding to each staff member; performing dot product of each staff member's satisfaction characteristics with the customer's target characteristics to obtain a third dot product value corresponding to each staff member; performing weighted summation of the first dot product value, the second dot product value and the third dot product value corresponding to each staff member to obtain the similarity between each staff member and the customer; and selecting the staff member with the greatest similarity as the staff member to be connected with the customer.
[0077] The method may further include: in response to the customer having never used the APP, but the customer's friend having used the APP, using the customer's friend's operating behavior on the APP as the customer's operating behavior on the APP; in response to the customer and the customer's friend having never used the APP, matching the customer's identity information with the identity information of the APP user to obtain an APP user that matches the customer, and using the operating behavior of the APP user that matches the customer on the APP as the customer's operating behavior on the APP.
[0078] For details about the staff determination method, please refer to the above description and will not be repeated here for the sake of brevity.
[0079] According to a further embodiment of the present disclosure, an electronic device is also provided, which may be a server 300, for example, a cloud server or a physical server. Fig. 9 The staff determination method shown.
[0080] Fig.10 The schematic block diagram of the structure of the electronic device is shown. The hardware structure of the electronic device 1000 of the present disclosure can be implemented using a bus architecture. The bus architecture may include any number of interconnected buses and bridges, depending on the specific application and overall design constraints of the hardware. The bus 1100 connects various circuits including one or more processors 1200, memory 1300 and / or hardware modules together. The bus 1100 may also connect various other circuits 1400 such as peripherals, voltage regulators, power management circuits, external antennas, etc. The bus 1100 may be an Industry Standard Architecture (ISA) bus, a Peripheral Component (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus 1100 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one connecting line is used in the figure, but it does not mean that there is only one bus or one type of bus. A computer program is stored in the memory 1300, and when the processor 1200 executes the computer program, the processor 1200 is able to perform the following process.
[0081] The target characteristics of the customer are determined according to the customer's identity information and the customer's operation behavior on the APP. This process can be implemented by the first determination module 1002.
[0082] The work result characteristics of each staff member are determined based on the identity information, work results and operation behaviors of successfully connected customers on the APP of each staff member among multiple staff members. This process can be achieved through the second determination module 1004.
[0083] The work result characteristics of each staff member are matched with the target characteristics of the customer respectively, and the staff member who is to be matched with the customer is determined. This process can be achieved through the matching module 1006.
[0084] The electronic device 1000 of the disclosed embodiment can determine the target features of a customer based on the customer's identity information and the customer's operating behaviors on the APP, and determine the work result features of each staff member based on the identity information, work results and operating behaviors of successfully connected customers on the APP of each staff member among multiple staff members, and then match the work result features of each staff member with the customer's target features respectively, and determine the staff member to be connected to the customer, so that the staff member to be connected to the customer is more in line with the customer's preferences, thereby improving the customer's transaction rate.
[0085] The contents related to the electronic device 1000 can refer to the above description, and will not be repeated here for the sake of brevity.
[0086] The present disclosure also provides a readable storage medium, in which execution instructions are stored, and when the execution instructions are executed by a processor, they are used to implement the above method. "Readable storage medium" can be any device that can contain, store, communicate, propagate or transmit programs for use in an instruction execution system, device or equipment or in combination with these instruction execution systems, devices or equipment. More specific examples of readable storage media include the following: an electrical connection portion (electronic device) with one or more wirings, a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and editable read-only memory (EPROM or flash memory), an optical fiber device, and a portable read-only memory (CDROM), etc.
[0087] The present disclosure also provides a computer program product. The method of the present disclosure can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer programs or instructions. When the computer program or instruction is loaded and executed, the process or function of the present disclosure is executed in whole or in part.
[0088] The computer program or instructions may be stored in a readable storage medium or transmitted from one readable storage medium to another readable storage medium, for example, the computer program or instructions may be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired or wireless means. The readable storage medium may be any available medium that can be accessed or a data storage device such as a server, data center, etc. that integrates one or more available media. The available medium may be a magnetic medium, such as a floppy disk, a hard disk, or a magnetic tape; it may also be an optical medium, such as a digital video disk; it may also be a semiconductor medium, such as a solid state drive. The computer readable storage medium may be a volatile or non-volatile storage medium, or may include both volatile and non-volatile types of storage media.
[0089] Those skilled in the art will appreciate that the embodiments of the present disclosure may be provided as methods, systems, or computer program products. Therefore, the present disclosure may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present disclosure may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0090] The present disclosure is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the present disclosure. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0091] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0092] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0093] It is understandable that before using the technical solutions disclosed in the embodiments of the present disclosure, the type, scope of use, usage scenarios, etc. of the personal information involved in the present disclosure should be informed to the user and the user's authorization should be obtained in an appropriate manner in accordance with relevant laws and regulations.
[0094] For example, in response to receiving an active request from a user, a prompt message is sent to the user to clearly prompt the user that the operation requested to be performed will require obtaining and using the user's personal information. Thus, the user can autonomously choose whether to provide personal information to software or hardware such as an electronic device, application, server, or storage medium that performs the operation of the technical solution of the present disclosure according to the prompt message.
[0095] As an optional but non-limiting implementation, in response to receiving an active request from the user, the prompt information may be sent to the user in the form of a pop-up window, in which the prompt information may be presented in text form. In addition, the pop-up window may also carry a selection control for the user to choose "agree" or "disagree" to provide personal information to the electronic device.
[0096] It is understandable that the above notification and the process of obtaining user authorization are merely illustrative and do not constitute a limitation on the implementation of the present disclosure. Other methods that meet the relevant laws and regulations may also be applied to the implementation of the present disclosure.
[0097] At the same time, it is understandable that the data involved in the technical solution of the present disclosure (including but not limited to the data itself, the acquisition or use of the data) shall comply with the requirements of relevant laws, regulations and relevant provisions.
[0098] In the description of this specification, the description with reference to the terms "one embodiment / method", "some embodiments / methods", "example", "specific example", or "some examples" etc. means that the specific features, structures, or characteristics described in conjunction with the embodiment / method or example are included in at least one embodiment / method or example of the present disclosure. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment / method or example. Moreover, the specific features, structures, or characteristics described may be combined in any one or more embodiments / methods or examples in a suitable manner. In addition, those skilled in the art may combine and combine different embodiments / methods or examples described in this specification and features of different embodiments / methods or examples, unless they are contradictory.
[0099] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of the present disclosure, "plurality" means at least two, such as two, three, etc., unless otherwise clearly and specifically defined.
[0100] Those skilled in the art should understand that the above embodiments are only for the purpose of clearly illustrating the present disclosure, and are not intended to limit the scope of the present disclosure. For those skilled in the art, other changes or modifications may be made based on the above disclosure, and these changes or modifications are still within the scope of the present disclosure.
Claims
1. A method for determining a staff member, characterized in that: include: Determine the target characteristics of the customer based on the customer's identity information and the customer's operation behavior on the APP; Determine the work result characteristics of each staff member according to the identity information, work results and operation behavior of successfully connected customers on the APP of each staff member among the multiple staff members; and The work result characteristics of each staff member are matched with the target characteristics of the customer respectively, and the staff member who is to be connected with the customer is determined.
2. The staff identification method according to claim 1, characterized in that: Determining the target characteristics of the customer based on the customer's identity information and the customer's operation behavior on the APP includes: Extracting features of the customer's identity information and the customer's operation behavior on the APP to obtain the customer's operation behavior features; and Feature extraction is performed on the customer's identity information and the customer's operation behavior characteristics to obtain the customer's target features.
3. The staff identification method according to claim 2, characterized in that: The feature extraction of the customer's identity information and the customer's operation behavior on the APP to obtain the customer's operation behavior features includes: Input the customer's identity information and the customer's operation behavior on the APP into the user tower of the first double-tower model, extract features of the customer's identity information and the customer's operation behavior on the APP through the user tower, and obtain the customer's operation behavior features, wherein the first double-tower model also includes an article tower, the article tower is a dnn structure, and the user tower is an mmoe structure; The extracting the identity information of the customer and the operation behavior characteristics of the customer to obtain the target characteristics of the customer includes: The customer's identity information and operation behavior characteristics are input into the customer tower of the second double-tower model, and the customer's identity information and operation behavior characteristics are extracted through the customer tower to obtain the target characteristics of the customer, wherein the second double-tower model also includes a staff tower, the customer tower is a DNN structure, and the staff tower is an MMOE structure.
4. The staff identification method according to claim 2, characterized in that: In response to the customer not having used the APP, and the customer's friend having used the APP, The method further comprises: Extracting features of the identity information of the customer's friends and the operation behaviors of the customer's friends on the APP to obtain operation behavior features of the customer's friends; and The operation behavior characteristics of multiple friends of the customer are integrated, and the integrated operation behavior characteristics are used as the operation behavior characteristics of the customer.
5. The staff identification method according to claim 2, characterized in that: In response to the fact that neither the customer nor the customer's friend has used the APP, The method further comprises: Matching the identity information of the customer with the identity information of the user of the APP to obtain the APP user matching the customer; Extracting features of the identity information of the APP user who matches the customer and his / her operation behavior on the APP to obtain operation behavior features of the APP user who matches the customer; and The operation behavior characteristics of the APP users matching the customer are integrated, and the integrated operation behavior characteristics are used as the operation behavior characteristics of the customer.
6. The staff identification method according to claim 2, characterized in that: The personnel relationship network includes APP users and friends of the APP users, and the friends of the APP users include at least one of a person who has used the APP and a person who has not used the APP. The method further comprises: Matching the customer's identity information with the identity information of people in the personnel relationship network; If the identity information of a person who has used the APP in the personnel relationship network is the same as the identity information of the customer, it is determined that the customer has used the APP; If the identity information of a person who has not used the APP in the personnel relationship network is the same as the identity information of the customer, it is determined that the customer has not used the APP, and the customer's friend has used the APP; If there is no person in the personnel relationship network whose identity information is the same as the identity information of the customer, it is determined that neither the customer nor the customer's friends have used the APP.
7. The staff identification method according to claim 1, characterized in that: Determining the work result characteristics of each staff member according to the identity information, work results and operation behavior of the successfully connected customers on the APP of each staff member among the multiple staff members includes: Extracting features of the identity information of the successfully connected customer and the operation behavior of the successfully connected customer on the APP to obtain the operation behavior features of the successfully connected customer; and Feature extraction is performed on the identity information, work results and operation behavior characteristics of the successfully connected customers of each of the multiple staff members to obtain the work result characteristics of each staff member.
8. The staff identification method according to claim 7, characterized in that: The feature extraction of the identity information of the successfully connected customer and the operation behavior of the successfully connected customer on the APP to obtain the operation behavior features of the successfully connected customer includes: Input the identity information of the successfully connected customer and the operation behavior of the successfully connected customer on the APP into the user tower of the first double-tower model, extract the identity information of the successfully connected customer and the operation behavior of the successfully connected customer on the APP through the user tower, and obtain the operation behavior characteristics of the successfully connected customer, wherein the first double-tower model also includes an article tower, the article tower is a dnn structure, and the user tower is an mmoe structure; The feature extraction of the identity information, work results and operation behavior characteristics of the successfully connected customers of each of the multiple staff members to obtain the work result characteristics of each staff member includes: The identity information, work results and operational behavior characteristics of the successfully served customers of each of the multiple staff members are input into the staff tower of the second double-tower model, and feature extraction is performed on the identity information, work results and operational behavior characteristics of the successfully served customers of each staff member through the staff tower to obtain the work result characteristics of each staff member, wherein the second double-tower model also includes a customer tower, the customer tower is a DNN structure, and the staff tower is an MMOE structure.
9. The staff identification method according to claim 1, characterized in that: The work result characteristics include transaction characteristics, test drive characteristics, and satisfaction characteristics. The work result characteristics of each staff member are matched with the target characteristics of the customer to determine the staff member to be connected with the customer, including: Performing dot product of the transaction characteristics of each staff member and the target characteristics of the customer, respectively, to obtain a first dot product value corresponding to each staff member; Performing dot product of the test drive feature of each staff member and the target feature of the customer to obtain a second dot product value corresponding to each staff member; Performing dot product of the satisfaction characteristic of each staff member and the target characteristic of the customer, respectively, to obtain a third dot product value corresponding to each staff member; Performing weighted summation on the first dot product value, the second dot product value, and the third dot product value corresponding to each staff member to obtain the similarity between each staff member and the customer; and The staff member with the greatest similarity is selected as the staff member to be matched with the customer.
10. The staff identification method according to any one of claims 1 to 9, characterized in that: Also includes: In response to the customer not having used the APP and the customer's friend having used the APP, taking the operation behavior of the customer's friend on the APP as the operation behavior of the customer on the APP; In response to the fact that neither the customer nor the customer's friend has used the APP, the customer's identity information is matched with the identity information of the user of the APP to obtain the APP user matching the customer, and the operating behavior of the APP user matching the customer on the APP is used as the operating behavior of the customer on the APP.
11. An electronic device, characterized in that: The invention comprises a processor and a memory, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the following process: Determine the target characteristics of the customer based on the customer's identity information and the customer's operation behavior on the APP; Determine the work result characteristics of each staff member according to the identity information, work results and operation behavior of successfully connected customers on the APP of each staff member among the multiple staff members; and The work result characteristics of each staff member are matched with the target characteristics of the customer respectively, and the staff member who is to be connected with the customer is determined.
12. A computer program product, characterized in that The computer program product comprises a computer program, which is used to implement at least the method according to any one of claims 1 to 10 when executed by a processor.
