Method and device for recommending outbound call leads, and electronic device

Abstract

The present application provides a method and device for recommending outbound call leads, an electronic device, and a computer-readable storage medium, and relates to the field of data processing technology. The method cross-processes at least two of the characteristic data of outbound call agents, characteristic data of outbound call leads, and communication data between outbound call agents and outbound call leads to generate outbound call cross-feature vector data; the outbound call cross-feature vector data is input into a pre-trained multimodal outbound call lead recommendation model, and the multimodal outbound call lead recommendation model is used to predict the order conversion rate of each outbound call lead corresponding to each outbound call seat, and the predicted order conversion rate of each outbound call lead corresponding to each outbound call seat is obtained; the outbound call lead recommendation queue corresponding to each outbound call seat is obtained according to the predicted order conversion rate of each outbound call lead corresponding to each outbound call seat, and the outbound call lead recommendation queue corresponding to each outbound call seat is recommended. This embodiment can improve the efficiency and order conversion rate of outbound calls.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method and device for recommending outbound call leads, an electronic device, and a computer-readable storage medium. Background Art

[0002] With the development of mobile Internet, various new sales models have emerged, and sales have gradually become online. The outbound call system is an indispensable and important part of many complex product services, such as insurance services, financial loan services, financial investment services, real estate services, car services, course services, travel route services, event invitations, etc. Faced with the one-to-many scenario of outbound call seats and outbound call leads, how to improve the efficiency of outbound call seat management and calling outbound call leads has become a technical problem that needs to be solved urgently. Summary of the invention

[0003] In view of the above problems, this application is proposed to provide a method and device for recommending outbound call leads, an electronic device, and a computer-readable storage medium that overcome the above problems or at least partially solve the above problems, and can improve the efficiency of outbound calls and the order conversion rate. The technical solution is as follows:

[0004] In a first aspect, a method for recommending outbound call leads is provided, comprising:

[0005] Obtaining feature data of outbound call agents, feature data of outbound call leads, and communication data between outbound call agents and outbound call leads;

[0006] Cross-process at least two of the feature data of the outbound call agent, the feature data of the outbound call clue, and the communication data between the outbound call agent and the outbound call clue to generate outbound call cross feature vector data;

[0007] Inputting the outbound call cross feature vector data into a pre-trained multimodal outbound call lead recommendation model, using the multimodal outbound call lead recommendation model to predict the order conversion rate of each outbound call lead corresponding to each outbound call seat, and obtaining the predicted order conversion rate of each outbound call lead corresponding to each outbound call seat;

[0008] According to the predicted order conversion rate of each outbound call lead corresponding to each outbound call seat, an outbound call lead recommendation queue corresponding to each outbound call seat is obtained, and outbound call leads are recommended based on the outbound call lead recommendation queue corresponding to each outbound call seat.

[0009] In a possible implementation, recommending outbound call leads based on the outbound call lead recommendation queues corresponding to the outbound call seats includes:

[0010] Based on the outbound call lead recommendation queue corresponding to each outbound call seat, the corresponding outbound call lead is allocated to each outbound call seat; or

[0011] The outbound call clue recommendation queue corresponding to each outbound call seat is provided to the corresponding outbound call seat, and the corresponding outbound call seat communicates with the outbound call clues in sequence according to the outbound call clue recommendation queue.

[0012] In a possible implementation, obtaining an outbound call lead recommendation queue corresponding to each outbound call agent according to the predicted order conversion rate of each outbound call lead corresponding to each outbound call agent includes:

[0013] Determine the highest predicted order conversion rate corresponding to each outbound call lead according to the predicted order conversion rate of each outbound call seat corresponding to each outbound call lead;

[0014] The outbound call seat corresponding to the highest predicted order conversion rate of each outbound call lead is used as the recommended outbound call seat corresponding to each outbound call lead;

[0015] The outbound call leads corresponding to each outbound call seat are counted according to the recommended outbound call seats corresponding to each outbound call lead, so as to obtain an outbound call lead recommendation queue corresponding to each outbound call seat.

[0016] In a possible implementation, the multimodal outbound call lead recommendation model is trained by the following steps:

[0017] Build an initial multimodal outbound call lead recommendation model;

[0018] Collect sample characteristic data of outbound call agents, sample characteristic data of outbound call leads, sample communication data between outbound call agents and outbound call leads, and historical order conversion rates of each outbound call lead corresponding to each outbound call agent;

[0019] Based on the sample feature data of the outbound call agents, the sample feature data of the outbound call leads, the sample communication data between the outbound call agents and the outbound call leads, and the historical order conversion rate of each outbound call lead corresponding to each outbound call agent, the initial multimodal outbound call lead recommendation model is trained to obtain a trained multimodal outbound call lead recommendation model.

[0020] In a possible implementation, the initial multimodal outbound call lead recommendation model is trained based on the sample feature data of the outbound call agent, the sample feature data of the outbound call lead, the sample communication data between the outbound call agent and the outbound call lead, and the historical order conversion rate of each outbound call lead corresponding to each outbound call agent to obtain the trained multimodal outbound call lead recommendation model, including:

[0021] Cross-process at least two of the sample feature data of the outbound call agent, the sample feature data of the outbound call clue, and the sample communication data of the outbound call agent and the outbound call clue to generate outbound cross-sample feature vector data;

[0022] The outbound call cross-sample feature vector data is used as input, and the historical order conversion rate of each outbound call lead corresponding to each outbound call seat is used as output, and the initial multimodal outbound call lead recommendation model is trained to obtain a trained multimodal outbound call lead recommendation model.

[0023] In a possible implementation, cross-processing is performed on at least two of the feature data of the outbound call agent, the feature data of the outbound call clue, and the communication data between the outbound call agent and the outbound call clue to generate outbound cross feature vector data, including:

[0024] Select at least two types of data from the feature data of the outbound call agent, the feature data of the outbound call clue, and the communication data between the outbound call agent and the outbound call clue, and construct a vector for each type of data;

[0025] Performing feature-level cross processing on data of at least two vectors to obtain a first processing result;

[0026] Performing element-level cross processing on data of at least two vectors to obtain a second processing result;

[0027] Combining data of at least two vectors in pairs, performing low-order cross processing, and obtaining a third processing result;

[0028] The first processing result, the second processing result and the third processing result are combined and transformed using a preset function to generate outbound call cross feature vector data.

[0029] In a second aspect, a device for recommending outbound call leads is provided, comprising:

[0030] An acquisition module, used to acquire characteristic data of outbound call agents, characteristic data of outbound call leads, and communication data between outbound call agents and outbound call leads;

[0031] A generating module, configured to perform cross processing on at least two of the feature data of the outbound call agent, the feature data of the outbound call clue, and the communication data between the outbound call agent and the outbound call clue, to generate outbound cross feature vector data;

[0032] A prediction module, used for inputting the outbound call cross feature vector data into a pre-trained multimodal outbound call lead recommendation model, using the multimodal outbound call lead recommendation model to predict the order conversion rate of each outbound call lead corresponding to each outbound call seat, and obtaining the predicted order conversion rate of each outbound call lead corresponding to each outbound call seat;

[0033] The recommendation module is used to obtain the outbound call lead recommendation queue corresponding to each outbound call seat according to the predicted order conversion rate of each outbound call lead corresponding to each outbound call seat, and recommend outbound call leads based on the outbound call lead recommendation queue corresponding to each outbound call seat.

[0034] In a possible implementation, the recommendation module is further used to:

[0035] Based on the outbound call lead recommendation queue corresponding to each outbound call seat, the corresponding outbound call lead is allocated to each outbound call seat; or

[0036] The outbound call clue recommendation queue corresponding to each outbound call seat is provided to the corresponding outbound call seat, and the corresponding outbound call seat communicates with the outbound call clues in sequence according to the outbound call clue recommendation queue.

[0037] In a possible implementation, the recommendation module is further used to:

[0038] Determine the highest predicted order conversion rate corresponding to each outbound call lead according to the predicted order conversion rate of each outbound call seat corresponding to each outbound call lead;

[0039] The outbound call seat corresponding to the highest predicted order conversion rate of each outbound call lead is used as the recommended outbound call seat corresponding to each outbound call lead;

[0040] The outbound call leads corresponding to each outbound call seat are counted according to the recommended outbound call seats corresponding to each outbound call lead, so as to obtain an outbound call lead recommendation queue corresponding to each outbound call seat.

[0041] In a possible implementation manner, the device further includes a training module, and the training module is used to:

[0042] Build an initial multimodal outbound call lead recommendation model;

[0043] Collect sample characteristic data of outbound call agents, sample characteristic data of outbound call leads, sample communication data between outbound call agents and outbound call leads, and historical order conversion rates of each outbound call lead corresponding to each outbound call agent;

[0044] Based on the sample feature data of the outbound call agents, the sample feature data of the outbound call leads, the sample communication data between the outbound call agents and the outbound call leads, and the historical order conversion rate of each outbound call lead corresponding to each outbound call agent, the initial multimodal outbound call lead recommendation model is trained to obtain a trained multimodal outbound call lead recommendation model.

[0045] In a possible implementation, the training module is further used to:

[0046] Cross-process at least two of the sample feature data of the outbound call agent, the sample feature data of the outbound call clue, and the sample communication data of the outbound call agent and the outbound call clue to generate outbound cross-sample feature vector data;

[0047] The outbound call cross-sample feature vector data is used as input, and the historical order conversion rate of each outbound call lead corresponding to each outbound call seat is used as output, and the initial multimodal outbound call lead recommendation model is trained to obtain a trained multimodal outbound call lead recommendation model.

[0048] In a possible implementation, the generating module is further used for:

[0049] Select at least two types of data from the feature data of the outbound call agent, the feature data of the outbound call clue, and the communication data between the outbound call agent and the outbound call clue, and construct a vector for each type of data;

[0050] Performing feature-level cross processing on data of at least two vectors to obtain a first processing result;

[0051] Performing element-level cross processing on data of at least two vectors to obtain a second processing result;

[0052] Combining data of at least two vectors in pairs, performing low-order cross processing, and obtaining a third processing result;

[0053] The first processing result, the second processing result and the third processing result are combined and transformed using a preset function to generate outbound call cross feature vector data.

[0054] According to a third aspect, an electronic device is provided, comprising a processor and a memory, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute any of the above-mentioned methods for recommending outbound call leads.

[0055] In a fourth aspect, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, wherein the computer program is configured to execute any of the above-mentioned methods for recommending outbound call leads when running.

[0056] By means of the above technical scheme, the outbound call lead recommendation method and device, electronic device and computer-readable storage medium provided in the embodiments of the present application can obtain the feature data of the outbound call agent, the feature data of the outbound call lead and the communication data between the outbound call agent and the outbound call lead; perform cross-processing on at least two of the feature data of the outbound call agent, the feature data of the outbound call lead and the communication data between the outbound call agent and the outbound call lead to generate outbound call cross-feature vector data; then input the outbound call cross-feature vector data into a pre-trained multimodal outbound call lead recommendation model, and use the multimodal outbound call lead recommendation model to predict the order conversion rate of each outbound call lead corresponding to each outbound call seat, and obtain the predicted order conversion rate of each outbound call lead corresponding to each outbound call seat; then, obtain the outbound call lead recommendation queue corresponding to each outbound call seat according to the predicted order conversion rate of each outbound call lead corresponding to each outbound call seat, and recommend outbound calls based on the outbound call lead recommendation queue corresponding to each outbound call seat. It can be seen that the embodiment of the present application can recommend some outbound call leads suitable for the outbound call agent in combination with some characteristics of the outbound call agent itself, thereby forming personalized recommendations for each outbound call agent, which can improve the efficiency of outbound calls and order conversion rate, and thus improve the efficiency of the entire outbound telephone call system. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in describing the embodiments of the present application are briefly introduced below.

[0058] Figure 1 The system structure diagram of outbound call clue distribution in the related art is shown;

[0059] Figure 2 A flowchart of a method for recommending outbound call leads provided in an embodiment of the present application is shown;

[0060] Figure 3 A system structure diagram of outbound call clue distribution in an embodiment of the present application is shown;

[0061] Figure 4 The following is an architecture diagram of outbound call clue distribution in an embodiment of the present application;

[0062] Figure 5 A structural diagram of a device for recommending outbound call leads provided in an embodiment of the present application is shown;

[0063] Figure 6 A structural diagram of an outbound call clue recommendation device provided by another embodiment of the present application is shown;

[0064] Figure 7 A structural diagram of an electronic device according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0065] The exemplary embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present application and to fully convey the scope of the present application to those skilled in the art.

[0066] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that such use is interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the term "including" and its variants are to be interpreted as open-ended terms meaning "including but not limited to".

[0067] Figure 1 The system structure diagram of outbound call clue distribution in the related art is shown. Figure 1 In the process, retained leads, new leads, and secondary contact leads are connected to the CRM (Customer Relationship Management) system. For the two leads with different time limits, real-time leads and yesterday's leads, three different allocation methods are set, namely number allocation, clock-in allocation, and operation allocation. The outbound call agent obtains the outbound call leads under the restrictions of these three allocation methods, and then performs the process of lead dialing, lead connection, and lead conversion. The number allocation here means that the outbound call agent sends a number allocation request and obtains the outbound call lead from the real-time lead or yesterday's lead queue; the clock-in allocation means that the outbound call agent obtains the outbound call lead from the real-time lead or yesterday's lead queue after clocking in; the operation allocation means that the operation personnel allocates the outbound call lead queue to each outbound call agent, so that each outbound call agent communicates with the outbound call lead according to the outbound call lead queue. The related technology has the following problems in the whole process:

[0068] (1) For channels with large delivery volumes, there is a situation where outbound call agents cannot make all the calls, resulting in a waste of outbound call leads;

[0069] (2) It is not possible to allocate outbound call leads suitable for the outbound call agent based on the outbound call agent's own characteristics, which cannot significantly improve the overall outbound call efficiency and order conversion rate.

[0070] In order to solve the above technical problems, the embodiment of the present application provides a method for recommending outbound call leads, which models outbound call seats and outbound call leads, and can form personalized recommendations for each outbound call seat. Figure 2As shown, the outbound call lead recommendation method can be applied to electronic devices such as servers, personal computers, smart phones, tablet computers, smart watches, etc., and can specifically include the following steps S201 to S204:

[0071] Step S201, acquiring feature data of an outbound call agent, feature data of an outbound call clue, and communication data between the outbound call agent and the outbound call clue;

[0072] Step S202, cross-processing at least two of the feature data of the outbound call agent, the feature data of the outbound call clue, and the communication data between the outbound call agent and the outbound call clue to generate outbound cross feature vector data;

[0073] Step S203, inputting the outbound call cross feature vector data into a pre-trained multimodal outbound call lead recommendation model, and using the multimodal outbound call lead recommendation model to predict the order conversion rate of each outbound call lead corresponding to each outbound call seat, to obtain the predicted order conversion rate of each outbound call lead corresponding to each outbound call seat;

[0074] Step S204, obtaining an outbound call lead recommendation queue corresponding to each outbound call seat according to the predicted order conversion rate of each outbound call lead corresponding to each outbound call seat, and recommending outbound call leads based on the outbound call lead recommendation queue corresponding to each outbound call seat.

[0075] The embodiment of the present application can obtain the characteristic data of the outbound call seat, the characteristic data of the outbound call clue, and the communication data between the outbound call seat and the outbound call clue; perform cross-processing on at least two of the characteristic data of the outbound call seat, the characteristic data of the outbound call clue, and the communication data between the outbound call seat and the outbound call clue to generate outbound cross-feature vector data; then input the outbound cross-feature vector data into the pre-trained multimodal outbound call clue recommendation model, and use the multimodal outbound call clue recommendation model to predict the order conversion rate of each outbound call clue corresponding to each outbound call seat, and obtain the predicted order conversion rate of each outbound call clue corresponding to each outbound call seat; then obtain the outbound call clue recommendation queue corresponding to each outbound call seat according to the predicted order conversion rate of each outbound call clue corresponding to each outbound call seat, and recommend outbound call clues based on the outbound call clue recommendation queue corresponding to each outbound call seat. It can be seen that the embodiment of the present application can recommend some outbound call clues suitable for the outbound call seat in combination with some characteristics of the outbound call seat itself, thereby forming personalized recommendations for each outbound call seat, which can improve the efficiency of outbound calls and order conversion rate, and thus improve the efficiency of the entire telephone outbound call system.

[0076] The characteristic data of the outbound call agent mentioned in step S201 above may be some characteristic data of the outbound call agent itself, such as age, work experience, educational experience, etc., and may also be the number of outbound calls, the number of outbound call leads, the outbound call completion order data, etc., which is not limited in this embodiment of the present application.

[0077] The outbound call leads here may be users who may have needs for insurance services, real estate services, car services, course services, etc. The characteristic data of the outbound call leads may be the age, work experience, education experience, social income, family situation, etc. of the outbound call leads, and may also be the configuration of the outbound call corresponding service, etc., which is not limited in the embodiments of the present application.

[0078] The communication data between the outbound call agent and the outbound call clue may be call voice data, call text data, instant messaging text data, etc., which is not limited in this embodiment of the present application.

[0079] In the above step S202, at least two of the feature data of the outbound agent, the feature data of the outbound clue, and the communication data between the outbound agent and the outbound clue are cross-processed. The feature data of the outbound agent and the feature data of the outbound clue may be cross-processed, or the feature data of the outbound agent and the communication data between the outbound agent and the outbound clue may be cross-processed, or the feature data of the outbound clue and the communication data between the outbound agent and the outbound clue may be cross-processed, or the feature data of the outbound clue and the communication data between the outbound agent and the outbound clue may be cross-processed, so as to generate outbound cross feature vector data.

[0080] The present application provides a possible implementation method, which can train a multimodal outbound call lead recommendation model by following the steps A1 to A3:

[0081] Step A1, constructing an initial multimodal outbound call lead recommendation model;

[0082] Step A2, collecting sample feature data of outbound call agents, sample feature data of outbound call leads, sample communication data between outbound call agents and outbound call leads, and historical order conversion rates of each outbound call lead corresponding to each outbound call agent;

[0083] Step A3, based on the sample feature data of the outbound call agents, the sample feature data of the outbound call leads, the sample communication data between the outbound call agents and the outbound call leads, and the historical order conversion rate of each outbound call lead corresponding to each outbound call agent, the initial multimodal outbound call lead recommendation model is trained to obtain a trained multimodal outbound call lead recommendation model.

[0084] In step A1, the initial multimodal outbound call lead recommendation model can be built based on xdeepfm (extremely deep factorization machine model). The two modules of CIN (Compressed Interaction Network) and DNN (Deep Neural Networks) integrated in xdeepfm can help the model learn high-order feature interactions in both explicit and implicit ways, and the integrated linear module and deep neural module also allow the model to have both memory and generalization learning capabilities. In order to improve the versatility of the model, different modules in xdeepfm share the same input data. In specific application scenarios, different modules can also access their own different input data. For example, the linear module can still access many cross-features extracted based on prior knowledge to improve memory capabilities, while in CIN or DNN, in order to reduce the computational complexity of the model, only a part of the sparse feature subset can be imported.

[0085] Corresponding to the characteristic data of the outbound call agent, the characteristic data of the outbound call clues, and the communication data between the outbound call agent and the outbound call clues introduced above, the sample characteristic data of the outbound call agent can be some characteristic data of the outbound call agent itself, such as age, work experience, educational experience, etc., and can also be the number of outbound calls, the number of outbound call clues, the outbound call completion order data, etc., which is not limited in this embodiment of the present application.

[0086] The outbound call leads here may be users who may have needs for insurance services, real estate services, car services, course services, etc. The sample feature data of the outbound call leads may be the age, work experience, education experience, social income, family situation, etc. of the outbound call leads, and may also be the configuration of the outbound call corresponding service, etc., which is not limited in the embodiments of the present application.

[0087] The sample communication data between the outbound call agent and the outbound call clue may be call voice data, call text data, instant messaging text data, etc., which is not limited in this embodiment of the present application.

[0088] In an embodiment of the present application, a possible implementation method is provided. In the above step A3, based on the sample feature data of the outbound call seat, the sample feature data of the outbound call lead, the sample communication data between the outbound call seat and the outbound call lead, and the historical order conversion rate of each outbound call lead corresponding to each outbound call seat, the initial multimodal outbound call lead recommendation model is trained to obtain the trained multimodal outbound call lead recommendation model, which may specifically include the following steps A3-1 and A3-2:

[0089] Step A3-1, cross-processing at least two of the sample feature data of the outbound call agent, the sample feature data of the outbound call clue, and the sample communication data of the outbound call agent and the outbound call clue to generate outbound cross-sample feature vector data;

[0090] It can be understood that cross-processing of at least two of the sample feature data of outbound agents, the sample feature data of outbound clues, and the sample communication data between outbound agents and outbound clues may be cross-processing of the sample feature data of outbound agents and the sample feature data of outbound clues, or cross-processing of the sample feature data of outbound agents and the sample communication data between outbound agents and outbound clues, or cross-processing of the sample feature data of outbound clues and the sample communication data between outbound agents and outbound clues, or cross-processing of the sample feature data of outbound agents, the sample feature data of outbound clues, and the sample communication data between outbound agents and outbound clues, thereby generating outbound cross-sample feature vector data.

[0091] Step A3-2, taking the outbound call cross sample feature vector data as input, taking the historical order conversion rate of each outbound call lead corresponding to each outbound call seat as output, training the initial multimodal outbound call lead recommendation model, and obtaining a trained multimodal outbound call lead recommendation model.

[0092] The embodiment of the present application continuously optimizes the adaptability of outbound call seats and outbound call leads by adjusting the model parameters, thereby forming personalized recommendations for each outbound call seat, which can improve the efficiency of outbound calls and the order conversion rate.

[0093] The present application provides a possible implementation method. In step S204, the outbound call lead recommendation queue corresponding to each outbound call seat is obtained according to the predicted order conversion rate of each outbound call lead corresponding to each outbound call seat, which may specifically include the following steps B1 to B3:

[0094] Step B1, determining the highest predicted order conversion rate corresponding to each outbound call lead according to the predicted order conversion rate of each outbound call seat corresponding to each outbound call lead;

[0095] Step B2, taking the outbound call seat corresponding to the highest predicted order conversion rate for each outbound call lead as the recommended outbound call seat corresponding to each outbound call lead;

[0096] Step B3, counting the outbound leads corresponding to each outbound call seat according to the recommended outbound call seats corresponding to each outbound call lead, and obtaining the outbound call lead recommendation queue corresponding to each outbound call seat.

[0097] For example, the outbound call seats are A, B, C, D, etc., and the outbound call leads are 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, etc. The predicted order conversion rates of outbound call lead 1 corresponding to the outbound call seats A, B, C, D are 0.5, 0.6, 0.4, 0.7, the predicted order conversion rates of outbound call lead 2 corresponding to the outbound call seats A, B, C, D are 0.4, 0.2, 0.5, 0.3, the predicted order conversion rates of outbound call lead 3 corresponding to the outbound call seats A, B, C, D are 0.2, 0.4, 0.6, 0.3, etc. The highest predicted order conversion rate corresponding to each outbound call lead can be determined. For example, the highest predicted order conversion rate corresponding to outbound call lead 1 is The measured order conversion rate is 0.7, the highest predicted order conversion rate corresponding to outbound lead 2 is 0.5, the highest predicted order conversion rate corresponding to outbound lead 3 is 0.6, and so on. The outbound seats corresponding to the highest predicted order conversion rates of each outbound lead can be used as recommended outbound seats corresponding to each outbound lead; and then the outbound leads corresponding to each outbound seat are counted according to the recommended outbound seats corresponding to each outbound clue, to obtain the outbound lead recommendation queue corresponding to each outbound seat, that is, outbound seat D calls outbound lead 1, outbound seat C calls outbound lead 2 and calls lead 3. It should be noted that this is only an example and does not limit the embodiments of the present application.

[0098] The embodiment of the present application provides a possible implementation method, in which the above step S204 recommends outbound leads based on the outbound lead recommendation queue corresponding to each outbound call seat, and specifically, the outbound lead recommendation queue corresponding to each outbound call seat is allocated to each outbound call seat; or the outbound lead recommendation queue corresponding to each outbound call seat is provided to the corresponding outbound call seat, and the corresponding outbound call seat communicates with the outbound lead in sequence according to the outbound lead recommendation queue. The embodiment of the present application can recommend some outbound leads suitable for the outbound call seat in combination with some characteristics of the outbound call seat itself, thereby forming personalized recommendations for each outbound call seat, which can improve the efficiency of outbound calls and the order conversion rate, and thus improve the efficiency of the entire telephone outbound call system.

[0099] The embodiment of the present application provides a possible implementation method. In the above step S202, at least two of the feature data of the outbound call agent, the feature data of the outbound call clue, and the communication data between the outbound call agent and the outbound call clue are cross-processed to generate outbound cross feature vector data. Specifically, xdeepfm can be used for cross processing, which can include the following steps C1 to C5:

[0100] Step C1, selecting at least two types of data from the feature data of the outbound call agent, the feature data of the outbound call clue, and the communication data between the outbound call agent and the outbound call clue, and constructing a vector for each type of data;

[0101] Step C2, performing feature-level cross processing on the data of at least two vectors to obtain a first processing result;

[0102] Step C3, performing element-level cross processing on the data of at least two vectors to obtain a second processing result;

[0103] Step C4, combining the data of at least two vectors in pairs and performing low-order cross processing to obtain a third processing result;

[0104] Step C5, combining the first processing result, the second processing result and the third processing result, using a preset function to perform transformation processing to generate outbound call cross feature vector data.

[0105] The implementation example of this application can make full use of the relationship between data, extract more implicit features, and take into account both high-order and low-order processing, so that data utilization is more complete and the subsequent prediction results are more accurate, meeting the needs of actual application scenarios.

[0106] In actual applications, the feature data of outbound agents, the feature data of outbound leads, and the communication data between outbound agents and outbound leads are different fields. The processing method is to embed the data of the three fields into vectors of the same latitude. For example, the feature vectors of the two fields are (a1, a2, a3) and (b1, b2, b3). The specific cross processing includes:

[0107] (1) Perform feature-level crossover between data of different fields, that is, perform Hadamard product on all elements between vectors, and then perform convolution transformation under certain weights, f(w*(a1*b1,a2*b2,a3*b3));

[0108] (2) All filed vector data are crossed at the element level, that is, each element between the vectors is Hadamard-producted, and different weight values ​​are assigned to the results of each product, and then linear transformation is performed, f(w1*a1*b1,w2*a2*b2,w3*a3*b3);

[0109] (3) All data are combined in pairs and low-order cross processing is performed f(w(a1,a2,a3,b1,b2,b3));

[0110] The above three processing results are combined together and transformed using a preset function to obtain the outbound call cross feature vector data. The preset function here can be set according to actual conditions, and this embodiment does not limit this. It should be noted that the above examples are only illustrative and do not limit the embodiments of the present application.

[0111] The above introduces Figure 2In the illustrated embodiment, there are multiple implementation methods for each link. The method for recommending outbound call leads provided in the embodiment of the present application is further described below through specific embodiments.

[0112] For a company, high-quality customers are valuable, and the time of sales staff is also valuable. It is particularly important to improve the order conversion rate of each sales staff and reduce unnecessary time waste. In order to solve the existing problems, this application builds a recommendation system with sales staff as the core to assign the best customers to sales staff. Here, sales staff can call outbound seats, and customers can be outbound leads.

[0113] This application improves upon the original system, adjusts the data flow method while adapting to the original CRM system, and adds a recommended link to solve the problems arising in the above system. Figure 3 The system structure diagram of outbound call clue distribution in the embodiment of the present application is shown. Figure 3 A recommendation system is added, which performs the above steps S201 to S204 to recommend outbound leads based on the outbound lead recommendation queues corresponding to each outbound agent, which can be further divided into real-time outbound leads corresponding to outbound agents and non-real-time outbound leads corresponding to outbound agents, and then connected to the CRM system for real-time recommendation. Here, the recommendation system is connected to the CRM system:

[0114] (1) Based on big data computing and processing, users from different channels are classified into recommendation models;

[0115] (2) Combined with the sales data of outbound call agents, the data flowing out of the recommendation system is connected to the CRM system to make real-time recommendations.

[0116] Optimize sales and adapt user recommendation algorithm:

[0117] (1) Periodically obtain sales staff’s statistical data and text data, perform part-of-speech analysis, and generate feature vectors to be incorporated into the recommendation model;

[0118] It is understandable that based on text data and statistical data, NLP (Natural Language Processing) and Bayesian algorithms can be used to efficiently complete the characterization of sales portraits.

[0119] (2) Process more user data and cross-process it with salesperson data;

[0120] (3) Parameter and structure optimization of the recommended model.

[0121] Figure 4 The architecture diagram of outbound call clue distribution in the embodiment of the present application is shown. Figure 4In the application end, the CRM system can be included, the application layer can include call text recall, sales portrait vector splicing, user portrait vector splicing, xdeepfm model prediction, lead re-arrangement and distribution, and the data layer can include sales data, user data, order data, call text data, etc.

[0122] Based on the existing system framework, the embodiments of the present application improve the efficiency of data processing and utilization by asynchronously processing massive text data, sales portrait data and user data, and transforming the data flow process. It introduces multi-dimensional data such as sales behavior and habits, models leads from a sales perspective, and rearranges and allocates them, thereby greatly improving the utilization rate of leads. By adjusting the model parameters, the compatibility between sales and users is continuously optimized, and the business results lead to an increase in the order conversion rate and APL (Average Product of Labor) of online sales.

[0123] It should be noted that, in practical applications, all possible implementation methods described above can be arbitrarily combined to form possible embodiments of the present application, which will not be described one by one here.

[0124] Based on the methods for recommending outbound call leads provided in the above embodiments, and based on the same inventive concept, an embodiment of the present application further provides a device for recommending outbound call leads.

[0125] Figure 5 FIG. 1 shows a structural diagram of a device for recommending outbound call leads provided in an embodiment of the present application. Figure 5 As shown, the outbound call lead recommendation device may include an acquisition module 510 , a generation module 520 , a prediction module 530 and a recommendation module 540 .

[0126] An acquisition module 510 is used to acquire characteristic data of an outbound call agent, characteristic data of an outbound call clue, and communication data between an outbound call agent and an outbound call clue;

[0127] A generating module 520, configured to perform cross-processing on at least two of the feature data of the outbound call agent, the feature data of the outbound call clue, and the communication data between the outbound call agent and the outbound call clue, to generate outbound cross-feature vector data;

[0128] The prediction module 530 is used to input the outbound call cross feature vector data into a pre-trained multimodal outbound call lead recommendation model, and use the multimodal outbound call lead recommendation model to predict the order conversion rate of each outbound call lead corresponding to each outbound call seat, so as to obtain the predicted order conversion rate of each outbound call lead corresponding to each outbound call seat;

[0129] The recommendation module 540 is used to obtain the outbound call lead recommendation queue corresponding to each outbound call seat according to the predicted order conversion rate of each outbound call lead corresponding to each outbound call seat, and recommend outbound call leads based on the outbound call lead recommendation queue corresponding to each outbound call seat.

[0130] A possible implementation method is provided in the embodiments of the present application. Figure 5 The illustrated recommendation module 540 is also used to:

[0131] Based on the outbound call lead recommendation queue corresponding to each outbound call seat, the corresponding outbound call lead is allocated to each outbound call seat; or

[0132] The outbound call lead recommendation queue corresponding to each outbound call seat is provided to the corresponding outbound call seat, and the corresponding outbound call seat communicates with the outbound call leads in sequence according to the outbound call lead recommendation queue.

[0133] A possible implementation method is provided in the embodiments of the present application. Figure 5 The illustrated recommendation module 540 is also used to:

[0134] According to the predicted order conversion rate of each outbound call lead corresponding to each outbound call seat, determine the highest predicted order conversion rate corresponding to each outbound call lead;

[0135] The outbound call agent corresponding to the highest predicted order conversion rate for each outbound call lead is used as the recommended outbound call agent corresponding to each outbound call lead;

[0136] According to the recommended outbound call seats corresponding to each outbound call lead, the outbound call leads corresponding to each outbound call seat are counted to obtain the outbound call lead recommendation queue corresponding to each outbound call seat.

[0137] A possible implementation method is provided in the embodiment of the present application, such as Figure 6 As shown above Figure 5 The device shown may also include a training module 610, the training module 610 is used to:

[0138] Build an initial multimodal outbound call lead recommendation model;

[0139] Collect sample characteristic data of outbound call agents, sample characteristic data of outbound call leads, sample communication data between outbound call agents and outbound call leads, and historical order conversion rates of each outbound call lead corresponding to each outbound call agent;

[0140] Based on the sample feature data of outbound call agents, the sample feature data of outbound call leads, the sample communication data between outbound call agents and outbound call leads, and the historical order conversion rate of each outbound call lead corresponding to each outbound call agent, the initial multimodal outbound call lead recommendation model is trained to obtain a trained multimodal outbound call lead recommendation model.

[0141] A possible implementation method is provided in the embodiment of the present application, and the training module 610 is further used to:

[0142] Cross-process at least two of the sample feature data of the outbound call agent, the sample feature data of the outbound call clue, and the sample communication data of the outbound call agent and the outbound call clue to generate outbound cross-sample feature vector data;

[0143] The outbound call cross-sample feature vector data is taken as input, and the historical order conversion rate of each outbound call lead corresponding to each outbound call seat is taken as output, and the initial multimodal outbound call lead recommendation model is trained to obtain a trained multimodal outbound call lead recommendation model.

[0144] A possible implementation method is provided in an embodiment of the present application, where the generating module 520 is further configured to:

[0145] Select at least two types of data from the feature data of the outbound call agent, the feature data of the outbound call clue, and the communication data between the outbound call agent and the outbound call clue, and construct a vector for each type of data;

[0146] Performing feature-level cross processing on data of at least two vectors to obtain a first processing result;

[0147] Performing element-level cross processing on data of at least two vectors to obtain a second processing result;

[0148] Combining data of at least two vectors in pairs, performing low-order cross processing, and obtaining a third processing result;

[0149] The first processing result, the second processing result and the third processing result are combined and transformed using a preset function to generate outbound call cross feature vector data.

[0150] Based on the same inventive concept, an embodiment of the present application further provides an electronic device, including a processor and a memory, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the method for recommending outbound call leads of any of the above embodiments.

[0151] In an exemplary embodiment, an electronic device is provided, such as Figure 7 As shown, Figure 7 The electronic device 700 shown includes: a processor 701 and a memory 703. The processor 701 and the memory 703 are connected, such as through a bus 702. Optionally, the electronic device 700 may also include a transceiver 704. It should be noted that in actual applications, the transceiver 704 is not limited to one, and the structure of the electronic device 700 does not constitute a limitation on the embodiments of the present application.

[0152] Processor 701 may be a CPU (Central Processing Unit), a general purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It may implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of this application. Processor 701 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0153] The bus 702 may include a path to transmit information between the above components. The bus 702 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The bus 702 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 7 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0154] The memory 703 can be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disk storage (including compressed optical disk, laser disk, optical disk, digital versatile disk, Blu-ray disk, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited to these.

[0155] The memory 703 is used to store the application code for executing the solution of the present application, and the execution is controlled by the processor 701. The processor 701 is used to execute the application code stored in the memory 703 to implement the contents shown in the above method embodiment.

[0156] Among them, electronic devices include but are not limited to: mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 7 The electronic device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0157] Based on the same inventive concept, an embodiment of the present application further provides a computer-readable storage medium, in which a computer program is stored, wherein the computer program is configured to execute the method for recommending outbound call leads of any of the above embodiments when running.

[0158] Those skilled in the art can clearly understand that the specific working processes of the systems, devices, and modules described above can refer to the corresponding processes in the aforementioned method embodiments, and for the sake of brevity, they are not further described here.

[0159] Those skilled in the art can understand that the technical solution of the present application can be essentially or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, which includes a number of program instructions, so that an electronic device (such as a personal computer, a server, or a network device, etc.) executes all or part of the steps of the method described in each embodiment of the present application when running the program instructions. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0160] Alternatively, all or part of the steps of implementing the aforementioned method embodiments may be accomplished by hardware associated with program instructions (such as electronic devices such as personal computers, servers, or network devices), and the program instructions may be stored in a computer-readable storage medium. When the program instructions are executed by a processor of an electronic device, the electronic device executes all or part of the steps of the methods described in the embodiments of the present application.

[0161] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that within the spirit and principles of the present application, they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not deviate from the protection scope of the present application.

Claims

1. A method for recommending outbound call leads, characterized in that: include: Obtaining feature data of outbound call agents, feature data of outbound call leads, and communication data between outbound call agents and outbound call leads; Cross-process at least two of the feature data of the outbound call agent, the feature data of the outbound call clue, and the communication data between the outbound call agent and the outbound call clue to generate outbound call cross feature vector data; Inputting the outbound call cross feature vector data into a pre-trained multimodal outbound call lead recommendation model, using the multimodal outbound call lead recommendation model to predict the order conversion rate of each outbound call lead corresponding to each outbound call seat, and obtaining the predicted order conversion rate of each outbound call lead corresponding to each outbound call seat; Obtaining an outbound call lead recommendation queue corresponding to each outbound call seat according to the predicted order conversion rate of each outbound call lead corresponding to each outbound call seat, and recommending outbound call leads based on the outbound call lead recommendation queue corresponding to each outbound call seat; The multimodal outbound call lead recommendation model is trained by the following steps: Build an initial multimodal outbound call lead recommendation model; Collect sample characteristic data of outbound call agents, sample characteristic data of outbound call leads, sample communication data between outbound call agents and outbound call leads, and historical order conversion rates of each outbound call lead corresponding to each outbound call agent; Based on the sample feature data of the outbound call agents, the sample feature data of the outbound call leads, the sample communication data between the outbound call agents and the outbound call leads, and the historical order conversion rate of each outbound call lead corresponding to each outbound call agent, the initial multimodal outbound call lead recommendation model is trained to obtain a trained multimodal outbound call lead recommendation model.

2. The method for recommending outbound call leads according to claim 1, characterized in that: Recommending outbound call leads based on the outbound call lead recommendation queues corresponding to the outbound call seats includes: Based on the outbound call lead recommendation queue corresponding to each outbound call seat, the corresponding outbound call lead is allocated to each outbound call seat; or The outbound call clue recommendation queue corresponding to each outbound call seat is provided to the corresponding outbound call seat, and the corresponding outbound call seat communicates with the outbound call clues in sequence according to the outbound call clue recommendation queue.

3. The method for recommending outbound call leads according to claim 1, characterized in that: According to the predicted order conversion rate of each outbound call lead corresponding to each outbound call seat, the outbound call lead recommendation queue corresponding to each outbound call seat is obtained, including: Determine the highest predicted order conversion rate corresponding to each outbound call lead according to the predicted order conversion rate of each outbound call seat corresponding to each outbound call lead; The outbound call seat corresponding to the highest predicted order conversion rate of each outbound call lead is used as the recommended outbound call seat corresponding to each outbound call lead; The outbound call leads corresponding to each outbound call seat are counted according to the recommended outbound call seats corresponding to each outbound call lead, so as to obtain an outbound call lead recommendation queue corresponding to each outbound call seat.

4. The method for recommending outbound call leads according to claim 1, characterized in that: Based on the sample feature data of the outbound call agents, the sample feature data of the outbound call leads, the sample communication data between the outbound call agents and the outbound call leads, and the historical order conversion rate of each outbound call lead corresponding to each outbound call agent, the initial multimodal outbound call lead recommendation model is trained to obtain a trained multimodal outbound call lead recommendation model, including: Cross-process at least two of the sample feature data of the outbound call agent, the sample feature data of the outbound call clue, and the sample communication data of the outbound call agent and the outbound call clue to generate outbound cross-sample feature vector data; The outbound call cross-sample feature vector data is used as input, and the historical order conversion rate of each outbound call lead corresponding to each outbound call seat is used as output, and the initial multimodal outbound call lead recommendation model is trained to obtain a trained multimodal outbound call lead recommendation model.

5. The method for recommending outbound call leads according to claim 1, characterized in that: Cross-processing at least two of the feature data of the outbound call agent, the feature data of the outbound call clue, and the communication data between the outbound call agent and the outbound call clue to generate outbound cross feature vector data, including: Select at least two types of data from the feature data of the outbound call agent, the feature data of the outbound call clue, and the communication data between the outbound call agent and the outbound call clue, and construct a vector for each type of data; Performing feature-level cross processing on data of at least two vectors to obtain a first processing result; Performing element-level cross processing on data of at least two vectors to obtain a second processing result; Combining data of at least two vectors in pairs, performing low-order cross processing, and obtaining a third processing result; The first processing result, the second processing result and the third processing result are combined and transformed using a preset function to generate outbound call cross feature vector data.

6. A device for recommending outbound call leads, characterized in that: include: An acquisition module, used to acquire characteristic data of outbound call agents, characteristic data of outbound call leads, and communication data between outbound call agents and outbound call leads; A generating module, configured to perform cross processing on at least two of the feature data of the outbound call agent, the feature data of the outbound call clue, and the communication data between the outbound call agent and the outbound call clue, to generate outbound cross feature vector data; A prediction module, used for inputting the outbound call cross feature vector data into a pre-trained multimodal outbound call lead recommendation model, using the multimodal outbound call lead recommendation model to predict the order conversion rate of each outbound call lead corresponding to each outbound call seat, and obtaining the predicted order conversion rate of each outbound call lead corresponding to each outbound call seat; A recommendation module, configured to obtain an outbound call lead recommendation queue corresponding to each outbound call seat according to the predicted order conversion rate of each outbound call lead corresponding to each outbound call seat, and recommend outbound call leads based on the outbound call lead recommendation queue corresponding to each outbound call seat; The device further comprises a training module, wherein the training module is used for: Build an initial multimodal outbound call lead recommendation model; Collect sample characteristic data of outbound call agents, sample characteristic data of outbound call leads, sample communication data between outbound call agents and outbound call leads, and historical order conversion rates of each outbound call lead corresponding to each outbound call agent; Based on the sample feature data of the outbound call agents, the sample feature data of the outbound call leads, the sample communication data between the outbound call agents and the outbound call leads, and the historical order conversion rate of each outbound call lead corresponding to each outbound call agent, the initial multimodal outbound call lead recommendation model is trained to obtain a trained multimodal outbound call lead recommendation model.

7. The outbound call lead recommendation device according to claim 6, characterized in that: The recommendation module is also used to: Based on the outbound call lead recommendation queue corresponding to each outbound call seat, the corresponding outbound call lead is allocated to each outbound call seat; or The outbound call clue recommendation queue corresponding to each outbound call seat is provided to the corresponding outbound call seat, and the corresponding outbound call seat communicates with the outbound call clues in sequence according to the outbound call clue recommendation queue.

8. The outbound call lead recommendation device according to claim 6, characterized in that: The recommendation module is also used to: Determine the highest predicted order conversion rate corresponding to each outbound call lead according to the predicted order conversion rate of each outbound call seat corresponding to each outbound call lead; The outbound call seat corresponding to the highest predicted order conversion rate of each outbound call lead is used as the recommended outbound call seat corresponding to each outbound call lead; The outbound call leads corresponding to each outbound call seat are counted according to the recommended outbound call seats corresponding to each outbound call lead, so as to obtain an outbound call lead recommendation queue corresponding to each outbound call seat.

9. The outbound call lead recommendation device according to claim 6, characterized in that: The training module is also used to: Cross-process at least two of the sample feature data of the outbound call agent, the sample feature data of the outbound call clue, and the sample communication data of the outbound call agent and the outbound call clue to generate outbound cross-sample feature vector data; The outbound call cross-sample feature vector data is used as input, and the historical order conversion rate of each outbound call lead corresponding to each outbound call seat is used as output, and the initial multimodal outbound call lead recommendation model is trained to obtain a trained multimodal outbound call lead recommendation model.

10. The outbound call lead recommendation device according to claim 6, characterized in that: The generation module is also used for: Select at least two types of data from the feature data of the outbound call agent, the feature data of the outbound call clue, and the communication data between the outbound call agent and the outbound call clue, and construct a vector for each type of data; Performing feature-level cross processing on data of at least two vectors to obtain a first processing result; Performing element-level cross processing on data of at least two vectors to obtain a second processing result; Combining data of at least two vectors in pairs, performing low-order cross processing, and obtaining a third processing result; The first processing result, the second processing result and the third processing result are combined and transformed using a preset function to generate outbound call cross feature vector data.

11. An electronic device, characterized in that: The invention comprises a processor and a memory, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the method for recommending outbound call leads according to any one of claims 1 to 5.

12. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein the computer program is configured to execute the method for recommending outbound call leads according to any one of claims 1 to 5 when running.

Images