Data processing method and device, storage medium and electronic equipment

Abstract

The invention relates to a data processing method and device, a storage medium and electronic equipment. The data processing method comprises the steps that a plurality of outbound numbers and outbound information of each outbound number are acquired, the outbound numbers comprise a first outbound number and a second outbound number, and the outbound information of the first outbound number meets a preset rule; deleting the first outbound number; according to the outbound information of the second outbound number, determining a complaint probability of the second outbound number, the complaint probability being used for representing a probability of receiving a complaint after outbound through the second outbound number; and screening the second call-out numbers according to the complaint probability of the second call-out numbers to obtain a third call-out number. According to the data processing method and device, the storage medium and the electronic equipment, on the basis of identifying the potential complaint risk, the accuracy of screening the outbound numbers can be improved, and then the outbound success rate is improved based on the screened outbound numbers.

Description

Technical Field

[0001] The present disclosure relates to the field of data processing technology, and in particular, to a data processing method, device, storage medium and electronic device. Background Art

[0002] With the development of modern communication technology and the improvement of data mining capabilities, telemarketing has also developed. In the telemarketing scenario, you can create an outbound call list for business marketing and promotion. The outbound call list can include multiple outbound call numbers. Customer service can perform outbound call tasks based on the outbound call list, thereby achieving business marketing and promotion. Summary of the invention

[0003] The purpose of the present disclosure is to provide a data processing method, device, storage medium and electronic device, which can improve the accuracy of outbound number screening on the basis of identifying potential complaint risks, and then improve the outbound call success rate based on the screened outbound numbers.

[0004] In order to achieve the above-mentioned purpose, in a first aspect, the present disclosure provides a data processing method, including: obtaining multiple outbound call numbers and outbound call information of each of the outbound call numbers, the outbound call numbers including a first outbound call number and a second outbound call number, and the outbound call information of the first outbound call number meets a preset rule; deleting the first outbound call number; determining the complaint probability of the second outbound call number based on the outbound call information of the second outbound call number, the complaint probability being used to characterize the probability of receiving a complaint after an outbound call is made through the second outbound call number; and screening the second outbound call number based on the complaint probability of the second outbound call number to obtain a third outbound call number.

[0005] In a second aspect, the present disclosure provides a data processing device, including: an acquisition unit, used to acquire multiple outbound call numbers and outbound call information of each of the outbound call numbers, the outbound call numbers including a first outbound call number and a second outbound call number, and the outbound call information of the first outbound call number meets a preset rule; a determination unit, used to delete the first outbound call number, and determine the complaint probability of the second outbound call number based on the outbound call information of the second outbound call number, and the complaint probability is used to characterize the probability of receiving a complaint after calling through the second outbound call number; a screening unit, used to screen the second outbound call number according to the complaint probability of the second outbound call number to obtain a third outbound call number.

[0006] In a third aspect, the present disclosure provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the data processing method as described in the first aspect.

[0007] In a fourth aspect, the present disclosure provides an electronic device, comprising: a memory on which a computer program is stored; and a processor, configured to execute the computer program in the memory to implement the data processing method as described in the first aspect.

[0008] Through the above technical solution, multiple outbound call numbers and outbound call information of each outbound call number are obtained. Among the multiple outbound call numbers, the outbound call information of the first outbound call number meets the preset rules and can be directly deleted; for the second outbound call number, the complaint probability can be determined based on the corresponding outbound call information, and the second outbound call number can be screened based on the complaint probability. This technical solution can improve the accuracy of outbound call number screening by screening outbound call numbers through outbound call information, preset rules and complaint probability. In addition, screening outbound call numbers based on complaint probability can identify the potential complaint risk of outbound call numbers. Therefore, this technical solution can further improve the accuracy of outbound call number screening on the basis of identifying potential complaint risks, and then improve the outbound call success rate based on the screened outbound call numbers.

[0009] Other features and advantages of the present disclosure will be described in detail in the following detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] The accompanying drawings are used to provide a further understanding of the present disclosure and constitute a part of the specification. Together with the following specific embodiments, they are used to explain the present disclosure but do not constitute a limitation of the present disclosure. In the accompanying drawings: Figure 1 The figure is a structural block diagram of an outbound calling system according to an exemplary embodiment.

[0011] Figure 2 The invention is a flowchart of a data processing method according to an exemplary embodiment.

[0012] Figure 3 The figure is a schematic diagram showing outbound call information according to an exemplary embodiment.

[0013] Figure 4 The figure is a schematic diagram of a display interface of an outbound call list according to an exemplary embodiment.

[0014] Figure 5 The figure is a flowchart of list processing according to an exemplary embodiment.

[0015] Figure 6 The present invention is a flowchart of a model-based list processing according to an exemplary embodiment.

[0016] Figure 7 It is a structural block diagram of a data processing device according to an exemplary embodiment.

[0017] Figure 8It is a block diagram of an electronic device according to an exemplary embodiment. DETAILED DESCRIPTION

[0018] The specific implementation of the present disclosure is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described herein is only used to illustrate and explain the present disclosure, and is not used to limit the present disclosure.

[0019] It should be noted that all actions of acquiring signals, information or data in the present disclosure are carried out in compliance with the relevant data protection laws and policies of the country where the device is located and with the authorization given by the owner of the corresponding device.

[0020] With the development of modern communication technology and the improvement of data mining capabilities, telemarketing has also developed. In the telemarketing scenario, you can create an outbound call list for business marketing and promotion, and customer service can perform outbound call tasks based on the outbound call list, thereby achieving business marketing and promotion. For example, for enterprises in the financial, education, and communications industries, when a new service is launched in the business system or a new customer appears in the business system, business marketing and promotion can be carried out to the customer. The enterprise will create an outbound call list, which can be used by customer service to perform outbound call tasks to complete business marketing and promotion.

[0021] With the development of telemarketing, the market will also regulate telemarketing. For customers, they can take the initiative to complain about some unsatisfactory or non-compliant telemarketing. For the market, some non-compliant marketing methods will also be detected, for example: detecting whether there are frequent invalid outbound calls. Furthermore, the market's supervision of telemarketing will lead to a large number of customer service numbers being blocked, affecting the normal operating efficiency of the telemarketing industry, and in serious cases, affecting the corporate image. Therefore, after creating an outbound call list, it is necessary to screen the outbound numbers in the outbound call list to reduce frequent invalid outbound calls, save communication resources, and reduce user complaints and non-compliant outbound calling behaviors.

[0022] It is understandable that the embodiments of the present disclosure can be applied to not only the scenario of telemarketing, but also the scenario of telephone promotion. In the scenario of telephone promotion, the purpose of outbound calls may not be to promote consumption behavior of users, but to promote some unknown information to users. For example, popular science safety knowledge, popular science safety protection measures, etc.

[0023] In the related technology, there are three ways to screen outbound numbers. The first is manual screening, which relies on manual records and experience judgment, is inefficient, and the screening accuracy is difficult to guarantee. The second is based on rule matching screening, which only screens the list according to preset rules (such as empty numbers, shutdown, etc.), and cannot dynamically identify potential complaining users and customer intentions. The third is an intelligent filtering system based on a single indicator. Although it can use certain data analysis, it is limited to simple statistical models, such as the setting of the number of calls threshold, ignoring the complexity of customer behavior patterns, and has a high misjudgment rate.

[0024] It can be seen that the outbound call number screening method of the relevant technology has technical problems such as being unable to identify potential customer complaint risks and difficulty in ensuring screening accuracy.

[0025] Based on this, the embodiment of the present disclosure provides a technical solution, in which the accuracy of outbound call list screening can be improved by screening outbound call numbers based on outbound call information, preset rules and complaint probability. In addition, screening outbound call numbers based on complaint probability can identify the potential complaint risk of outbound call numbers. Therefore, the technical solution can improve the accuracy of outbound call number screening on the basis of identifying potential complaint risks, and further improve the outbound call success rate based on the screened outbound call numbers.

[0026] It is understandable that if outbound numbers are not screened, outbound calls may be made to users with a high probability of complaints, and the probability of rejection is relatively high, wasting communication resources during the outbound call process. However, after outbound number screening, the probability of rejection can be reduced, thereby saving communication resources. In addition, on the basis of reducing outbound calls to users with a high probability of complaints, the user experience can be improved.

[0027] Figure 1 is a structural block diagram of an outbound call system according to an exemplary embodiment. Figure 1 As shown, the outbound call system 100 includes: a client 101 and a server 102 .

[0028] The server 102 may be connected to the client 101 via a network. The network is used to provide a medium for a communication link between the server 102 and the client 101. The network may include various connection types, such as a wired communication link, a wireless communication link, etc., which is not limited in the embodiments of the present disclosure.

[0029] It should be understood that Figure 1The server 102, network and client 101 are merely illustrative. Depending on the implementation requirements, there may be any number of servers, networks and terminal devices. Exemplarily, the server 102 may be a physical server or a server cluster composed of multiple servers, and the client 101 may be installed on electronic devices such as mobile phones, tablets, desktop computers, and laptop computers. It is understood that the embodiments of the present disclosure may also allow multiple clients 101 to access the server 102 at the same time.

[0030] It can be understood that the outbound calling system can be applied to any enterprise or individual with telemarketing needs, and different enterprises or individuals can personalize the outbound calling system according to their own needs.

[0031] In some embodiments, the client 101 can serve as a customer service terminal, and the client 101 can use intelligent customer service or manual customer service.

[0032] In some embodiments, the user implements the execution of the outbound call task through the client 101 . After the outbound call task is executed, the corresponding outbound call information is recorded, and the recorded outbound call information can be synchronized to the server 102 .

[0033] In some embodiments, the server 102 can maintain the outbound call numbers. In the early telemarketing scenario, multiple preset outbound call numbers can be used. As the telemarketing business develops, the preset multiple outbound call numbers can be updated to respond to the changes in the outbound call numbers with the telemarketing application. Among them, different telemarketing businesses may correspond to different outbound call numbers, but the outbound call numbers corresponding to different telemarketing businesses can change with the telemarketing application.

[0034] In the embodiment of the present disclosure, multiple outbound call numbers may be stored in the form of an outbound call list, that is, multiple outbound call numbers are configured in the outbound call list.

[0035] In some embodiments, the outbound calling system 100 can be a part of the telemarketing platform, or a system independent of the telemarketing platform, and can have different configurations according to different application scenarios.

[0036] Therefore, in the outbound call system 100, it is important for the server 102 to effectively maintain the outbound call number. The maintenance method of the server 102 includes: outbound call number screening, outbound call number addition and outbound call number reset. Based on this, the embodiment of the present disclosure provides an outbound call number screening solution, which can be applied to the outbound call system 100.

[0037] Figure 2 is a flow chart of a data processing method according to an exemplary embodiment, the data processing method comprising the following steps: Step S21, obtaining multiple outbound call numbers and outbound call information of each outbound call number. The outbound call numbers include a first outbound call number and a second outbound call number, and the outbound call information of the first outbound call number meets a preset rule.

[0038] Step S22, deleting the first outbound calling number.

[0039] Step S23, determining the complaint probability of the second outbound number according to the outbound call information of the second outbound call number, where the complaint probability is used to characterize the probability of receiving a complaint after an outbound call is made through the second outbound call number.

[0040] Step S24, screening the second outbound number according to the complaint probability of the second outbound number to obtain a third outbound number.

[0041] It can be understood that step S21 to step S24 correspond to a complete screening process of an outbound number. In the outbound call system 100, the complete screening process of the outbound number can be executed multiple times.

[0042] For example, for an enterprise, the customer list it maintains will be updated as the business system develops. Based on the update of the customer list, the outbound call list can be updated accordingly. Also, after the customer service executes the outbound call task, a corresponding outbound call task execution result will be generated. Based on the execution result, the outbound call list can also be updated accordingly. Based on the need to update the outbound call list, the outbound call numbers can be screened. Therefore, in these examples, the complete screening process of the outbound call numbers of the embodiment of the present disclosure can be executed.

[0043] Therefore, the server 102 may configure corresponding outbound call number update conditions, and when the outbound call number update conditions are met, the server 102 is triggered to update the outbound call number.

[0044] In step S21, the outbound call information of the first outbound call number satisfies the preset rule. Therefore, it can be determined whether the outbound call number belongs to the first outbound call number or the second outbound call number through the preset rule and the outbound call information of each outbound call number.

[0045] In some embodiments, the outbound call information may include outbound call information, and the outbound call information may be outbound call information recorded in a historical record.

[0046] For example, after executing an outbound call task, corresponding outbound call information is recorded, and the recorded information can be used as outbound call information, which can be used to describe the outbound call status of the outbound call number.

[0047] In the early stage of telemarketing, the outbound call information may contain no data or only contain less data. As the telemarketing continues, the outbound call information is updated and the amount of data gradually increases. For example, whenever a client 101 generates a new outbound call record, the outbound call number and call information corresponding to the outbound call record are stored as the outbound call information.

[0048] Regarding outbound call information, it can be implemented in different ways according to the requirements of different application scenarios.

[0049] In some embodiments, the outbound call information includes at least one of the following: call information, number attribute information, and event marker information.

[0050] In some embodiments, the outbound call information may be information that can be directly obtained or information obtained through data analysis. For example, number attribute information is information that can be directly obtained; call information and event marker information may be information that can be directly obtained or information obtained through data analysis.

[0051] The multiple outbound call numbers may include the phone numbers of new customers and the phone numbers of old customers. New customers may be customers who have not been called outbound; and old customers may be customers who have been called outbound.

[0052] In some embodiments, if the outbound call number is a new outbound call number, the outbound call information of other outbound call numbers can be used to determine the outbound call information of the new outbound call number. Alternatively, the outbound call information of the new outbound call number can be determined as blank outbound call information. Then, when performing the preset rule judgment, it can be determined by default that the preset rule is not satisfied, so as to retain the new outbound call number as much as possible.

[0053] In some embodiments, a similar outbound call number to the new outbound call number is determined from a plurality of outbound call numbers having outbound call information, and the outbound call information of the similar outbound call number is determined as the outbound call information of the new outbound call number.

[0054] In some embodiments, it is first determined whether the new outbound call number has a similar outbound call number. If so, the outbound call information corresponding to the similar outbound call number is determined as the outbound call information corresponding to the new outbound call number. If not, the outbound call information of the new outbound call number is determined as blank outbound call information.

[0055] In some embodiments, each outbound call number is also configured with some basic information, such as: the location of the outbound call number and the user to which the outbound call number belongs, etc. These basic information can be obtained without making an outbound call, and these basic information can be used as part of the outbound call information.

[0056] Therefore, whether the outbound numbers are similar can be determined based on the similarity between the basic information of the outbound numbers. For example, the similarity between the basic information of two outbound numbers is determined, and when the similarity is greater than a preset similarity (for example, 95%), the two outbound numbers are determined to be similar outbound numbers. Through the above implementation, when the outbound call number corresponds to a new customer, the outbound call information of the outbound call number can also be determined, and the outbound call information may be a predicted outbound call information, which can be used to determine whether the outbound call number should be deleted. Thus, complaints from new customers can be reduced as much as possible, so as to improve the success rate of telemarketing for new customers.

[0057] This implementation method can be applicable to situations where there are many new customers. If the number of outbound numbers corresponding to new customers is greater than the number of outbound numbers corresponding to old customers, this implementation method can be adopted to minimize complaints from new customers and increase the success rate of telemarketing to new customers.

[0058] Since the outbound call information can represent the outbound call history of the outbound call number, multiple outbound call numbers can be preliminarily screened by using preset rules and combining the outbound call history of the outbound call number to filter out outbound call numbers that are obviously not valuable for outbound calls. For example, outbound call numbers that are likely to receive complaints after outbound calls and outbound calls that are obviously not answered.

[0059] Therefore, among the multiple outbound call numbers, the outbound call information of the first outbound call number meets the preset rule, and accordingly, in step S22, the first outbound call number is deleted.

[0060] As an optional implementation, the preset rules include at least one of the following: the outbound number cannot be connected; the location of the outbound number is the preset location; the outbound time is within the preset outbound time range; the user type to which the outbound number belongs is the preset user type; the outbound number has a complaint event mark; the outbound number has a rejection event mark.

[0061] In the above preset rules, the outbound number cannot be connected and the outbound call time belongs to the preset outbound call time range, which are preset rules for call information; the place of origin of the outbound number is the preset place of origin and the user type of the outbound number is the preset user type, which are preset rules for number attribute information; the outbound number has a complaint event mark and the outbound number has a rejection event mark, which are preset rules for event mark information.

[0062] The preset location may be a location corresponding to a region where outbound calls are not allowed, and may be preset accordingly according to different application scenarios. The preset outbound call time range may be a time period where outbound calls are not allowed, and may be preset accordingly according to different application scenarios. The preset user type may be a special group of people who are not allowed to make outbound calls, such as older people and younger people.

[0063] Based on the above preset rules, the outbound call information may correspondingly include at least one of the following information: call records, location of the outbound call number, user type, complaint event mark, and rejection event mark.

[0064] Regarding call records, they may include information such as whether the call was connected, whether the call was answered, call duration, and outbound call time. Whether the call was connected is different from whether the call was answered. For example, after an outbound call, if the number is prompted to be unavailable or out of service, it can be recorded as a missed call; after an outbound call, if the number is prompted to be connected (for example, a connected prompt tone is heard), but the call is not answered, it can be recorded as a missed call.

[0065] Thus, an empty number or a suspended number that cannot be answered can be the first outbound number. Also, an outbound number whose outbound time falls within a preset outbound time range can be the first outbound number.

[0066] The location of the outbound call number is the basic information of the outbound call number. The user type is the basic information of the user to which the outbound call number belongs.

[0067] Thus, the outbound call number of the area where outbound calls are not allowed can be the first outbound call number. Also, the outbound call number of the user type that does not allow outbound calls can be the first outbound call number.

[0068] The complaint event mark and the rejection event mark can be information recorded by the outbound call task executor. The complaint event mark can indicate whether there is a complaint event for the outbound call number, and the rejection event mark can indicate whether there is a rejection mark for the outbound call number. Among them, the complaint event can occur during the outbound call process or after the outbound call. The rejection event can occur during the outbound call process, and can be a rejection event expressed by the call content or a rejection event represented by the hang-up behavior.

[0069] Regarding the complaint event mark and the rejection event mark, it can also be the mark information obtained by the outbound call system 100 through data analysis. For example, for each outbound call, the client 101 will record and generate a call recording. By converting the call recording into text, the text information of the call content can be obtained. By checking whether the call content text information includes the complaint event key field and the rejection event key field, the corresponding complaint event mark and rejection event mark can be generated. The key fields of the complaint event are, for example: "complaint", "report to the leader", "Your attitude is really bad", "This service is very inconvenient to use", etc. The key fields of the rejection event are, for example: "Don't call me again", "Not interested", etc.

[0070] For example, Figure 3 is a schematic diagram showing an outbound call message according to an exemplary embodiment. Figure 3As shown, the outbound call information is provided with a plurality of outbound call information items, and different information is recorded under the plurality of outbound call information items respectively. Figure 3 The multiple outbound call information items include: outbound call number, outbound call number location, outbound call number user type, outbound call time, whether the call is connected (if the call is not connected, the reason needs to be noted), whether the call is answered (if the call is not answered, the reason needs to be noted), complaint event mark, and rejection event mark.

[0071] It should be understood that for an outbound call number, there may be more than one outbound call. If there are multiple outbound calls, the multiple outbound calls will be recorded separately. Therefore, the outbound call number, the location of the outbound call number, and the user type to which the outbound call number belongs may be fixed information. The outbound call time, whether it is connected (if not answered, the reason needs to be noted), whether it is answered (if not answered, the reason needs to be noted), complaint event mark, rejection event mark, etc. may include multiple data respectively.

[0072] For outbound number X1, the location is XX region, the user type of the outbound number is the outbound call type, the outbound time is time T1, whether it is connected is connected, whether it is answered is not answered, the complaint event mark is 0 (indicating no complaint event), and the rejection event mark is 1 (indicating a rejection event). Accordingly, if you need to delete an outbound number with a rejection event mark, the outbound number X1 will be deleted.

[0073] For outbound number X2, the location is YY region, the user type of the outbound number is the outbound call type, the outbound time is time T2, whether it is connected is not connected (empty number), whether it is answered is not answered, the complaint event mark is 0 (indicating no complaint event), and the rejection event mark is 0 (indicating a rejection event). Accordingly, if you need to delete the outbound number that cannot be connected, the outbound number X2 will be deleted.

[0074] For outbound number X3, the location is ZZ region, the user type of the outbound number is the outbound call type, the outbound time is time T3, whether connected is connected, whether answered is answered, the complaint event mark is 1 (indicating a complaint event), and the rejection event mark is 0 (indicating no rejection event). Accordingly, if you need to delete an outbound number with a complaint event mark, the outbound number X3 will be deleted.

[0075] Understandably, Figure 3 This is only an exemplary introduction. In actual application scenarios, more outbound call numbers and corresponding outbound call information may be included, which are not listed here one by one.

[0076] In addition, in some embodiments, the multiple outbound call numbers may also include an incorrect outbound call number, for example, user A's phone number is X1, but the recorded phone number is X2. In this case, the outbound call information may also include an incorrect number mark. The marking method of the incorrect number mark may refer to the marking method of the complaint event and the rejection event. Among them, the key field of the incorrect number event may be, for example, "You dialed the wrong number", "I am not XX", etc.

[0077] Correspondingly, the preset rules may also include: there is a wrong number mark for the outbound call number.

[0078] It can be understood that the outbound call information may include more information, and the preset rules may include more rules satisfied by the outbound call numbers to be deleted, which are not limited in the embodiments of the present disclosure.

[0079] It can be seen from the introduction of the above implementation modes that, through the preset rules, some outbound call numbers that obviously have no outbound call value can be initially deleted.

[0080] For example: Invalid number filtering: Check whether the phone numbers in the list are empty, out of service, wrong numbers, etc. Such numbers can be deleted directly. Compliance review: According to relevant regulatory provisions or industry regulations, the list is screened for compliance, and phone numbers of time periods, regions or special groups that are not allowed to be dialed are deleted. Historical complaint records: Filter out customers with a history of complaints. These customers may be more resistant to telemarketing activities, so they will be temporarily removed in the initial filtering stage. Multiple rejections or hang-ups: Phone numbers that have been rejected multiple times recently or frequently hung up in a short period of time indicate that the customer may be unwilling to accept telemarketing and should also be temporarily removed from the list.

[0081] Therefore, on this basis, the screening accuracy of outbound numbers can be improved, more accurate outbound numbers can be obtained, and the success rate of telephone marketing can be improved.

[0082] In the disclosed embodiment, the first outbound number may be an outbound number with an obvious complaint tendency, and correspondingly, the second outbound number may be an outbound number with no obvious complaint tendency.

[0083] After step S22, a preprocessing operation may be performed for the second outbound call number, and the preprocessing operation is used to reduce abnormal or invalid outbound call numbers to obtain a final outbound call number.

[0084] As an optional implementation, the pre-processing operation may include: deleting invalid or erroneous outbound call numbers, filling missing values, and identifying and processing abnormal values.

[0085] Invalid or incorrect outbound call numbers can be duplicate outbound call numbers, or, for the same outbound call number, there are different information corresponding to it. In this case, the different information is checked and only the correct information is retained. Fill in missing values, for example, for an outbound call number that is missing some outbound call information, complete the outbound call information of the outbound call number. Identify and process abnormal values, for example, if the number of outbound calls of an outbound call number is less than a certain value, then the outbound call number with a high number of outbound calls can be deleted.

[0086] In step S23, the complaint probability is used to characterize the probability of receiving a complaint after the outbound call is made through the second outbound call number. The complaint may be received during the outbound call or after the outbound call is ended.

[0087] As an optional implementation, step S23 includes: determining the outbound calling features of the second outbound calling number based on the outbound calling information of the second outbound calling number; inputting the outbound calling features of the second outbound calling number into a pre-trained model, and outputting the complaint probability of the second outbound calling number.

[0088] In this embodiment, feature extraction is performed on the outbound call information corresponding to the plurality of second outbound call numbers to obtain the outbound call feature of the second outbound call number. The outbound call feature may be a feature related to outbound call complaints, and the feature extraction process may be understood as screening out information used for model prediction of complaint probability from the outbound call information.

[0089] As an optional implementation, the target call features include at least one of the following features: the outbound call features include at least one of the following features: the outbound call duration; the outbound call interval duration, the outbound call interval time is used to characterize the interval duration between two adjacent outbound calls; the total number of outbound calls and the number of outbound calls answered; the evaluation of the outbound calling user; the evaluation of the outbound answering user and the number of historical outbound call complaints; the preset outbound call content, the preset outbound call content is used to characterize the negative evaluation of the outbound answering user.

[0090] The duration of an outbound call can be determined based on the outbound call information. If the outbound call duration is short, the probability of complaint may be low. If the outbound call duration is long, the probability of complaint may be further determined based on the specific call content. Therefore, the duration of an outbound call can be used as an outbound call feature.

[0091] The duration between outbound calls can be determined by the outbound call times in the outbound call information. If the duration between two consecutive outbound calls is long, the user is unlikely to be disgusted, and the probability of complaint may be low. If the duration between two consecutive outbound calls is short, the user may be disgusted, and the probability of complaint may be high. Therefore, the duration between outbound calls can be used as an outbound call feature.

[0092] The number of outbound calls and the number of answered outbound calls can be determined through outbound call information. The number of outbound calls is the total number of outbound calls, and the number of answered outbound calls is the number of answered outbound calls. It can be understood that if the number of outbound calls is large and the number of answered outbound calls is also large, the user may have a high degree of cooperation and a low probability of complaint. If the number of outbound calls is large and the number of answered outbound calls is small, the user may not have a high degree of cooperation and a high probability of complaint. Therefore, the ratio of the number of answered outbound calls to the number of outbound calls can reflect the probability of complaints, and the number of outbound calls and the number of answered outbound calls can be used as outbound call features.

[0093] The evaluation of the outbound call user may be the evaluation of the outbound call user feedback. The outbound call user may be a customer service representative, an agent, etc. For example, after each outbound call, the outbound call user may record the evaluation. Therefore, according to the specific content of the outbound call user evaluation, the complaint probability may be predicted, and the outbound call user evaluation may be used as an outbound call feature.

[0094] The outbound call answering user rating can be the rating given by the outbound call answering user after the outbound call. The number of historical complaints can be the information obtained by analyzing the historical call information. It can be understood that the higher the outbound call answering user rating, the higher the user satisfaction and the lower the probability of complaints. The more historical complaints, the lower the user satisfaction and the higher the probability of complaints. Therefore, both of these two pieces of information can be used as outbound call features.

[0095] The preset outbound call content can be obtained by analyzing the outbound call information. For example, by converting the call recording into text call content, the text call content is analyzed. If it contains some fields representing negative comments, these fields are extracted as the preset outbound call content. The call content representing negative comments may include: swear words, dissatisfied interjections, disgusted interjections, etc. It can be understood that when the outbound call answering user has negative comments, the probability of complaint is higher. Therefore, the preset outbound call content can be used as an outbound call feature.

[0096] From the introduction of the outbound call features above, we can see that the outbound call features can reflect the user's complaint tendency or possibility. Therefore, based on the outbound call features, the pre-trained model can predict the complaint probability.

[0097] In different application scenarios, more outbound call features can be configured according to different needs, which are not limited here.

[0098] In some embodiments, the pre-trained prediction model may be a logistic regression model; or, it may be a deep learning model, a large language model, etc., which is not limited in the present disclosure.

[0099] Furthermore, the pre-trained model is used to output the complaint probability based on the input outbound call features.

[0100] As an optional implementation, the training process of the pre-trained model may include: obtaining a training data set, the training data set includes multiple training samples, each training sample includes a sample outbound call feature and a complaint label of the sample outbound call feature, the sample outbound call feature is a feature extracted from the outbound call information of the sample outbound call number, and the complaint label is used to characterize the outbound call complaint probability of the sample outbound call number; according to the training data set, training the model to be trained to obtain the pre-trained model. .

[0101] In this implementation, the sample outbound call feature may correspond to the aforementioned outbound call feature. Therefore, the implementation of the sample outbound call feature may refer to the implementation of the aforementioned outbound call feature.

[0102] In some embodiments, the number of features of the sample outbound call features may be greater than the number of the aforementioned outbound call features. That is, when training the model, more outbound call features may be selected as much as possible to improve the generalization ability of the model, thereby ensuring the accuracy of the complaint probability prediction.

[0103] In some embodiments, the complaint label may be a manually marked complaint probability, which is used to characterize the outbound complaint situation of the sample outbound number corresponding to the sample outbound feature.

[0104] In some embodiments, multiple training samples can be obtained by selecting samples based on multiple sample outbound call numbers and corresponding sample outbound call information. For example, outbound call information corresponding to an outbound call number without a complaint event mark is selected, and sample outbound call features are extracted from the outbound call information corresponding to the outbound call number without a complaint event mark. In addition, complaint labels are set manually.

[0105] The training data set is used to train the model to be trained to obtain a pre-trained model, so that the pre-trained model can identify potential complaint tendencies. Since the second outbound number is an outbound number without obvious complaint tendencies, the pre-trained model can be used to predict the complaint probability from the level of potential complaint tendencies, and then the second outbound number can be further screened based on the complaint probability.

[0106] Regarding the pre-trained model, some model verification and tuning implementation methods can also be adopted during its training process to ensure the accuracy of the model as much as possible.

[0107] For example, a cross-validation method can be used to verify the performance of the model. For example, obtain a training data set and a validation data set (refer to the method for obtaining a training data set). In order to avoid overfitting and ensure the reliability and effectiveness of the validation results, a k-fold cross-validation strategy can be used. Specifically, the data set is evenly divided into k subsets, and the model is trained with k-1 subsets each time, and the remaining subsets are used to verify the model performance. This process is repeated k times, and each subset will have the opportunity to be verified as a validation set, and finally the overall performance of the model is evaluated by combining the k validation results.

[0108] For example, model tuning can be achieved through hyperparameter adjustment. For example, the learning rate hyperparameter is used for model tuning. The learning rate determines the step size of the model to update the weights during the training process, and its optimization is a key step to improve model performance. First, set an initial learning rate for the logistic regression model, and then adjust the learning rate by observing the performance of the model on the validation set during the training process. Specifically, the learning rate is dynamically adjusted according to the changing trend of the evaluation indicators on the validation set (such as AUC (Area UnderCurve, defined as the area under the ROC (Receiver Operating Characteristic Curve) curve and the coordinate axis), precision, recall, etc.). When the model performance on the validation set no longer improves significantly or begins to fluctuate, reduce the learning rate to seek a more refined parameter space search; conversely, if the model converges too slowly or underfitting occurs, try to increase the learning rate to speed up the training process.

[0109] Through cross-validation and learning rate tuning strategies, the accuracy of the model in predicting the probability of complaints can be continuously improved, thereby effectively reducing invalid and high-risk calls and improving the success rate and compliance of telemarketing activities.

[0110] In the disclosed embodiment, the pre-trained model is updated accordingly with the update of the outbound call data. For example, when new outbound call data is added to the outbound call data, training samples are generated according to the new data to synchronously update the model, so that the model can be continuously improved and the accuracy of the model is guaranteed.

[0111] Furthermore, in step S24, the second outbound number is screened according to the complaint probability of the second outbound number to obtain a third outbound number.

[0112] As an optional implementation, step S24 includes: deleting the second outbound call number whose complaint probability is greater than the preset complaint probability to obtain a third outbound call number.

[0113] In this implementation, the second outbound call number with a higher complaint probability is directly deleted. The preset complaint probability may be a complaint probability that indicates a high complaint tendency. For example, if the complaint probability ranges from 0 to 100%, the preset complaint probability may be a probability value between 90% and 95%.

[0114] In other embodiments, the second outbound calling numbers may be sorted from high to low according to the complaint probability, and then the second outbound calling numbers with high complaint probability rankings (eg, top 20%) may be deleted.

[0115] Alternatively, based on the complaint probability, other screening methods may be used, which are not limited here.

[0116] It is understandable that for some new customers, the predicted complaint probability is low due to insufficient or no outbound call features. Also, for some old customers, their needs are changing at any time, and the predicted complaint probability is not high, but there may actually be a tendency to complain. Therefore, there is uncertainty for these outbound call numbers.

[0117] Therefore, a corresponding identifier may be configured for the second outbound call number as a kind of indicative information.

[0118] Therefore, the data processing method may also include: in response to the complaint probability of the second outbound number being greater than the first preset probability, configuring a first preset identifier for the second outbound number, the first preset identifier being used to characterize that the user type to which the outbound number belongs is a first type of user; and / or, in response to the complaint probability of the second outbound number being less than the second preset probability, configuring a second preset identifier for the second outbound number, the second preset identifier being used to characterize that the user to which the outbound number belongs is a second type of user, and the second complaint probability is less than the first complaint probability.

[0119] In this implementation, if the preset complaint probability is a value between 90% and 95%, the first preset probability may be a value between 50% and 90%. It can be understood that when the complaint probability is greater than the first preset probability, it indicates that the customer to whom the outbound number belongs is uncertain, and that the customer to whom the outbound number belongs is likely to be an old customer. Based on these two aspects, the customer may be marked as a high-quality customer, and the success rate of telemarketing is higher than that of other customers.

[0120] If the preset complaint probability is a value between 90% and 95%, and the first preset probability is a value between 50% and 90%, then the second preset probability may be a value between 0% and 50%. It can be understood that when the complaint probability is less than the second preset probability, it means that the customer to whom the outbound number belongs has a smaller complaint probability, and the customer to whom the outbound number belongs is more likely to be a new customer. Based on these two aspects, it can be marked as a new customer.

[0121] Therefore, the first type of users may be high-quality customers, and the second type of users may be new customers.

[0122] Furthermore, the first preset identifier and the second preset identifier may have different forms, which are not limited in the embodiments of the present disclosure. For example, if the outbound call number is displayed in the form of a list, the first preset identifier and the second preset identifier may be used as a reference item in the list, which is displayed corresponding to the corresponding outbound call number. Alternatively, the first preset identifier and the second preset identifier may be the identification information displayed after the user triggers the corresponding outbound call number. Furthermore, after configuring a corresponding identifier for the outbound call number, multiple outbound call numbers can be output so that the corresponding user can perform outbound call tasks according to the multiple outbound call numbers.

[0123] For example, multiple outbound numbers are fed back to the user who performs the outbound call task in the form of a list. In some embodiments, the outbound call task is generally divided, so when the list is fed back, the list can be synchronously divided according to the division of the outbound call task, so that the multiple outbound call lists after division can be fed back to the relevant personnel respectively.

[0124] Figure 4 is a schematic diagram of a display interface of an outbound call list according to an exemplary embodiment. Figure 4 As shown, an outbound call list interface 40 is displayed on the client 101, and at least multiple outbound call numbers are displayed in the outbound call list interface 40. The display form here is only a schematic and does not constitute a limitation on the embodiment of the present disclosure. When the user clicks on one of the outbound call numbers, if the outbound call number corresponds to the first preset identifier or the second preset identifier, an identification information interface 401 is displayed on the client 101, and the first preset identifier or the second preset identifier is displayed in the identification information interface 401. If the outbound call number does not correspond to the first preset identifier or the second preset identifier, when the user clicks on the outbound call number, no corresponding response will be made. In this way, the user can adopt a more reasonable outbound call strategy based on the displayed identification information.

[0125] For example, if the identification information of the outbound call number is the first preset identification, the user who performs the outbound call task can give priority to recommending new services to the user. Also, during the call, the user is guided to give corresponding feedback or evaluation, and by recording these feedback or evaluations, the success rate of the telemarketing business can be improved.

[0126] For example, if the identification information of the outbound call number is the second preset identification, the user who performs the outbound call task can first introduce the enterprise service to the user, and after gaining the trust of the user, introduce or recommend new services, etc. Through more tactful telemarketing methods, the trust of new customers can be gained as much as possible, and the coverage of telemarketing services can be continuously expanded.

[0127] Figure 5 FIG. 1 is a flowchart of a list processing according to an exemplary embodiment. The list processing flow can be applied to the processing of an outbound call list in a telemarketing scenario. Figure 5 As shown in the figure, the list processing process includes: Input original list: (Outbound call) The system receives the original customer list (which may correspond to the original outbound call list) containing the phone number to be dialed (which may correspond to the original outbound call number) and the dialing record data (which may correspond to the historical outbound call data).

[0128] Call record analysis: In-depth analysis of received call record data, including but not limited to key indicators such as call duration, answer rate, hang-up reason, customer feedback, and whether there is swearing.

[0129] Preliminary filter list: Screen and remove numbers with obvious complaint tendencies, high frequency of rejections, etc.

[0130] Secondary filtering list: further use the call record data for refined analysis to identify customers who appear normal on the surface but may actually be at potential risk of complaints.

[0131] Comprehensive assessment module: Integrate multiple analysis results to establish a multi-dimensional risk scoring system, and provide a comprehensive risk assessment score for each number. The comprehensive assessment module can be understood as an assessment module based on the probability of complaints, and the risk assessment score can be determined based on the probability of complaints. For example, the higher the probability of complaints, the higher the risk assessment score, indicating a greater risk.

[0132] Generate optimized list: Based on the risk assessment score, the system automatically eliminates high-risk numbers and generates an optimized list of calls to be dialed (corresponding to the target outbound call list).

[0133] Output the final available list: Provide the strictly filtered and optimized list to the telemarketing team as the actual dialing target.

[0134] Figure 6 is a flowchart of a list processing based on a model according to an exemplary embodiment. Figure 5 The list processing flow shown in the figure adds the model application flow. Figure 6 As shown in the figure, the list processing process includes: Import historical data: The (outbound call) system first imports rich and detailed historical call record data from the telemarketing database, covering a variety of information such as call duration, answering status, reasons for hanging up, customer feedback and handling intentions.

[0135] Data initial screening and preprocessing are used to implement initial screening and preprocessing of data. The specific implementation method can refer to the above-mentioned embodiment. Feature Engineering: In this phase, based on business understanding and data insights, additional features are created that help predict the target variable (complaint likelihood), such as call intervals, call frequency in different time periods, customer service reviews, swear words, etc.

[0136] Model construction: Logistic regression model is used for model construction. Logistic regression is a statistical modeling technique widely used in binary classification problems, especially in predicting the probability of event occurrence, which has the advantage of being intuitive and easy to understand.

[0137] Model building can specifically include: Feature selection: Select factors such as call duration, service quality score, and historical complaint records from the cleaned and organized data set as feature inputs. Model training: Use the training data to fit a logistic regression model, where the model learns the relationship between features and complaints, and outputs the probability of complaints for each sample. In turn, the trained logistic regression model can be used to predict new telemarketing activity data, output the probability value of each customer complaint, and then set a complaint probability threshold to distinguish high-risk and low-risk customers.

[0138] Model training: Use the labeled historical data to train the constructed model, and continuously optimize the model capabilities through model verification and tuning. For the specific implementation of model verification and tuning, refer to the above embodiments.

[0139] Real-time monitoring and data flow: The system is connected to the telemarketing platform in real time to obtain new data after each call, and incorporate this data into the model for real-time analysis, building a real-time data flow between the two systems.

[0140] Real-time data acquisition and transmission: Real-time monitoring first involves seamless integration with the telemarketing platform. The system needs to access the new call record data stream generated by the telemarketing platform in real time. Whenever a new call ends, whether it is successfully connected, no one answers, or the customer actively hangs up, the system will immediately capture all relevant information about the call, such as call duration, customer response, reason for hanging up, customer emotions, etc.

[0141] Real-time data processing: The real-time call data obtained needs to go through the same preprocessing steps as when training the model, including data cleaning, missing value filling, outlier detection and processing, feature engineering, etc., to ensure that the new data is consistent with the data format used during model training to facilitate predictive analysis.

[0142] Real-time prediction and evaluation: The processed real-time data is fed into the pre-trained logistic regression model for real-time prediction. The model will calculate the probability of customer complaints based on the characteristics of the new data. The logistic regression model updates the model parameters, especially the adjustment of the learning rate, so that the model can timely optimize its prediction ability based on the latest data feedback.

[0143] Dynamic list filtering: The call record data obtained in real time is predicted by the model in real time to obtain the probability of each customer complaint or comprehensive risk score. The system dynamically adjusts the telemarketing list accordingly, and directly removes high-risk customers from the call list to reduce the potential complaint risk. The real-time monitoring and dynamic filtering mechanism constitute a complete feedback loop. The system can not only respond to the real-time changes in the telemarketing list marketing process in real time, but also continuously optimize the model performance based on the new data that is constantly accumulated. This means that as the telemarketing activities proceed, the model's prediction accuracy of customer complaint risks will continue to improve, thereby further improving the effect of list filtering and the work efficiency of the telemarketing team.

[0144] Generate optimized lists: Based on the model prediction results and the company's telemarketing strategy, the system automatically generates an optimized telemarketing list to ensure that the customers on the list not only have high potential purchase intentions but also have a low overall complaint risk, thereby improving the telemarketing team's work efficiency and customer satisfaction.

[0145] Through the introduction of the above embodiments, the technical solutions of the embodiments of the present disclosure can bring the following technical effects in practical applications: Improve telemarketing efficiency: By intelligently analyzing historical outbound call data and filtering out invalid numbers, customers with high complaint risks, and customers who are unwilling to answer calls in the short term, the telemarketing team can focus on potentially valuable customers, thereby significantly improving call success rate and business conversion rate.

[0146] Reduce the complaint rate: By predicting the possibility of customer complaints in real time and screening the list accordingly, it reduces the disturbance to customers who are sensitive or disgusted with telemarketing, thereby significantly reducing the occurrence of customer complaints and maintaining the corporate image.

[0147] Save resources: Avoid invalid calls and high-risk calls, reduce invalid labor costs and unnecessary communication expenses, while also reducing the pressure on the customer service team to handle complaints and optimizing resource allocation.

[0148] Improve customer experience: Only customers who are more receptive to telemarketing will receive calls, making telemarketing activities more targeted and personalized, thereby improving customer satisfaction.

[0149] Strengthen compliance: follow relevant market supervision rules, avoid harassing those who are prohibited or restricted from telemarketing, and enhance the competitiveness, development and stability of enterprises in the market.

[0150] Figure 7 is a structural block diagram of a data processing device according to an exemplary embodiment. Figure 7 As shown, the data processing device includes: an acquisition unit 701, used to acquire multiple outbound call numbers and outbound call information of each of the outbound call numbers, the outbound call numbers include a first outbound call number and a second outbound call number, and the outbound call information of the first outbound call number meets a preset rule; a determination unit 702, used to delete the first outbound call number, and determine the complaint probability of the second outbound call number according to the outbound call information of the second outbound call number, and the complaint probability is used to characterize the probability of receiving a complaint after calling through the second outbound call number; a screening unit 703, used to screen the second outbound call number according to the complaint probability of the second outbound call number to obtain a third outbound call number.

[0151] Optionally, the determination unit 702 is further used to determine the outbound calling features of the second outbound calling number based on the outbound calling information of the second outbound calling number; input the outbound calling features of the second outbound calling number into a pre-trained model, and output the complaint probability of the second outbound calling number.

[0152] Optionally, the device includes a training unit, which is used to: obtain a training data set, the training data set includes multiple training samples, each training sample includes a sample outbound call feature and a complaint label of the sample outbound call feature, the sample outbound call feature is a feature extracted from the outbound call information of the sample outbound call number, and the complaint label is used to characterize the outbound call complaint probability of the sample outbound call number; according to the training data set, train the model to be trained to obtain the pre-trained model.

[0153] Optionally, the outbound call features include at least one of the following features: outbound call duration; outbound call interval duration, the outbound call interval time is used to characterize the interval duration between two adjacent outbound calls; the total number of outbound calls and the number of outbound calls answered; the evaluation of the outbound calling user; the evaluation of the outbound calling user and the number of historical outbound calling complaints; preset outbound call content, the preset outbound call content is used to characterize the negative evaluation of the outbound calling user. Optionally, the device also includes a configuration unit for configuring a first preset identifier for the second outbound number in response to the complaint probability of the second outbound number being greater than a first preset probability, the first preset identifier being used to characterize that the user type to which the outbound number belongs is a first type of user; and / or, in response to the complaint probability of the second outbound number being less than a second preset probability, configuring a second preset identifier for the second outbound number, the second preset identifier being used to characterize that the user to which the outbound number belongs is a second type of user, and the second complaint probability is less than the first complaint probability.

[0154] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0155] Figure 8 FIG. 8 is a block diagram of an electronic device 800 according to an exemplary embodiment. Figure 8 As shown, the electronic device 800 may include: a processor 801 , a memory 802 . The electronic device 800 may also include one or more of a multimedia component 803 , an input / output (I / O) interface 804 , and a communication component 805 .

[0156] The processor 801 is used to control the overall operation of the electronic device 800 to complete all or part of the steps in the above-mentioned data processing method. The memory 802 is used to store various types of data to support the operation of the electronic device 800. For example, these data may include instructions for any application or method used to operate on the electronic device 800, and application-related data, such as contact data, messages sent and received, pictures, audio, video, etc. The memory 802 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read-Only Memory, referred to as EPROM), programmable read-only memory (Programmable Read-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic memory, flash memory, magnetic disk or optical disk. The multimedia component 803 may include a screen and an audio component. The screen may be, for example, a touch screen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signal may be further stored in the memory 802 or sent through the communication component 805. The audio component also includes at least one speaker for outputting audio signals. The I / O interface 804 provides an interface between the processor 801 and other interface modules, and the above-mentioned other interface modules may be a keyboard, a mouse, buttons, etc. These buttons may be virtual buttons or physical buttons. The communication component 805 is used for wired or wireless communication between the electronic device 800 and other devices. Wireless communication, such as Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G or 4G, or a combination of one or more of them, so the corresponding communication component 805 may include: Wi-Fi module, Bluetooth module, NFC module.

[0157] In an exemplary embodiment, the electronic device 800 can be implemented by one or more application specific integrated circuits (ASIC), digital signal processors (DSP), digital signal processing devices (DSPD), programmable logic devices (PLD), field programmable gate arrays (FPGA), controllers, microcontrollers, microprocessors or other electronic components to execute the above-mentioned data processing method.

[0158] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided, and when the program instructions are executed by the processor, the steps of the above-mentioned data processing method are implemented. For example, the computer-readable storage medium can be the above-mentioned memory 802 including program instructions, and the above-mentioned program instructions can be executed by the processor 801 of the electronic device 800 to complete the above-mentioned data processing method.

[0159] The preferred embodiments of the present disclosure are described in detail above in conjunction with the accompanying drawings; however, the present disclosure is not limited to the specific details in the above embodiments. Within the technical concept of the present disclosure, a variety of simple modifications can be made to the technical solution of the present disclosure, and these simple modifications all fall within the protection scope of the present disclosure.

[0160] It should also be noted that the various specific technical features described in the above specific embodiments can be combined in any suitable manner without contradiction. In order to avoid unnecessary repetition, the present disclosure will not further describe various possible combinations.

[0161] In addition, various embodiments of the present disclosure may be arbitrarily combined, and as long as they do not violate the concept of the present disclosure, they should also be regarded as the contents disclosed by the present disclosure.

Claims

1. A data processing method, characterized in that: include: Acquire multiple outbound call numbers and outbound call information of each of the outbound call numbers, the outbound call numbers include a first outbound call number and a second outbound call number, and the outbound call information of the first outbound call number meets a preset rule; Deleting the first outbound calling number; Determine, according to the outbound call information of the second outbound call number, a complaint probability of the second outbound call number, wherein the complaint probability is used to characterize the probability of receiving a complaint after an outbound call is made through the second outbound call number; According to the complaint probability of the second outbound calling number, the second outbound calling number is screened to obtain a third outbound calling number.

2. The data processing method according to claim 1, characterized in that: The determining, according to the outbound call information of the second outbound call number, the complaint probability of the second outbound call number includes: Determining an outbound calling feature of the second outbound calling number according to the outbound calling information of the second outbound calling number; The outbound call features of the second outbound call number are input into a pre-trained model, and the complaint probability of the second outbound call number is output.

3. The data processing method according to claim 1, characterized in that: The method further comprises: Obtain a training data set, the training data set including a plurality of training samples, each training sample including a sample outbound call feature and a complaint label of the sample outbound call feature, the sample outbound call feature being a feature extracted from outbound call information of a sample outbound call number, and the complaint label being used to characterize an outbound call complaint probability of the sample outbound call number; The model to be trained is trained according to the training data set to obtain the pre-trained model.

4. The data processing method according to claim 3, characterized in that: The outbound call feature includes at least one of the following features: Outbound call duration; Outbound call interval duration, which is used to represent the interval duration between two adjacent outbound calls; The total number of outbound calls and the number of outbound calls answered; Outbound calls to user reviews; User evaluation of outbound calls and the number of historical outbound complaints; The preset outbound call content is used to represent the negative evaluation of the outbound call receiving user.

5. The data processing method according to claim 1, characterized in that: The data processing method further includes: In response to the complaint probability of the second outbound number being greater than a first preset probability, configuring a first preset identifier for the second outbound number, wherein the first preset identifier is used to indicate that the user type to which the outbound number belongs is a first type of user; and / or, In response to the complaint probability of the second outbound number being less than a second preset probability, a second preset identifier is configured for the second outbound number, the second preset identifier being used to characterize that the user to which the outbound number belongs is a second type of user, and the second complaint probability is less than the first complaint probability.

6. The data processing method according to claim 1, characterized in that: The outbound call information includes at least one of the following: call information, number attribute information, and event mark information.

7. A data processing device, characterized in that: include: An acquisition unit, configured to acquire multiple outbound call numbers and outbound call information of each of the outbound call numbers, wherein the outbound call numbers include a first outbound call number and a second outbound call number, and the outbound call information of the first outbound call number satisfies a preset rule; a determination unit, configured to delete the first outbound call number, and determine a complaint probability of the second outbound call number according to the outbound call information of the second outbound call number, wherein the complaint probability is used to characterize a probability of receiving a complaint after an outbound call is made through the second outbound call number; The screening unit is used to screen the second outbound number according to the complaint probability of the second outbound number to obtain a third outbound number.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the data processing method according to any one of claims 1 to 6 is implemented.

9. An electronic device, characterized in that: include: a memory having a computer program stored thereon; A processor, configured to execute the computer program in the memory to implement the data processing method according to any one of claims 1 to 6.

Images