Queuing and calling methods and systems

Abstract

The application provides a queuing number calling method and system, and relates to the technical field of queuing number calling. The interval time length of the current moment from the starting moment of the business handling time period and the pre-recorded number of appointments of the business handling time period are input into a first number prediction model to predict a first total number of appointments of the business handling time period; the actual number of appointments in the business handling time period in the latest N days and the number change characteristics are input into a second number prediction model to predict a second total number of appointments of the business handling time period; the first total number of appointments and the second total number of appointments of the business handling time period are used to predict a final total number of appointments of the business handling time period; and the final total number of appointments and the proportion of each business type of business are input into a first waiting time length prediction model to predict the business handling waiting time length that a user needs to wait after handling on-site check-in in each check-in time slice of the check-in time period corresponding to the business handling time period.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of queuing for a number, in particular to a queuing for a number method and system. BACKGROUND

[0002] In life, in many scenarios, it is necessary to queue for a number to handle a business. For example, going to a bank to handle a business, going to a government office to do something or going to a large hospital to see a doctor, etc. all need to queue for a number.

[0003] At present, a user can make an appointment for a business handling period in advance, then arrive at a business handling site to check in and queue for a number before the end of the appointment business handling period, and then rest and wait for a number at a candidate position, which can make people get rid of the tiredness of standing in line and make the business handling more reasonable and orderly.

[0004] However, in some cases, when a user arrives at a business handling site, due to a large number of people queuing for a number, the user needs to wait for a long time, and even gives up handling the business, which results in a low user experience. SUMMARY

[0005] The present application provides a queuing for a number method and system, which is used to solve the problem that in some cases, when a user arrives at a business handling site, due to a large number of people queuing for a number, the user needs to wait for a long time, and even gives up handling the business, which results in a low user experience.

[0006] In a first aspect, the present application provides a queuing for a number method applied to a server, which comprises the following steps:

[0007] receiving queuing appointment information sent by a user terminal based on a target application program, wherein the queuing appointment information comprises an appointment business handling time period, a business handling type and user identity information;

[0008] in a case where the user identity information belongs to a preset senior user identity information set, inputting an interval time length of a current time from a start time of the business handling time period and a pre-recorded number of people who have made an appointment for the business handling time period into a pre-trained first number of people prediction model to predict a first total number of people who have made an appointment for the business handling time period, wherein the first number of people prediction model is obtained by inputting a plurality of first training samples into a first neural network for training, and each first training sample comprises a historical interval time length of a current time from a start time of a historical business handling time period, a pre-recorded number of people who have made an appointment for the historical business handling time period in the past, and a historical actual total number of people who have made an appointment for the historical business handling time period;

[0009] obtaining an actual number of people who have made an appointment for the business handling time period in the last N days, and extracting a number of people change feature of the actual number of people who have made an appointment for the business handling time period in the last N days;

[0010] inputting the actual number of appointments in the business handling time period in the recent N days and the number change feature into the pre-trained second number prediction model to predict a second number of appointments in the business handling time period, wherein the second number prediction model is trained by inputting a plurality of second training samples into a long short-term memory network, and each second training sample includes an actual number of appointments in the business handling time period in the recent N days in history, a corresponding historical number change feature, and a historical actual total number of appointments of the corresponding historical business handling time period;

[0011] predicting a final total number of appointments in the business handling time period according to the first total number of appointments and the second total number of appointments in the business handling time period;

[0012] determining a proportion of handling each business type in the business handling time period according to the business handling time period;

[0013] inputting the final total number of appointments and the proportion of handling each business type into the pre-trained first waiting time prediction model to predict a business handling waiting time that a user needs to wait after handling on-site check-in in each check-in time slice in a check-in time period corresponding to the business handling time period, wherein the first waiting time prediction model is trained by inputting a plurality of third training samples into a second neural network, and each third training sample includes a historical total number of appointments, a proportion of handling each business type in history, and a historical actual business handling waiting time that a user needs to wait after handling on-site check-in in each historical check-in time slice in a historical check-in time period corresponding to a historical business handling time period;

[0014] recommending a target check-in time slice associated with a minimum business handling waiting time to the user terminal;

[0015] after receiving a check-in instruction of the user terminal based on the user identity information sent by the check-in machine of the business handling site in the target check-in time slice, generating a business queue number associated with the user identity information according to the number of people who have checked in but have not handled the business, and queuing and calling the user identity information according to the business queue number.

[0016] In some embodiments, after generating the business queue number associated with the user identity information according to the number of people who have checked in but have not handled the business, the method provided by the present application further includes:

[0017] inputting the number of people who have signed in but have not handled the business and the business type of each person corresponding to the business to be handled into the pre-trained second waiting time length prediction model to predict the waiting time length that the user corresponding to the user identity information still needs to wait, wherein the second waiting time length prediction model is trained by inputting a plurality of fourth training samples into a third neural network, and each fourth training sample includes a historical number of people who have signed in but have not handled the business, a historical business type corresponding to each person to be handled, and a historical actual waiting time length of a user corresponding to the historical user identity information;

[0018] sending the waiting time length that the user corresponding to the user identity information still needs to wait to the user terminal;

[0019] in response to a temporary leaving instruction input by the user from the user terminal, determining the remaining waiting time length of the user in real time, and receiving user location information sent by the user terminal in real time, wherein the temporary leaving instruction carries a leaving traffic mode;

[0020] determining a return time length required for the user to return to the business handling site in real time according to the user location information, the location information of the business handling site, and the traffic mode;

[0021] in the case where the return time length is equal to the remaining waiting time length of the user, sending prompt information representing returning to the business handling site as soon as possible to the user terminal.

[0022] In some embodiments, after queuing the user identity information according to the business number, the method provided by the application further comprises:

[0023] recording the time when the user identity information is called, and determining an actual business handling waiting time length from the time when the user identity information is called to the time when the sign-in instruction is received;

[0024] in the case where the actual business handling waiting time length and the predicted business handling waiting time length have a time length difference greater than a set time length difference threshold, updating the network parameters of the first number prediction model, the network parameters of the second number prediction model, and the network parameters of the first waiting time length prediction model according to the time length difference.

[0025] In some embodiments, according to the first total number of reservations and the second total number of reservations of the business handling time period, the final total number of reservations of the business handling time period is predicted, comprising:

[0026] input the first total number of appointments and the second total number of appointments of the service handling time period into a third number prediction model pre-trained, to predict the final total number of appointments of the service handling time period, wherein the third number prediction model is obtained by inputting a plurality of fifth training samples into a fourth neural network, and each fifth training sample comprises a historical first total number of appointments, a historical second total number of appointments, and a corresponding historical final total number of appointments of a historical service handling time period.

[0027] In some embodiments, predicting the final total number of appointments of the service handling time period according to the first total number of appointments and the second total number of appointments of the service handling time period comprises:

[0028] predicting the final total number of appointments of the service handling time period according to the formula N=k1N1+k2N2, wherein N is the final total number of appointments of the service handling time period, N1 is the first total number of appointments, N2 is the second total number of appointments, k1 is a first weighting coefficient, k2 is a second weighting coefficient, and 0<k1<1, 0<k2<1, and k1+k2=1.

[0029] In a second aspect, the present application provides a queuing number calling system configured in a server, which comprises:

[0030] an information receiving unit configured to receive queuing appointment information sent by a user terminal based on a target application program, wherein the queuing appointment information comprises an appointment service handling time period, a service handling type, and user identity information;

[0031] a first number prediction unit configured to, in a case where the user identity information belongs to a preset senior user identity information set, input an interval time length from a current time to a start time of the service handling time period and a pre-recorded number of appointments of the service handling time period into a first number prediction model pre-trained, to predict a first total number of appointments of the service handling time period, wherein the first number prediction model is obtained by inputting a plurality of first training samples into a first neural network, and each first training sample comprises a historical interval time length from a current time to a start time of a historical service handling time period, a pre-recorded number of appointments of the historical service handling time period, and a corresponding historical actual total number of appointments of the historical service handling time period;

[0032] a feature extraction unit configured to acquire actual numbers of appointments in the service handling time period in the last N days, and extract number change features of the actual numbers of appointments in the service handling time period in the last N days;

[0033] a second number prediction unit configured to input the actual number of appointments in the service handling time period in the last N days and the number change feature into a pre-trained second number prediction model to predict a second number of appointments in the service handling time period, wherein the second number prediction model is trained by inputting a plurality of second training samples into a long short-term memory network, and each second training sample includes an actual number of appointments in the service handling time period in the last N days in history, a corresponding historical number change feature, and a historical actual total number of appointments in a corresponding historical service handling time period;

[0034] a third number prediction unit configured to predict a final total number of appointments in the service handling time period according to the first total number of appointments and the second total number of appointments in the service handling time period;

[0035] a service proportion determination unit configured to determine a proportion of handling each service type in the service handling time period according to the service handling time period;

[0036] a first waiting time prediction unit configured to input the final total number of appointments and the proportion of handling each service type into a pre-trained first waiting time prediction model to predict a service handling waiting time that a user needs to wait after handling on-site check-in in a corresponding check-in time period of the service handling time period, wherein the first waiting time prediction model is trained by inputting a plurality of third training samples into a second neural network, and each third training sample includes a historical total number of appointments, a proportion of handling each service type in history, and a historical actual service handling waiting time that a user needs to wait after handling on-site check-in in each historical check-in time slice in a corresponding historical check-in time period of a historical service handling time period;

[0037] a time slice recommendation unit configured to recommend a target check-in time slice associated with a minimum service handling waiting time to a user terminal;

[0038] a queuing and calling unit configured to, after receiving a check-in instruction of a user terminal based on user identity information sent by a check-in machine of a service handling site in the target check-in time slice, generate a service number associated with the user identity information according to the number of people who have checked in but have not handled services, and call the user identity information in a queue according to the service number.

[0039] In some embodiments, the system provided by the present application further comprises:

[0040] a second waiting time length prediction unit, configured to input the number of people who have signed in but have not handled the business and the business type corresponding to each person into a pre-trained second waiting time length prediction model to predict the waiting time length that the user corresponding to the user identity information still needs to wait;

[0041] an information sending unit, configured to send the waiting time length that the user corresponding to the user identity information still needs to wait to the user terminal;

[0042] a remaining waiting time length determination unit, configured to determine the remaining waiting time length that the user still needs to wait in real time in response to a temporary leaving instruction input by the user from the user terminal;

[0043] an information receiving unit, further configured to receive the user location information sent by the user terminal in real time, wherein the temporary leaving instruction carries a leaving traffic mode;

[0044] a return time length determination unit, configured to determine the return time length that the user needs to return to the business handling site in real time according to the user location information, the location information of the business handling site and the traffic mode;

[0045] a return prompting unit, configured to send prompt information representing returning to the business handling site as soon as possible to the user terminal in a case where the return time length is equal to the remaining waiting time length that the user still needs to wait.

[0046] In some embodiments, the system provided by the present application further comprises:

[0047] an actual waiting time length determination unit, configured to record the time when the user identity information is called and determine the actual business handling waiting time length from the time when the user identity information is called to the time when the sign-in instruction is received;

[0048] a parameter updating unit, configured to update the network parameters configured by the first number prediction model, the network parameters configured by the second number prediction model and the network parameters configured by the first waiting time length prediction model according to the time length difference in a case where the time length difference between the actual business handling waiting time length and the predicted business handling waiting time length that needs to be waited is greater than a set time length difference threshold.

[0049] In some embodiments, the third number of people prediction unit is specifically configured to input the first total number of appointments and the second total number of appointments of the service handling time period into a pre-trained third number of people prediction model to predict the final total number of appointments of the service handling time period, wherein the third number of people prediction model is trained by inputting a plurality of fifth training samples into a fourth neural network, and each fifth training sample includes a historical first total number of appointments, a historical second total number of appointments, and a historical final total number of appointments of a corresponding historical service handling time period.

[0050] In some embodiments, the third number of people prediction unit is specifically configured to predict the final total number of appointments of the service handling time period according to the formula N=k1N1+k2N2, wherein N is the final total number of appointments of the service handling time period, N1 is the first total number of appointments, N2 is the second total number of appointments, k1 is a first weighting coefficient, k2 is a second weighting coefficient, and 0<k1<1, 0<k2<1, and k1+k2=1.

[0051] The application provides a queuing number calling method and system. In the case that the user identity information belongs to a preset high-level user identity information set, the interval time length from the current time to the start time of the service handling time period and the pre-recorded number of appointments of the service handling time period are input into a pre-trained first number of people prediction model to predict the first total number of appointments of the service handling time period. Since the first training samples for training the first number of people prediction model include a historical interval time length from the current time to the start time of the historical service handling time period, a historical pre-recorded number of appointments of the historical service handling time period, and a historical actual total number of appointments of the corresponding historical service handling time period, the accuracy of the predicted first total number of appointments of the service handling time period is high.

[0052] The actual number of appointments in the service handling time period in the latest N days is obtained, and the number of people change characteristics of the actual number of appointments in the service handling time period in the latest N days is extracted. It can be understood that the number of people change characteristics is used to represent the evolution law of the actual number of appointments in the service handling time period in the latest N days.

[0053] The actual number of appointments in the service handling time period in the latest N days and the number of people change characteristics are input into a pre-trained second number of people prediction model to predict the second number of appointments of the service handling time period. Since the plurality of second training samples for training the second number of people prediction model include the actual number of appointments in the service handling time period in the latest N days, the corresponding historical number of people change characteristics, and the historical actual total number of appointments of the corresponding historical service handling time period, the accuracy of the obtained second number of appointments of the service handling time period is high.

[0054] Since the accuracy of the obtained first number of appointments for the service handling time period is high, and the accuracy of the obtained second number of appointments for the service handling time period is also high. In this way, according to the first total number of appointments and the second total number of appointments for the service handling time period, the accuracy of predicting the final total number of appointments for the service handling time period is higher.

[0055] According to the service handling time period, the proportion of handling each service type of business in the service handling time period is determined. Understandably, according to the summarized historical law, the proportion of each service type of business corresponding to different service handling time periods is different. For example, for bank business, in the morning 9:00-10:00, the proportion of handling pension collection, regular deposit maturity transfer and other businesses by the elderly is high, and in the noon 13:00-14:00, the proportion of handling transfer business, credit card repayment business and other businesses by the working class using rest time is high. For the housing management bureau business, in the morning 9:00-10:00, the proportion of handling property transfer, mortgage registration and other businesses is high, and in the noon 13:00-14:00, the proportion of handling tax deduction consultation, tax payment and other businesses by the working class using lunch break is high. Therefore, according to the service handling time period, the proportion of handling each service type of business in the service handling time period can be determined.

[0056] The final total number of appointments and the proportion of each service type of business are input into the pre-trained first waiting time length prediction model to predict the business handling waiting time length that the user needs to wait after handling on-site check-in in each check-in time slice in the check-in time period corresponding to the service handling time period. Since the first waiting time length prediction model is trained by the third training sample including the historical total number of appointments, the proportion of each service type of business in the history and the corresponding historical actual business handling waiting time length that the user needs to wait after handling on-site check-in in each historical check-in time slice in the historical check-in time period corresponding to the historical service handling time period. Therefore, the accuracy of the obtained business handling waiting time length that the user needs to wait after handling on-site check-in in each check-in time slice is high.

[0057] The target check-in time slice associated with the minimum business handling waiting time length is recommended to the user terminal. The user can browse the target check-in time slice associated with the minimum business handling waiting time length of himself, and reasonably plan the time to arrive at the business handling site for check-in in the target check-in time slice, so as to make the personal business handling time length shortest, reduce the proportion of giving up business handling, and improve the user's experience. BRIEF DESCRIPTION OF DRAWINGS

[0058] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0059] Figure 1 This is one of the flowcharts for the queuing and calling method provided in the embodiments of this application;

[0060] Figure 2 A functional block diagram of the queuing and calling system provided in the embodiments of this application. Detailed Implementation

[0061] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments made by those skilled in the art under the guidance of these embodiments are within the scope of protection of this application.

[0062] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a particular order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0063] Please see Figure 1 This application provides a queuing and calling method, which is applied to a server.

[0064] like Figure 1 As shown, the method provided in this application embodiment includes:

[0065] S101: Receive queuing reservation information sent by the user terminal based on the target application.

[0066] The appointment information includes the appointment time slot, the type of service, and the user's identity information.

[0067] For example, the service handling time period (e.g., 13:00-14:00) of the bank service, the service handling type (e.g., credit card repayment service), and the user identity information (e.g., name and ID number) sent by the bank APP

[0068] S102: In a case where the user identity information belongs to the preset high-level user identity information set, inputting the interval time length of the current time from the start time of the service handling time period and the pre-recorded number of people who have made appointments for the service handling time period into the pre-trained first number of people prediction model to predict the first total number of people who have made appointments for the service handling time period.

[0069] Since the first number of people prediction model is trained by inputting a plurality of first training samples into the first neural network, each first training sample includes a historical interval time length of a historical current time from a start time of a historical service handling time period, a historical number of people who have made appointments for the historical service handling time period, and a historical actual total number of people who have made appointments for the corresponding historical service handling time period, and thus the accuracy of the first total number of people who have made appointments for the service handling time period predicted in this way is high.

[0070] For example, in a case where the user identity information belongs to the preset high-level user identity information set (e.g., the VIP user identity information set), inputting the interval time length (e.g., 3 hours) of the current time (e.g., 10:00) from the start time (e.g., 13:00) of the service handling time period (e.g., 13:00-14:00) and the pre-recorded number of people who have made appointments for the service handling time period into the pre-trained first number of people prediction model to predict the first total number of people who have made appointments for the service handling time period (e.g., 13 people).

[0071] S103: Obtaining the actual number of people who have made appointments for the service handling time period in the last N days and extracting the number of people change features of the actual number of people who have made appointments for the service handling time period in the last N days. Understandably, the number of people change features are used to represent the evolution law of the actual number of people who have made appointments for the service handling time period in the last N days.

[0072] For example, the dependence relationship in time sequence of the actual number of people who have made appointments for the service handling time period in the last N days can be extracted according to the deep belief DBN network; and the number of people change features of the actual number of people who have made appointments for the service handling time period in the last N days can be extracted according to the dependence relationship in time sequence of the actual number of people who have made appointments for the service handling time period in the last N days.

[0073] S104: Inputting the actual number of people who have made appointments for the service handling time period in the last N days and the number of people change features into the pre-trained second number of people prediction model to predict the second number of people who have made appointments for the service handling time period.

[0074] The second predicted total number of appointments in the service handling time period is also high in accuracy, because the second number prediction model is trained by inputting a plurality of second training samples into a long short-term memory network, each of the second training samples including an actual number of appointments in the service handling time period in the last N days in history, corresponding historical number change features, and a historical actual total number of appointments in the corresponding historical service handling time period.

[0075] S105: predicting a final total number of appointments in the service handling time period according to the first total number of appointments and the second total number of appointments in the service handling time period.

[0076] Exemplarily, the specific implementation of S105 includes but is not limited to the following two kinds:

[0077] The first kind: inputting the first total number of appointments and the second total number of appointments in the service handling time period into a pre-trained third number prediction model to predict the final total number of appointments in the service handling time period, wherein the third number prediction model is trained by inputting a plurality of fifth training samples into a fourth neural network, wherein each of the fifth training samples includes a historical first total number of appointments, a historical second total number of appointments, and a historical final total number of appointments in the corresponding historical service handling time period.

[0078] The second kind: predicting the final total number of appointments in the service handling time period according to the formula N=k1N1+k2N2, wherein N is the final total number of appointments in the service handling time period, N1 is the first total number of appointments, N2 is the second total number of appointments, k1 is a first weighting coefficient, k2 is a second weighting coefficient, and 0<k1<1, 0<k2<1, k1+k2=1. For example, k1=0.5, k2=0.5.

[0079] S106: determining the proportion of each service type in the service handling time period according to the service handling time period.

[0080] It can be understood that according to the summarized historical law, the proportions of each service type in different service handling time periods are different. For example, for bank services, the proportion of old people handling pension collection, regular deposit maturity transfer and other services is high from 9:00 to 10:00 in the morning, and the proportion of working people handling transfer services, credit card repayment services and other services is high from 13:00 to 14:00 in the afternoon. For housing management bureau services, the proportion of property transfer, mortgage registration and other services is high from 9:00 to 10:00 in the morning, and the proportion of working people handling tax deduction consultation, tax payment and other services is high from 13:00 to 14:00 in the afternoon. Therefore, the proportion of each service type in the service handling time period can be determined according to the service handling time period.

[0081] For example, the business processing time period can be input into a pre-trained business proportion prediction model to predict the proportion of each business type processed during the business processing time period. The business proportion prediction model is trained by inputting multiple historical business processing time periods and the proportion of each business type corresponding to each business processing time period into the network to be trained.

[0082] S107: Input the total number of final appointments and the proportion of each business type into the pre-trained first waiting time prediction model to predict the waiting time for users to complete their business processing after signing in at each sign-in time slot within the corresponding sign-in time slot of the business processing time slot.

[0083] The first waiting time prediction model is obtained by inputting multiple third training samples into the second neural network. Each third training sample includes the total number of historical appointments, the proportion of each business type in history, and the actual historical waiting time for users to wait after signing in at the site in each historical sign-in time slot corresponding to the historical business handling time period.

[0084] For example, if the business processing time is 13:00-14:00, the corresponding check-in time is 12:00-14:00. A check-in time slice can be any 5-minute or 10-minute sub-time period within the 12:00-14:00 check-in time period. For example, check-in time slices could be 12:00-12:05, 12:40-12:45, and 13:20-13:25, etc. When the check-in time slice is 12:00-12:05, the predicted waiting time for business processing can be 1 hour; when the check-in time slice is 12:40-12:45, the predicted waiting time for business processing can be 25 minutes; and when the check-in time slice is 13:00-13:05, the predicted waiting time for business processing can be 40 minutes.

[0085] S108: Recommend the target check-in time slice associated with the minimum service processing wait time to the user terminal.

[0086] Based on the above S106, the user terminal can be recommended a check-in time slot of 12:40-12:45 with a predicted service processing waiting time of 25 minutes.

[0087] S109: After receiving the sign-in instruction based on user identity information from the sign-in machine at the business processing site during the target sign-in time slot, generate a business queue number associated with the user identity information based on the number of people who have signed in but have not yet processed their business, and then queue and call the users' identity information according to the business queue number.

[0088] In some embodiments, after S108, the method provided by the embodiments of the present application further comprises:

[0089] Step 1: inputting the number of people who have signed in but have not handled the business and the business type corresponding to each person into the pre-trained second waiting time prediction model to predict the waiting time that the user corresponding to the user identity information still needs to wait.

[0090] Understandably, since the second waiting time prediction model is trained by inputting a plurality of fourth training samples into the third neural network, each fourth training sample includes a historical number of people who have signed in but have not handled the business, a historical business type corresponding to each person, and a historical actual waiting time of a user corresponding to the historical user identity information. In this way, the accuracy of the predicted waiting time that the user corresponding to the user identity information still needs to wait can be high.

[0091] Step 2: sending the waiting time that the user corresponding to the user identity information still needs to wait to the user terminal.

[0092] Step 3: in response to a temporary leaving instruction input by the user from the user terminal, determining the remaining waiting time that the user still needs to wait in real time, and receiving user location information sent by the user terminal in real time, wherein the temporary leaving instruction carries a leaving traffic mode.

[0093] For example, when the user arrives at the business handling site and finds that the waiting time that still needs to be waited is long (such as 30 min), the user needs to temporarily leave the business handling site, and can input a temporary leaving instruction based on a target application program in the user terminal. The user location information can be recorded in real time by the GPS module of the user terminal when the user leaves the business handling site, and sent to the server.

[0094] Step 4: determining the return time required for the user to return to the business handling site in real time according to the user location information, the location information of the business handling site, and the traffic mode.

[0095] For example, when the user leaves the business handling site to go shopping or sightseeing by riding a shared bicycle, the return time of the user riding the shared bicycle from the user location information to the business handling site can be determined.

[0096] Step 5: in the case where the return time is equal to the remaining waiting time that the user still needs to wait, sending prompt information representing returning to the business handling site as soon as possible to the user terminal.

[0097] Understandably, as time goes by and the number of people who have signed in but have not handled business decreases, the remaining waiting time of the user becomes shorter and shorter, and when the return time is equal to the remaining waiting time of the user, the prompt information for indicating to return to the business handling site as soon as possible is sent to the user terminal. In this way, the user returns to the business handling site after perceiving the prompt information, and after returning to the business handling site, no waiting is needed, and business handling can be performed, meeting the diversified needs of the user.

[0098] In some embodiments, after S108, the method provided by the embodiment of the application further includes: recording the time when the user identity information is called, and determining the actual business handling waiting time from the time when the called number is called to the time when the sign-in instruction is received; in the case that the time difference between the actual business handling waiting time and the predicted business handling waiting time is greater than the set time difference threshold, the network parameters configured by the first number prediction model, the network parameters configured by the second number prediction model, and the network parameters configured by the first waiting time prediction model are updated according to the time difference. In this way, the accuracy of the predicted business handling waiting time of the user in the subsequent sign-in time period corresponding to the business handling time period can be high.

[0099] In summary, the queuing and calling method provided by the embodiment of the application inputs the interval time from the current time to the start time of the business handling time period and the pre-recorded number of people who have made appointments in the business handling time period to the pre-trained first number prediction model in the case that the user identity information belongs to the pre-set high-level user identity information set, and predicts the first total number of appointments in the business handling time period. Since the first training sample for training the first number prediction model includes the historical interval time from the current time to the start time of the historical business handling time period, the historical number of people who have made appointments in the historical business handling time period, and the historical actual total number of appointments in the corresponding historical business handling time period, the accuracy of the predicted first total number of appointments in the business handling time period is high.

[0100] The actual number of appointments in the business handling time period in the last N days is obtained, and the number change feature of the actual number of appointments in the business handling time period in the last N days is extracted. Understandably, the number change feature is used to represent the evolution law of the actual number of appointments in the business handling time period in the last N days.

[0101] The actual number of appointments in the recent N days in the service handling time period and the number change characteristics are input into the pre-trained second number prediction model to predict the second number of appointments in the service handling time period. Since the plurality of second training samples for training the second number prediction model include the actual number of appointments in the recent N days in the service handling time period in history, the corresponding historical number change characteristics, and the historical actual total number of appointments in the corresponding historical service handling time period. Therefore, the accuracy of the obtained second number of appointments in the service handling time period is high.

[0102] Since the accuracy of the obtained first number of appointments in the service handling time period is high, and the accuracy of the obtained second number of appointments in the service handling time period is also high. Therefore, according to the first total number of appointments and the second total number of appointments in the service handling time period, the accuracy of predicting the final total number of appointments in the service handling time period is higher.

[0103] According to the service handling time period, the proportion of handling each business type of business in the service handling time period is determined. It can be understood that according to the summarized historical law, the proportion of each business type of business corresponding to different service handling time periods is different. For example, for bank business, in the morning 9:00-10:00, the proportion of handling pension collection, regular deposit maturity transfer and other businesses by the elderly is high, and in the noon 13:00-14:00, the proportion of handling transfer business, credit card repayment business and other businesses by the working class during the break time is high. For the housing management bureau business, in the morning 9:00-10:00, the proportion of handling property transfer, mortgage registration and other businesses is high, and in the noon 13:00-14:00, the proportion of handling tax deduction consultation, tax payment and other businesses by the working class during the lunch break is high. Therefore, according to the service handling time period, the proportion of handling each business type of business in the service handling time period can be determined.

[0104] The final total number of appointments and the proportion of each business type of business are input into the pre-trained first waiting time prediction model to predict the business handling waiting time that the user needs to wait after handling the on-site check-in in each check-in time slice in the check-in time period corresponding to the service handling time period. Since the third training sample for training the first waiting time prediction model includes the historical total number of appointments, the proportion of each business type of business in history, and the historical actual business handling waiting time that the user needs to wait after handling the on-site check-in in each historical check-in time slice in the historical check-in time period corresponding to the historical service handling time period. Therefore, the accuracy of the obtained business handling waiting time that the user needs to wait after handling the on-site check-in in each check-in time slice is high.

[0105] The user terminal is recommended a target check-in time slice associated with a minimum service handling waiting time. The user can browse the target check-in time slice associated with the minimum service handling waiting time, and reasonably plan to arrive at the service handling site for check-in at the target check-in time slice, so that the personal service handling time is the shortest, the proportion of giving up service handling is reduced, and the user experience is improved.

[0106] Please refer to Figure 2 The embodiments of the present application provide a queuing number calling system configured in a server. It should be noted that the queuing number calling system provided by the embodiments of the present application has the same basic principles and technical effects as the above-mentioned embodiments. For brevity, the part of the embodiments of the present application not mentioned can refer to the corresponding content in the above-mentioned embodiments. The system provided by the embodiments of the present application includes an information receiving unit, a first number prediction unit, a feature extraction unit, a second number prediction unit, a third number prediction unit, a service proportion determination unit, a first waiting time prediction unit, a time slice recommendation unit, and a queuing number calling unit, wherein,

[0107] The information receiving unit is configured to receive queuing reservation information sent by a user terminal based on a target application program, wherein the queuing reservation information includes a reserved service handling time period, a service handling type, and user identity information.

[0108] The first number prediction unit is configured to, in a case where the user identity information belongs to a preset senior user identity information set, input an interval time length from a current time to a start time of the service handling time period and a pre-recorded number of reserved people of the service handling time period into a pre-trained first number prediction model to predict a first total number of reservations of the service handling time period, wherein the first number prediction model is obtained by inputting a plurality of first training samples into a first neural network for training, and each first training sample includes a historical interval time length from a current time to a start time of a historical service handling time period, a pre-recorded number of historical reservations of the historical service handling time period, and a historical total number of actual reservations of the corresponding historical service handling time period.

[0109] The feature extraction unit is configured to obtain actual reservation numbers in the service handling time period in the last N days, and extract number change features of the actual reservation numbers in the service handling time period in the last N days.

[0110] a second number prediction unit configured to input the actual number of appointments in the service handling time period in the last N days and the number change feature into a pre-trained second number prediction model to predict a second number of appointments in the service handling time period, wherein the second number prediction model is trained by inputting a plurality of second training samples into a long short-term memory network, and each second training sample includes an actual number of appointments in the service handling time period in the last N days, a corresponding historical number change feature, and a historical actual total number of appointments in a corresponding historical service handling time period;

[0111] a third number prediction unit configured to predict a final total number of appointments in the service handling time period according to the first total number of appointments and the second total number of appointments in the service handling time period;

[0112] a service proportion determination unit configured to determine a proportion of handling each service type in the service handling time period according to the service handling time period;

[0113] a first waiting time prediction unit configured to input the final total number of appointments and the proportion of handling each service type into a pre-trained first waiting time prediction model to predict a service handling waiting time that a user needs to wait after handling on-site check-in in each check-in time slice in a check-in time period corresponding to the service handling time period, wherein the first waiting time prediction model is trained by inputting a plurality of third training samples into a second neural network, and each third training sample includes a historical total number of appointments, a proportion of handling each service type in history, and a historical actual service handling waiting time that a user needs to wait after handling on-site check-in in each historical check-in time slice in a historical check-in time period corresponding to a historical service handling time period;

[0114] a time slice recommendation unit configured to recommend a target check-in time slice associated with a minimum service handling waiting time to a user terminal;

[0115] a queuing and calling unit configured to generate a service number associated with user identity information according to a number of people who have checked in but have not handled services after receiving a check-in instruction based on the user identity information sent by a check-in machine of the service handling site in the target check-in time slice, and call the user identity information in a queue according to the service number.

[0116] In some embodiments, the system provided by the application further comprises a second waiting time prediction unit configured to input the number of people who have signed in but have not handled the business and the corresponding business type of each person into a pre-trained second waiting time prediction model to predict the waiting time that the user corresponding to the user identity information still needs to wait, wherein the second waiting time prediction model is trained by inputting a plurality of fourth training samples into a third neural network, and each fourth training sample comprises a historical number of people who have signed in but have not handled the business, a corresponding historical business type of each person, and a historical actual waiting time of a user corresponding to the historical user identity information.

[0117] an information sending unit configured to send the waiting time that the user corresponding to the user identity information still needs to wait to the user terminal;

[0118] a remaining waiting time determination unit configured to determine the remaining waiting time that the user still needs to wait in real time in response to a temporary leaving instruction input by the user from the user terminal;

[0119] the information receiving unit is further configured to receive user location information sent by the user terminal in real time, wherein the temporary leaving instruction carries a leaving traffic mode;

[0120] a return time determination unit configured to determine a return time that the user needs to return to the business handling site in real time according to the user location information, location information of the business handling site, and the traffic mode;

[0121] a return prompting unit configured to send prompt information representing returning to the business handling site as soon as possible to the user terminal in a case where the return time is equal to the remaining waiting time that the user still needs to wait.

[0122] In some embodiments, the system provided by the application further comprises an actual waiting time determination unit configured to record a time when the user identity information is called and determine an actual business handling waiting time from the time when the user identity information is called to the time when the sign-in instruction is received.

[0123] a parameter updating unit configured to update network parameters configured by the first number prediction model, network parameters configured by the second number prediction model, and network parameters configured by the first waiting time prediction model according to the time difference in a case where the time difference between the actual business handling waiting time and the predicted business handling waiting time is greater than a set time difference threshold.

[0124] In some embodiments, the third number prediction unit is specifically configured to input the first total number of appointments and the second total number of appointments of the service handling time period into a pre-trained third number prediction model to predict the final total number of appointments of the service handling time period, wherein the third number prediction model is trained by inputting a plurality of fifth training samples into a fourth neural network, and each fifth training sample comprises a historical first total number of appointments, a historical second total number of appointments, and a historical final total number of appointments of a corresponding historical service handling time period.

[0125] In some embodiments, the third number prediction unit is specifically configured to predict the final total number of appointments of the service handling time period according to the formula N = k1N1 + k2N2, wherein N is the final total number of appointments of the service handling time period, N1 is the first total number of appointments, N2 is the second total number of appointments, k1 is a first weighting coefficient, k2 is a second weighting coefficient, and 0 < k1 < 1, 0 < k2 < 1, and k1 + k2 = 1.

[0126] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A method of queuing for a number, characterized by, Applied to a server, the method comprises: Receiving queuing reservation information sent by a user terminal based on a target application, wherein the queuing reservation information comprises a reserved service handling time period, a service handling type, and user identity information; In a case where the user identity information belongs to a preset senior user identity information set, inputting an interval time length from a current time to a start time of the service handling time period and a pre-recorded number of people who have reserved the service handling time period into a pre-trained first number of people prediction model to predict a first total number of people who have reserved the service handling time period, wherein the first number of people prediction model is obtained by inputting a plurality of first training samples into a first neural network, each of the first training samples comprising a historical interval time length from a historical current time to a start time of a historical service handling time period, a historical number of people who have pre-recorded the historical service handling time period, and a historical actual total number of people who have reserved the corresponding historical service handling time period; Obtaining an actual number of people who have reserved the service handling time period in the last N days and extracting a number of people change feature of the actual number of people who have reserved the service handling time period in the last N days; Inputting the actual number of people who have reserved the service handling time period in the last N days and the number of people change feature into a pre-trained second number of people prediction model to predict a second number of people who have reserved the service handling time period, wherein the second number of people prediction model is obtained by inputting a plurality of second training samples into a long short-term memory network, each of the second training samples comprising an actual number of people who have reserved the service handling time period in the last N days, a corresponding historical number of people change feature, and a historical actual total number of people who have reserved the corresponding historical service handling time period; Predicting a final total number of people who have reserved the service handling time period according to the first total number of people who have reserved the service handling time period and the second total number of people who have reserved the service handling time period; Determining a proportion of each service type in the service handling time period according to the service handling time period; Inputting the final total number of people who have reserved the service handling time period and the proportion of each service type into a pre-trained first waiting time length prediction model to predict a service handling waiting time length that a user needs to wait after handling on-site check-in in each check-in time slice in a check-in time period corresponding to the service handling time period, wherein the first waiting time length prediction model is obtained by inputting a plurality of third training samples into a second neural network, each of the third training samples comprising a historical total number of people who have reserved, a proportion of each service type in the past, and a historical actual service handling waiting time length that a user needs to wait after handling on-site check-in in each historical check-in time slice in a historical check-in time period corresponding to a historical service handling time period; Recommending a target check-in time slice associated with a minimum service handling waiting time length to the user terminal; After receiving the check-in instruction based on the user identity information sent by the check-in machine of the service site at the target check-in time slice, a service queue number associated with the user identity information is generated according to the number of people who have checked in but not handled the service, and the user identity information is queued according to the service queue number.

2. The method of claim 1, wherein, After the service queue number associated with the user identity information is generated according to the number of people who have checked in but not handled the service, the method further comprises: inputting the number of people who have checked in but not handled the service and the service type to be handled by each person into a pre-trained second waiting time length prediction model to predict the waiting time length that the user corresponding to the user identity information still needs to wait, wherein the second waiting time length prediction model is trained by inputting a plurality of fourth training samples into a third neural network, and each fourth training sample includes a historical number of people who have checked in but not handled the service, a historical service type to be handled by each person, and a historical actual waiting time length of a user corresponding to a historical user identity information. sending the waiting time length that the user corresponding to the user identity information still needs to wait to the user terminal; in response to a temporary leave instruction input by the user from the user terminal, determining the remaining waiting time length that the user still needs to wait in real time, and receiving user location information sent by the user terminal in real time, wherein the temporary leave instruction carries a transportation mode of leaving; determining a return time length required for the user to return to the service site according to the user location information, location information of the service site, and the transportation mode; in the case where the return time length is equal to the remaining waiting time length of the user, sending prompt information to the user terminal to indicate returning to the service site as soon as possible.

3. The method of claim 1, wherein, After the user identity information is queued according to the service queue number, the method further comprises: recording the time when the user identity information is called, and determining an actual service handling waiting time length from the time when the user identity information is called to the time when the check-in instruction is received; in the case where the actual service handling waiting time length and the predicted service handling waiting time length have a time length difference greater than a set time length difference threshold, updating the network parameters of the first number prediction model, the network parameters of the second number prediction model, and the network parameters of the first waiting time length prediction model according to the time length difference.

4. The method of claim 1, wherein, The prediction of the final total number of reservations for the service handling time period according to the first total number of reservations and the second total number of reservations for the service handling time period comprises: inputting the first total number of reservations and the second total number of reservations for the service handling time period into a pre-trained third number prediction model to predict the final total number of reservations for the service handling time period, wherein the third number prediction model is trained by inputting a plurality of fifth training samples into a fourth neural network, and each fifth training sample includes a historical first total number of reservations, a historical second total number of reservations, and a historical final total number of reservations for a corresponding historical service handling time period.

5. The method of claim 1, wherein, The final total number of appointments of the service handling time period is predicted according to the first total number of appointments and the second total number of appointments of the service handling time period, and the method comprises the steps of: The final total number of appointments of the service handling time period is predicted according to the formula N=k1N1+k2N2, wherein N is the final total number of appointments of the service handling time period, N1 is the first total number of appointments, N2 is the second total number of appointments, k1 is the first weighting coefficient, k2 is the second weighting coefficient, and 0k1<1, 0<k2<1, and k1+k2=1.

6. A queuing call system characterized by, The system comprises a server configured to: An information receiving unit is configured to receive queuing appointment information sent by a user terminal based on a target application program, wherein the queuing appointment information comprises an appointment service handling time period, a service handling type, and user identity information; A first number prediction unit is configured to, in a case where the user identity information belongs to a preset senior user identity information set, input an interval time length from a current time to a start time of the service handling time period and a pre-recorded number of appointments of the service handling time period into a pre-trained first number prediction model to predict a first total number of appointments of the service handling time period, wherein the first number prediction model is obtained by inputting a plurality of first training samples into a first neural network, and each first training sample comprises a historical interval time length from a historical current time to a start time of a historical service handling time period, a pre-recorded number of historical appointments of the historical service handling time period, and a historical actual total number of appointments of a corresponding historical service handling time period; A feature extraction unit is configured to obtain actual numbers of appointments in the service handling time period in the last N days and extract number change features of the actual numbers of appointments in the service handling time period in the last N days; A second number prediction unit is configured to input the actual numbers of appointments in the service handling time period in the last N days and the number change features into a pre-trained second number prediction model to predict a second number of appointments of the service handling time period, wherein the second number prediction model is obtained by inputting a plurality of second training samples into a long short-term memory network, and each second training sample comprises actual numbers of appointments in the service handling time period in the last N days, corresponding historical number change features, and a historical actual total number of appointments of a corresponding historical service handling time period; A third number prediction unit is configured to predict a final total number of appointments of the service handling time period according to the first total number of appointments and the second total number of appointments of the service handling time period; A service proportion determination unit is configured to determine proportions of handling each service type in the service handling time period according to the service handling time period. The first waiting time length prediction unit is configured to input the final total number of appointments and the proportion of each service type into a pre-trained first waiting time length prediction model to predict a waiting time length that a user needs to wait after conducting on-site check-in in each check-in time slice in a check-in time period corresponding to the service conducting time period, wherein the first waiting time length prediction model is obtained by inputting a plurality of third training samples into a second neural network, and each third training sample includes a historical total number of appointments, a proportion of each service type, and a historical actual service conducting waiting time length that a user needs to wait after conducting on-site check-in in each historical check-in time slice in a historical check-in time period corresponding to a historical service conducting time period; The time slice recommendation unit is configured to recommend a target check-in time slice associated with a minimum service conducting waiting time length to the user terminal; The queuing and calling unit is configured to generate a service queue number associated with the user identity information according to the number of people who have checked in but have not conducted services, and call the user identity information according to the service queue number after receiving a check-in instruction based on the user identity information sent by a check-in machine of the service conducting site in the target check-in time slice.

7. The system of claim 6, wherein, The system further comprises: The second waiting time length prediction unit is configured to input the number of people who have checked in but have not conducted services and the service type of each person into a pre-trained second waiting time length prediction model to predict a waiting time length that a user corresponding to the user identity information still needs to wait, wherein the second waiting time length prediction model is obtained by inputting a plurality of fourth training samples into a third neural network, and each fourth training sample includes a historical number of people who have checked in but have not conducted services, a historical service type of each person, and a historical actual waiting time length of a user corresponding to a historical user identity information; The information sending unit is configured to send the waiting time length that the user corresponding to the user identity information still needs to wait to the user terminal; The remaining waiting time length determination unit is configured to determine a remaining waiting time length that the user still needs to wait in real time in response to a temporary leaving instruction input by the user from the user terminal; The information receiving unit is further configured to receive user location information sent by the user terminal in real time, wherein the temporary leaving instruction carries a leaving transportation mode; The return time length determination unit is configured to determine a return time length that the user needs to return to the service conducting site in real time according to the user location information, location information of the service conducting site, and the transportation mode; The return prompting unit is configured to send prompt information representing returning to the service conducting site as soon as possible to the user terminal in a case where the return time length is equal to the remaining waiting time length that the user still needs to wait.

8. The system of claim 6, wherein, The system further comprises: The actual waiting time length determination unit is configured to record a time when the user identity information is called, and determine an actual service conducting waiting time length from the time when the user identity information is called to a time when the check-in instruction is received. The parameter updating unit is configured to update the network parameters of the first people number prediction model, the network parameters of the second people number prediction model, and the network parameters of the first waiting time prediction model according to the time length difference when the time length difference between the actual service handling waiting time and the predicted service handling waiting time is greater than the set time length difference threshold.

9. The system of claim 6, wherein, The third people number prediction unit is specifically configured to input the first total number of appointments and the second total number of appointments of the service handling time period into a pre-trained third people number prediction model to predict a final total number of appointments of the service handling time period, wherein the third people number prediction model is trained by inputting a plurality of fifth training samples into a fourth neural network, and each of the fifth training samples includes a historical first total number of appointments, a historical second total number of appointments, and a historical final total number of appointments of a corresponding historical service handling time period.

10. The system of claim 6, wherein, The third people number prediction unit is specifically configured to predict the final total number of appointments of the service handling time period according to the formula N=k1N1+k2N2, wherein N is the final total number of appointments of the service handling time period, N1 is the first total number of appointments, N2 is the second total number of appointments, k1 is a first weighting coefficient, k2 is a second weighting coefficient, and 0

Images