Call center call waiting prediction system and call center call waiting prediction method
The call waiting prediction system addresses the challenge of predicting waiting times in call centers by using voice recognition and waiting time management algorithms to play tailored automated messages, enhancing user experience and operational efficiency.
Patent Information
- Application Number
- JP2023200494
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2023-11-28
- Publication Date
- 2025-06-09
AI Technical Summary
Existing call center systems cannot predict the waiting time for a call with an operator when a user is put on hold due to an automated response, making it difficult to play an appropriate automated voice to manage user expectations.
A call waiting prediction system that includes a voice recognition device to analyze call content and a waiting time management device to predict waiting time using current time zone, call content, pre-registered statistical information, and call model data, allowing for the playback of automated voices based on predicted wait times.
Enables accurate prediction of waiting time until a call with an operator can be initiated, allowing the system to play appropriate automated messages to users, improving user experience and operational efficiency in call centers.
Smart Images

Figure 2025086492000001_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a call waiting prediction system for a call center and a call waiting prediction method for a call center.
Background Art
[0002] Conventionally, in a call center, requests and inquiries from users who use products, services, etc. are received by an operator using a telephone. In a call center, when a large number of calls from users occur, a situation where calls cannot be answered occurs. Therefore, a technique for detecting an operator who has ended a call and starting a call from a waiting user, for example, Patent Document 1 has been proposed.
[0003] Patent Document 1 describes that "when an incoming call is included in the utterances constituting each call, each call is divided immediately before the incoming call, and when a disconnection call is included in the utterances constituting each call, each call is divided immediately after the disconnection call, and when there is a call in which the last utterance constituting the previous call is not a disconnection call and the first utterance is not an incoming call, the call and the previous call are combined."
Prior Art Documents
Patent Documents
[0004]
Patent Document 1
Summary of the Invention
Problems to be Solved by the Invention
[0005] As described above, conventionally, a technique (Patent Document 1) has been proposed for determining that an operator who receives a call in a call center will soon end the conversation and become available to receive the next call. By the way, when a user calls a crowded call center, the call center system first receives the call using an automated voice response and then either waits or plays a message prompting the user to hang up and call again. However, the technique described in Patent Document 1 cannot predict the time until the start of a call with an operator, i.e., the waiting time, when a user calls the call center and is waiting due to an automated response. Therefore, the technique described in Patent Document 1 cannot play an automated voice that appropriately prompts the user to either wait or hang up and call again according to the waiting time.
[0006] The present invention has been made to solve the above problems, and an object of the present invention is to predict the waiting time until the start of a call with an operator in a call center and play an automated voice according to the waiting time.
Means for Solving the Problems
[0007] To solve the above problems, a call waiting prediction system for a call center according to the present invention is a call waiting prediction system for a call center including a plurality of telephones and a switching center having a function of allocating calls to the plurality of telephones and a function of playing an automated voice, the system comprising: a voice recognition device for recognizing the content of a call in progress on each of the plurality of telephones; and a waiting time management device for predicting the waiting time until the start of the call when a call is received at the call center and the call is being waited for by an automated voice, based on the current time zone, the call content recognized by the voice recognition device, pre-registered call statistical information, and call model information. Note that the above call waiting prediction system for a call center is an aspect of the present invention, and a call waiting prediction method for a call center reflecting one aspect of the present invention is also configured in the same manner as the above call waiting prediction system for a call center.
Advantages of the Invention
[0008] According to the present invention configured as described above, it is possible to predict the waiting time until the start of a call with an operator of a call center and reproduce an automatic voice according to the waiting time. Problems, configurations, and effects other than those described above will be clarified by the description of the following embodiments.
Brief Description of the Drawings
[0009]
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Mode for Carrying Out the Invention
[0010] Hereinafter, embodiments for carrying out the present invention will be described with reference to the accompanying drawings. In this specification and the drawings, components having substantially the same function or configuration are denoted by the same reference numerals, and redundant description is omitted.
[0011] <One Embodiment> [Configuration of Call Waiting Prediction System for Call Center] First, the configuration of the call waiting prediction system of the call center according to an embodiment of the present invention will be described. FIG. 1 is a diagram showing a configuration example of the call waiting prediction system 10 of the call center according to the present embodiment. As shown in FIG. 1, the call waiting prediction system 10 includes a plurality of telephones (telephones 11a to 11c) arranged in the call center 1, a line switch 12, a waiting time management device 13, a voice recognition device 14, a conversation statistical database 15, and a conversation model database 16. In addition, there are operators 20a to 20c corresponding to each of the plurality of telephones in the call center 1. Further, as shown in FIG. 1, the call waiting prediction system 10 is connected to external telephones 3a to 3n via the telephone network 2. Users 4a to 4n, who are users of each of the telephones 3a to 3n, make calls to the call center 1 using the telephones 3a to 3n.
[0012] In the call center 1, the plurality of telephones are connected to the line switch 12, and the line switch 12 is connected to the telephone network 2. Further, the line switch 12, the waiting time management device 13, and the voice recognition device 14 are connected in this order. In addition, the waiting time management device 13 is also connected to each of the conversation statistical database 15 and the conversation model database 16.
[0013] The line switch 12 has a function of allocating calls to a plurality of telephones (telephones 3a to 3n) and a function of playing back an automatic voice. Further, the line switch 12 reads the call voice between the user on the call and the operator and outputs it to the waiting time management device 13.
[0014] When a user makes a call, in the call center 1, when there is no available operator, that is, when all operators are in a state of being on a call with a user, the line switch 12 plays back an automatic voice. Here, the line switch 12 selects and plays back the instructed automatic voice from a plurality of pre-recorded automatic voices according to the automatic voice playback instruction from the waiting time management device 13. The user can listen to the automatic voice and decide whether to wait until an operator becomes available or to hang up the phone once and call again.
[0015] The waiting time management device 13 is composed of, for example, a CPU (Central Processing Unit), a RAM (Random Access Memory), a ROM (Read Only Memory), etc., which are not shown in the figure. The waiting time management device 13 acquires the call voice between the user on the call and the operator from the circuit switch 12 and outputs it to the voice recognition device 14. Further, when a call is made to the call center 1 and the call is on hold, the waiting time management device 13 predicts the waiting time until the start of the call based on the current time zone, the call content recognized by the voice recognition device 14, the conversation statistical information registered in advance (conversation trend data D1 described later), and the conversation model information (conversation position determination data D2 and conversation content determination data D3 described later).
[0016] Specifically, the waiting time management device 13 calculates the average waiting time of calls in the current time zone as the first waiting time based on the current time zone and the conversation statistical information. The calculation of the first waiting time will be described in detail with reference to FIG. 2, which will be described later. Further, the waiting time management device 13 calculates the average time until the end of each call of a plurality of telephones as the second waiting time based on the call content recognized by the voice recognition device 14 and the conversation model information. Further, the waiting time management device 13 predicts the total time of the calculated first waiting time and the second waiting time as the waiting time. The calculation of the second waiting time will be described in detail with reference to FIGS. 3 and 4, which will be described later.
[0017] The waiting time management device 13 determines whether the predicted call waiting time is within a predetermined time range. When the waiting time management device 13 determines that the predicted call waiting time is within the predetermined time range, it outputs to the circuit switch 12 a reproduction instruction for automatic voice that reproduces a message prompting to wait for the call. Also, when the waiting time management device 13 determines that the predicted call waiting time is not within the predetermined time range, it outputs to the circuit switch 12 a reproduction instruction for automatic voice that reproduces a message prompting that it is currently congested and to hang up and call again. Note that the message prompting to wait for the call is, for example, a message such as "Please wait as it is for a while". The message prompting that it is currently congested and to hang up and call again is, for example, a message such as "It is currently congested. Please call again". Note that the automatic voice message is not limited to the above-described messages, and any message may be used as long as it expresses the intention to prompt.
[0018] The voice recognition device 14 is composed of, for example, a CPU, RAM, ROM, etc. (not shown), inputs the call voice from the waiting time management device 13, and uses artificial intelligence (AI) to recognize the call content of each of the plurality of telephones from the content of the call voice. Here, the call content recognized by the voice recognition device 14 includes the patterned conversation content and phrase pattern during the call (see FIGS. 3 and 4 described later). Also, the voice recognition device 14 outputs the recognized call content of each of the plurality of telephones to the waiting time management device 13.
[0019] The conversation statistical database 15 stores the conversation trend data D1 shown in FIG. 2 described later. The conversation trend data D1 includes statistical information on past conversation content in units such as daily, by time zone, by day of the week, etc. This statistical information includes the patterned conversation content, the average call time for each call corresponding to each conversation content, and the ratio of each conversation content to all the calls in that time zone in time zones such as day, time zone, and day of the week. FIG. 2 is a diagram showing an example of the conversation trend data D1 in the call waiting prediction system 10 of the call center according to the present embodiment.
[0020] As shown in FIG. 2, the conversation tendency data D1 includes a time zone column D11, a conversation content column D12, a ratio column D13, and an average time column D14. In the time zone column D11, information representing the time zone, such as "Monday daytime", "Monday nighttime", etc. is stored. In the conversation content column D12, information representing the patterned conversation content, such as "estimate request", "claim", etc. is stored. In the ratio column D13, information representing the ratio of the number of calls including the conversation content to the number of all calls in that time zone, such as "50%", "30%", etc. is stored. In the average time column D14, information representing the average time of the calls corresponding to the conversation content, such as "5 minutes", "10 minutes", etc. is stored.
[0021] When a call is received, for example, on Monday daytime, the waiting time management device 13 acquires "estimate request", "repair request", and "claim" which are the conversation contents corresponding to the time zone of "Monday daytime" in the conversation tendency data D1. Also, the waiting time management device 13 acquires the ratios ("50%", "30%", and "20%") and the average times ("5 minutes", "10 minutes", and "15 minutes") corresponding to "estimate request", "repair request", and "claim" respectively. Then, the waiting time management device 13 calculates the waiting time T1 for Monday daytime according to the following formula (1).
[0022] [Equation 1] T1 = a1×(P1 / 100) + a2×(P2 / 100) + a3×(P3 / 100) …(1)
[0023] In the above formula (1), P1, P2, and P3 are the ratios "50%", "30%", and "20%" corresponding to "estimate request", "repair request", and "claim" respectively. a1, a2, and a3 are the average times "5 minutes", "10 minutes", and "15 minutes" corresponding to "estimate request", "repair request", and "claim" respectively.
[0024] In the above example, the calculation result according to Equation (1) is that the waiting time T1 = 5 (minutes) × 0.5 + 10 (minutes) × 0.3 + 15 (minutes) × 0.2 = 13 (minutes). Further, the waiting time management device 13 calculates the number of users waiting for a call × the waiting time T1 as the total waiting time. Then, the waiting time management device 13 calculates the average waiting time obtained by dividing the total waiting time by the number of operators as the first predicted time.
[0025] Note that the calculation method of the waiting time T1 is not limited to Equation (1). For example, the maximum value of the average time corresponding to each of the conversation contents "estimate request", "repair request", and "claim" ( "15 minutes") may be used as the waiting time T1.
[0026] The conversation model database 16 stores conversation position determination data D2 and conversation content determination data D3 that have been learned by artificial intelligence from past conversation contents, that is, conversation model information. FIG. 3 is a diagram showing an example of the conversation position determination data D2 in the call waiting prediction system 10 according to the present embodiment.
[0027] As shown in FIG. 3, the conversation position determination data D2 includes a pattern identifier column D21, a phrase pattern column D22, and a conversation position column D23. In the pattern identifier column D21, identification information of the conversation content, for example, "avuech20", "Ree4Eib5", etc. is stored. In the phrase pattern column D22, information representing the phrase pattern, for example, "service name confirmation", "repetition", etc. is stored. In the conversation position column D23, the conversation position corresponding to the phrase pattern, that is, information representing the position of the phrase pattern on the time axis from the start to the end of the conversation (the time axis of "average time" shown in FIG. 4 described later), for example, "20%", "5%", etc. is stored.
[0028] FIG. 4 is a diagram showing an example of conversation content determination data D3 in the call waiting prediction system 10 of the call center according to the present embodiment. As shown in FIG. 4, the conversation content determination data D3 includes a pattern identifier column D31, a conversation content column D32, and an average time column D33. In the pattern identifier column D31, identification information of the conversation content, for example, "Ree4Eib5", "avuech20", etc. are stored. In the conversation content column D32, information representing the patterned conversation content, for example, "product inquiry", "claim", etc. are stored. In the average time column D33, the average time of the call including the conversation content, for example, "7 minutes", "15 minutes", etc. are stored.
[0029] Here, for example, it is assumed that the conversation content of the telephone 11a acquired from the voice recognition device 14 is "claim" and the phrase pattern is "service name confirmation". In this case, the waiting time management device 13 extracts the pattern identifier "avuech20" and the average time "15 minutes" corresponding to "claim" from the conversation content determination data D3. Further, the waiting time management device 13 extracts the conversation position "20%" corresponding to the pattern identifiers "avuech20" and "service name confirmation" from the conversation position determination data D2 (see FIG. 3). The waiting time management device 13 calculates the time T2 until the current call ends using the extracted average time "15 minutes" and conversation position "20%". In the case of the above example, the time T2a until the call of the telephone 11a ends is 15 minutes × (1 - 20%) = 12 minutes. The waiting time management device 13 calculates the times T2a to T2c until the respective calls of the plurality of telephones (telephone 11a to telephone 11c) end as described above.
[0030] In addition, the waiting time management device 13 calculates the average time of times T2a to T2c until the respective calls of a plurality of telephones (telephones 11a to 11c) end as the second predicted time. Then, the waiting time management device 13 calculates (predicts) the total time of the first predicted time and the second predicted time as the call waiting time. Note that the calculation method of the second predicted time is not limited to the above-described calculation method. For example, the maximum value of times T2a to T2c until the respective calls of a plurality of telephones (telephones 11a to 11c) end may be used as the second predicted time.
[0031] [Call Waiting Time Prediction Processing] Next, a procedure of call waiting time prediction processing in the call waiting prediction system 10 of the call center will be described. FIG. 5 is a flowchart showing the procedure of call waiting time prediction processing in the call waiting prediction system 10 of the call center according to the present embodiment. The processing described below starts when a call from a user is received at the call center 1.
[0032] First, the line switch 12 receives a call from the user (step S1).
[0033] Next, the line switch 12 determines whether there is an available operator (step S2). In this process, the line switch 12 grasps the call status of the connected telephones 11a to 11c. When all the telephones are in a call, it is determined that there is no available operator, and step S2 becomes a No determination. On the other hand, when there is a telephone that is not in a call, the line switch 12 determines that there is an available operator, and step S2 becomes a Yes determination. Also, in this case, the line switch 12 acquires the number of telephones that are not in a call as the number of available operators.
[0034] In the process of step S2, when the line switch 12 determines that there is an available operator (Yes determination in step S2), any available operator connects to the corresponding telephone and starts a call (step S3).
[0035] On the other hand, in the process of step S2, when the circuit switch 12 determines that there is no available operator (in the case of No determination in step S2), it holds the connection to the operator and requests the waiting time management device 13 to predict the waiting time (step S4).
[0036] Next, the waiting time management device 13 calculates the average waiting time in the current time period (step S5). In this process, the waiting time management device 13 calculates the waiting time T1 in the current time period using the method described in FIG. 2 above. Also, the waiting time management device 13 calculates the first predicted time, which is the average waiting time in the current time period, using the number of users waiting for a call, the calculated waiting time T1, and the number of operators.
[0037] Next, the waiting time management device 13 calculates the time until the end of the call in progress (step S6). In this process, the waiting time management device 13 sends the call voice flowing in real time from the circuit switch 12 to the voice recognition device 14. The voice recognition device 14 outputs the patterned conversation content and phrase pattern (call content) recognized from the call voice to the waiting time management device 13. The waiting time management device 13 calculates the times T2a to T2c until the end of each call of each telephone during the call using the call content, the conversation position determination data D2, and the conversation content determination data D3 by the methods described in FIGS. 3 and 4 above. Also, the waiting time management device 13 calculates the average value of the times T2a to T2c as the second predicted time.
[0038] Next, the waiting time management device 13 predicts the total time of the calculated first predicted time and the second predicted time as the waiting time (step S7).
[0039] Next, based on the predicted waiting time and a predetermined time range registered in advance, the waiting time management device 13 instructs the circuit switch 12 to play back an automated voice (step S8). In this process, when the waiting time is within the predetermined time range, the waiting time management device 13 instructs the playback of an automated voice that plays a message prompting the user to wait for the call. Also, when the waiting time is not within the predetermined time range, the waiting time management device 13 instructs the playback of an automated voice that plays a message prompting the user to hang up and call again.
[0040] Next, the circuit switch 12 plays back an automated voice to the user according to the instruction from the waiting time management device 13 to play back an automated voice (step S9).
[0041] Next, the circuit switch 12 determines whether all the call waiting calls, that is, whether the calls waiting for a call with the operator have been disconnected (step S10).
[0042] In the process of step S10, when the circuit switch 12 determines that there are call waiting calls that have not been disconnected (in the case of a No determination in step S10), after a certain period of time has elapsed, it returns to the process of step S2 and repeatedly executes the processes of steps S2 to S10. Here, the certain period of time is, for example, appropriately registered in advance according to the call status in each time zone.
[0043] On the other hand, in the process of step S10, when the circuit switch 12 determines that all the call waiting calls have been disconnected (in the case of a Yes determination in step S10), or after the process of step S3, the call waiting time prediction process ends.
[0044] [Effect] As described above, in the call waiting prediction system 10 of the call center according to the present embodiment, when a user calls the call center 1 and is put on hold, the waiting time management device 13 calculates a first waiting time based on the current time zone and pre-registered conversation trend information. Further, the waiting time management device 13 calculates, as a second waiting time, the average time until each call of a plurality of telephones ends based on the telephone content recognized by the voice recognition device 14 and pre-registered conversation model information. Further, the waiting time management device 13 predicts the total time of the calculated first waiting time and the second waiting time as the call waiting time. Therefore, according to the call waiting prediction system 10 of the call center according to the present embodiment, it is possible to predict the call waiting time until the start of a call with an operator of the call center and reproduce an automatic voice according to the call waiting time.
[0045] Note that the present invention is not limited to the above-described embodiments, and it goes without saying that various other application examples and modification examples can be taken without departing from the gist of the present invention described in the claims. For example, the above-described embodiment has described the configuration of the call waiting prediction system 10 of the call center in detail and specifically for the purpose of explaining the present invention clearly, and is not necessarily limited to the one having all the configurations described. Further, it is possible to replace a part of the configuration of the embodiment described here with the configuration of another embodiment, and further, it is also possible to add the configuration of another embodiment to the configuration of a certain embodiment. Also, it is possible to add, delete, or replace other configurations for a part of the configuration of the embodiment. Also, the control lines and information lines show those considered necessary for explanation, and not necessarily all the control lines and information lines are shown on the product. In fact, it may be considered that almost all the components are interconnected.
Description of Reference Numerals
[0046] 1…Call center, 10…Call waiting prediction system, 11a~11c…Telephone sets, 12…Line switch, 13…Waiting time management device, 14…Speech recognition device, 15…Conversation statistical database, 16…Conversation model database
Claims
1. A call waiting prediction system for a call center comprising a plurality of telephones and a circuit switch having a function of allocating calls to the plurality of telephones and a function of playing back an automatic voice, a voice recognition device that recognizes the telephone content during a call of each of the plurality of telephones, a waiting time management device that predicts the call waiting time until the start of the call based on the current time zone, the telephone content recognized by the voice recognition device, pre-registered conversation statistical information, and conversation model information when a call is made to the call center and the call is put on hold, A call waiting prediction system for a call center.
2. The waiting time management device calculates, as a first waiting time, the average waiting time of calls in the current time zone based on the current time zone and the conversation statistical information The call waiting prediction system for a call center according to Claim 1.
3. The waiting time management device calculates, as a second waiting time, the average time until the end of each call of the plurality of telephones based on the telephone content recognized by the voice recognition device and the conversation model information The call waiting prediction system for a call center according to Claim 2.
4. The waiting time management device predicts the total time of the first waiting time and the second waiting time as the call waiting time The call waiting prediction system for a call center according to Claim 3.
5. When the waiting time management device determines that the predicted call waiting time is within a predetermined time range, it outputs a reproduction instruction for automatic voice that plays back a message prompting to wait for the call to the circuit switch The call waiting prediction system for a call center according to Claim 4.
6. When the waiting time management device determines that the predicted call waiting time is not within a predetermined time range, it outputs a reproduction instruction for automatic voice that plays back a message prompting that it is currently congested and to cut off the call and call again to the circuit switch The call waiting prediction system for a call center according to Claim 4.
7. The telephone content includes patterned telephone conversation content and phrase patterns The call waiting prediction system for a call center according to Claim 1.
8. A call waiting prediction method in a call center call waiting prediction system including a plurality of telephones and a circuit switch having a function of allocating calls to the plurality of telephones and a function of playing back automatic voice, comprising: recognizing the telephone content during a call for each of the plurality of telephones; when a call is made to the call center and the call is put on hold, predicting the call waiting time until the start of the call based on the current time zone, the recognized telephone content, pre-registered conversation statistical information, and conversation model information. A call waiting prediction method for a call center.
Citation Information
Patent Citations
Call unit detection apparatus, method and program
JP2011259127A