JP5424332B2 - Telephone response result prediction apparatus, method, and program thereof - Google Patents

Description

  The present invention relates to a telephone answering work management system for supporting a telephone answering work, and in particular, an apparatus for sequentially analyzing call voice during a call between a customer and an operator and predicting a result of the telephone answer during the call. , Methods, and programs.

  Improving customer satisfaction is an important management issue for companies. In particular, the telephone answering service at the call center is a window where customers and companies directly contact each other. For this reason, there is an increasing need to improve the contents of telephone response work in order to improve customer satisfaction.

  On the other hand, in the telephone answering work in the call center, the operator in charge of the work may have to make a call with the customer who makes a call in a state of dissatisfaction. Such communication with customers often imposes a psychological burden on the operator. Because of this psychological burden, the operator turnover rate is generally high. In order to promote the establishment of the operator, maintaining the employee satisfaction of the operator is also an issue.

  Further, even if the customer satisfaction and employee satisfaction can be improved by the measures, if the response time for one customer becomes longer as a result, the labor cost of the operator per customer increases. Since this results in pressure on the management of the call center, it is also an issue to reduce the response time for one customer to an appropriate level.

  Here, in this specification, customer satisfaction regarding calls, employee satisfaction regarding calls, and call duration of calls are collectively referred to as “telephone response results”.

  Various technical ideas for measuring customer satisfaction have been proposed for telephone response operations in call centers.

  For example, in Japanese Patent Laid-Open No. 2002-189837 (Patent Document 1), at the end of a telephone response of a responder to an inquiry telephone, voice guidance for inquiring customer satisfaction to a connected telephone is performed from the voice response device. A technical idea for receiving response signal information for voice guidance by a telephone and recording the response signal information as customer satisfaction data is disclosed.

  Japanese Patent Laid-Open No. 2003-316944 (Patent Document 2) presents a question to a customer in order to receive additional feedback regarding customer satisfaction with respect to a call before the call is disconnected, and touch tone. -The technical idea of receiving data and obtaining customer satisfaction is disclosed.

  Japanese Patent Laying-Open No. 2005-252845 (Patent Document 3) stores call contents in a server, measures the volume and tone of the stored call contents, extracts a keyword by voice morphological analysis, and extracts the voice volume. Further, a technical idea of measuring a CS (customer satisfaction) level based on a change in tone and a detection result of a specific keyword is disclosed.

JP 2002-189837 A JP 2003-316944 A JP 2005-252845 A

  The techniques disclosed in Patent Document 1 and Patent Document 2 acquire customer satisfaction regarding the business by asking the customer after the telephone response business at the call center is completed. Therefore, with the techniques of Patent Document 1 and Patent Document 2, it is impossible to sequentially measure or predict customer satisfaction during the telephone response work. Therefore, these technologies are used to predict customer satisfaction (call response results) during the call using the results of sequential analysis during the call and to present it to the operator or supervisor who is the supervisor. It is impossible to use it.

  The technique disclosed in Patent Document 3 sets a predetermined condition and measures customer satisfaction based on the volume of voice and tone of accumulated call contents and the detection result of a specific keyword. Is. Here, the predetermined condition to be set is, for example, that the CS level decreases by one step when the voice volume or tone increases by 10%, and the CS level decreases by two steps when the voice volume or tone increases by 20%. If detected, the CS level is increased by one step. However, this method cannot verify whether the CS-level evaluation criteria are correct or incorrect. Here, whether the CS level evaluation standard is correct is, for example, that the CS level decreases by one step when the voice volume or tone increases by 10%, and the CS level decreases by two steps when the voice level or tone increases by 20%. Whether the CS level is simply proportional or how the CS level is measured when the same keyword is repeated many times. Therefore, this technique cannot accurately calculate customer satisfaction.

  Furthermore, in any of the techniques disclosed in Patent Documents 1 to 3, only customer satisfaction is measured. Employee satisfaction and call duration are not measured by these technologies, although they are factors that are a challenge for call center management.

  In view of this, it is conceivable to provide a technique for improving customer satisfaction with respect to telephone answering operations in a call center, improving employee satisfaction, or preventing an increase in call time, and thus contributing to improvement of call center management. As one of such technologies, for example, the call voice is sequentially analyzed during a call between a customer and an operator, and a telephone response result (customer satisfaction, employee satisfaction, and A technique of sequentially predicting (call time) in the middle of a call and presenting it to the operator and the supervisor who is the supervisor is conceivable. If the technology with such functions can be used in the call center, the operator and supervisor can quickly improve customer satisfaction by referring to the prediction result of the telephone response result presented during the telephone response work. In addition, it is considered that some measures can be taken to improve employee satisfaction or prevent an increase in call time.

  Therefore, the problem to be solved by the present invention is that telephone response results (customer satisfaction, employee satisfaction, and call time) can be sequentially predicted and presented in the middle of a call with respect to the telephone response work in a call center. Is to do so.

  The telephone response result prediction method according to the present invention sequentially analyzes each call voice while the customer and the operator are talking, calculates and records the attribute, and after the call ends, the telephone response result relating to each call voice Input and record the collective attribute, analyze and record the telephone response relationship between the sequential attribute and the collective attribute, and the sequential analysis result (sequential attribute) of the call voice during the current call and the recorded telephone response relationship The predicted value of the telephone response result during the current call is predicted and presented during the current call.

  The telephone response result prediction apparatus according to the present invention includes a sequential attribute calculation unit that sequentially analyzes a call voice of a call while a customer and an operator are talking and calculates a sequential attribute, and individual call voices after the call ends. A batch attribute calculation unit that inputs a telephone response result as a batch attribute, an attribute recording unit that records a sequential attribute and a batch attribute as a past sequential attribute and a past batch attribute, respectively, a past sequential attribute and a past attribute The relationship calculation unit that analyzes the telephone response relationship with the collective attribute, the relationship recording unit that records the analysis result of the telephone response relationship as the past telephone response relationship, and the sequential attribute during the current call calculated by the sequential attribute calculation unit In response, a prediction presentation unit that sequentially predicts and presents a predicted value of the telephone response result during the current call based on the past telephone response relationship recorded in the relationship recording unit.

  According to the present invention, on the assumption (rule of thumb) of the telephone response relationship between the sequential attribute at the end of the call and the collective attribute accumulated until the start of the call, the call voice during the current call between the customer and the operator is By sequentially measuring and calculating the attribute during the call, the predicted value of the telephone response result during the current call can be sequentially calculated and presented during the call.

It is a functional block diagram which shows the telephone response result prediction apparatus which concerns on one Embodiment of this invention. It is a block diagram which shows the example which implement | achieved the telephone response result prediction apparatus of FIG. 1 using the general computer (electronic computer). It is drawing which showed the example of the measurement result by various sequential measurement means with a table | surface. It is drawing which showed the example of the calculation result of the sequential attribute calculated from the measurement result illustrated by FIG. 3 with the table | surface. It is drawing which showed the example of the recording content which the attribute recording part 131 records and accumulate | stores with the table | surface. 6 is a diagram illustrating an example of a relationship between a sequential attribute and a collective attribute at a call end time calculated by a relationship calculation unit 141. It is explanatory drawing which shows the main operation | movement procedure of this Embodiment with a flowchart.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  A first embodiment of the present invention will be described with reference to FIGS.

  In the drawings, parts according to the present invention are shown, and some of the related essential components are not shown. Further, changes such as division / merging and replacement before and after the procedure by the function of the illustrated block are free as long as the functions obtained from the gist of the present invention and the description of the claims are satisfied, and the following description is provided by the present invention. It is not intended to limit.

  In the description of the present invention, a functional block diagram is used in order to clarify the invention. In the functional block diagram, only important parts are shown, and some of the related components are omitted.

  In addition, since the present invention is realized by software, various functions / means are described as functional blocks. However, changes such as division / merging of functional blocks and replacement before and after processing are the gist of the present invention and claims. As long as the functions obtained from the description are satisfied, the description is free, and the following description of the embodiment does not limit the present invention.

  First, terms used to describe the embodiment are defined.

  A “customer” is a person who calls a call center for inquiries, applications, and the like.

  An “operator” is a call center employee who handles calls from customers.

  “Customer satisfaction” is the degree of satisfaction felt by the customer regarding the telephone conversation between the customer and the operator.

  “Employee satisfaction” is the degree of satisfaction felt by an operator regarding a telephone conversation between a customer and the operator.

  The “call time” is the length of time from when a customer calls the call center with a request until the customer hangs up.

  Next, a first embodiment of the present invention will be described with reference to FIG. FIG. 1 is a functional block diagram showing a telephone response result prediction apparatus 100.

  1 includes a sequential attribute calculation unit 111, a collective attribute calculation unit 121, an attribute recording unit 131, a relationship calculation unit 141, a relationship recording unit 151, and a prediction presentation unit 161. Is done.

  Note that the telephone response result prediction apparatus 100 is realized using an electronic computer. The electronic computer may be a general computer or may be realized by using a dedicated hardware configuration. If the example which implement | achieves customer satisfaction, employee satisfaction, and the telephone call time prediction apparatus 100 using a general computer is shown, the structure of FIG. 2 will be mentioned.

  FIG. 2 is a block diagram showing an example in which the telephone response result prediction apparatus 100 is realized using a general computer.

  2 includes a control unit 210 that performs various types of information processing, a ROM (read-only memory) 220, a RAM (random access memory) 230, an input unit 240, an output unit 250, and a storage device 110. And a network interface 260 or the like as necessary.

  The storage device 110 of the telephone response result prediction apparatus 100 stores various programs such as an operating system (OS) 270 and an application program 280 that realizes various functions and means, and is expanded in the RAM 230 as necessary. To be executed. The storage device 110 can store various types of information, and in addition to the program 280 described above, a data file in which sample data 290 and the like are recorded is also recorded. An HDD (hard disk drive) can be used as the storage device, but is not limited thereto, and any storage means that can reliably store necessary information may be used. For example, an empty storage area of flash memory or RAM may be used.

  In this example, the sequential attribute calculation unit 111, the collective attribute calculation unit 121, the attribute recording unit 131, the relationship calculation unit 141, the relationship recording unit 151, and the prediction presentation unit 161 are controlled by executing the application program 280. This is realized by causing the unit 210 to function as various means. That is, the application program 280 includes a program for causing the computer to function as the sequential attribute calculation unit 111, the collective attribute calculation unit 121, the attribute recording unit 131, the relationship calculation unit 141, the relationship recording unit 151, and the prediction presentation unit 161. included.

  Further, the application program 280 causes the control unit 210 to function as the sequential attribute calculation unit 111, the collective attribute calculation unit 121, the attribute recording unit 131, the relationship calculation unit 141, the relationship recording unit 151, and the prediction presentation unit 161. A program for processing the sample data input by the unit 240, outputting the calculation result to the output unit 250 as necessary, and outputting it to a printer or the like is included.

  Next, returning to FIG. 1, each functional block will be described in detail. Each functional block operates when the control unit 210 performs arithmetic processing according to the program 280.

  The sequential attribute calculation unit 111 calculates sequential attributes related to individual call voices. In other words, the sequential attribute calculation unit 111 sequentially measures the call voice in the middle of each call, and calculates the sequential attribute for each call from the measured amount. That is, the sequential attribute calculation unit 111 sequentially analyzes the call voice of the call while the customer and the operator are talking, and calculates the sequential attribute for each call.

  Here, the sequential attribute is an attribute calculated from the measured amount by sequentially measuring the call voice during the call between the customer and the operator.

  The sequential attribute calculation unit 111 includes sequential measurement means that sequentially measures call voice during a call. Examples of the sequential measurement means include a sound volume measurement means, an utterance speed measurement means, and a speech morpheme analysis means.

  The sound volume measuring means is a means for measuring the volume (volume) of the call voice, for example, at intervals of 1 second.

  The speech speed measuring means is a means for measuring the number of vowel appearances (speech speed) in the call voice, for example, at intervals of 1 second.

  The voice morpheme analysis means is means for converting a call voice into character data using a voice recognition tool and detecting a word from the character data using a morpheme analysis tool.

  In the illustrated embodiment, the sequential measurement means includes a combination of volume measurement means, speech rate measurement means, and speech morpheme analysis means, but is not limited to this. That is, the sequential measurement unit may be composed of any one of a volume measurement unit, a speech rate measurement unit, and a speech morpheme analysis unit, or a combination thereof. Specifically, the sequential measurement means may be composed of only the sound volume measurement means, may be composed only of the speech speed measurement means, or may be composed only of the speech morphological analysis means, and the sound volume measurement means and the speech It may be composed of a combination of speed measurement means, may be composed of a combination of sound volume measurement means and speech morpheme analysis means, or may be composed of a combination of speech speed measurement means and speech morpheme analysis means.

  The sequential attribute is an attribute calculated from the amount measured by these sequential measurement means. The sequential attribute is, for example, a maximum volume value, an average volume value, and a ratio between the maximum volume value and the average value as an attribute calculated from the volume (volume) of the call voice measured by the volume measuring unit. , The maximum value of the change rate of the volume, the average value of the change rate of the volume, and the like. The sequential attribute is, for example, an attribute calculated from the number of appearances of vowels (speech speed) in the call voice measured by the speech speed measuring means, such as the maximum value of the speech speed, the average value of the speech speed, and the speech speed. There are the ratio between the maximum value and the average value, the maximum value of the change rate of the speech rate, the average value of the change rate of the speech rate, and the like. The sequential attribute includes, for example, an accumulated value of the number of times a specific word is detected as an attribute calculated from the word detected by the speech morphological analysis unit.

  Note that the sequential attribute calculation unit 111 obtains the call voice between the customer and the operator by mixing the input voice from the customer side and the input voice from the operator side, and calculates the sequential attribute from the call voice. good. Alternatively, the sequential attribute calculation unit 111 obtains the call voice between the customer and the operator by separating the input voice from the customer side and the input voice from the operator side, and calculates the sequential attribute from the call voice. good.

  Here, the operation of the sequential attribute calculation unit 111 is illustrated.

  FIG. 3 shows examples of measurement results obtained by various sequential measurement means. The volume, the speech rate, and the specific word are measured at intervals of 1 second by the sequential measurement unit. The measurement is continuously performed from the start to the end of the call. In the example shown in FIG. 3, it is assumed that the measurement is continuously performed at intervals of 1 second until the end of the call after 50 seconds have passed since the start of the call.

  FIG. 4 is an example of calculation results of sequential attributes calculated from the measurement results illustrated in FIG. As the sequential attribute, here the maximum value of the volume, the maximum value of the change rate of the volume, the maximum value of the utterance speed, the maximum value of the change rate of the utterance speed, the cumulative number of detection of "Oi", the cumulative detection of "Iikengeniro" The number of times has been calculated.

  As illustrated in FIG. 4, the maximum value of the sound volume, the maximum value of the change rate of the sound volume, the maximum value of the utterance speed, the maximum value of the change rate of the utterance speed, the cumulative number of detections of “oi”, The sequential attribute of the cumulative number of times of detection is calculated as 1.0, no value, 3.0, no value, 0 times, and 0 times after 1 second from the start of the call. After 1 second has passed, and 2 seconds after the start of the call, it is calculated as 3.0, 2.0 / 4.0, 1.0, 1.0 / second, and 0 / second. Further, 1 second elapses, and 3 seconds after the start of the call, it is calculated as 3.0, 2.0 / 4.0, 4.0, 1.0 / second, 0 times / second. Further, 1 second elapses, and 4 seconds after the start of the call, 3.0, 2.0, 6.0, 3.0, 2, and 0 are calculated per second. Further, 1 second elapses, and after 5 seconds from the start of the call, it is calculated as 5.0, 4.0, 6.0 per second, 3.0 per second, 2 times, and 1 time. Thereafter, the maximum value of the volume, the maximum value of the change rate of the volume, the maximum value of the speech rate, the maximum value of the change rate of the speech rate, the cumulative number of detections of “Oi”, and the cumulative number of detections of “Iikengeniro” It is assumed that the sequential attribute has not changed from the value after 5 seconds from the start of the call, but has changed until the end of the call after 50 seconds have passed since the start of the call. At the end of the call, it is calculated as 5.0, 4.0, 6.0 per second, 3.0 per second, twice, and once.

  The collective attribute calculation unit 121 calculates collective attributes related to individual call voices. In other words, the collective attribute calculation unit 121 calculates a telephone response result regarding each call as a collective attribute for each call. That is, the collective attribute calculation unit 121 calculates a telephone response result regarding each call voice after the call is finished as a collective attribute for each call.

  The collective attribute (telephone response result) is an attribute calculated for the call after the call between the customer and the operator ends. In the illustrated example, the collective attribute (telephone response result) includes customer satisfaction regarding the call, employee satisfaction regarding the call, and call duration of the call.

  In the illustrated embodiment, the collective attribute (phone response result) includes a combination of customer satisfaction, employee satisfaction, and call time, but is not limited thereto. In other words, the collective attribute (telephone response result) may be configured by any one or a combination of customer satisfaction, employee satisfaction, and call time. Specifically, the collective attribute (telephone response result) may be composed only of customer satisfaction, may be composed only of employee satisfaction, may be composed only of call time, It may consist of a combination of customer satisfaction and employee satisfaction, may consist of a combination of customer satisfaction and talk time, or a combination of employee satisfaction and talk time Also good.

  As a method for calculating customer satisfaction regarding a call, for example, a method in which an operator asks the customer about the degree of satisfaction after the call ends and obtains an answer, and after the call ends, the customer is asked about the degree of satisfaction with a voice response device. There is a method of obtaining an answer and a method of sending a document to a customer after a call and asking a degree of satisfaction to obtain an answer.

  The customer satisfaction is expressed by a numerical value with a perfect score of 100 points, for example, and the numerical value is obtained by asking the customer how many points were given when the satisfaction level was 100 points.

  As a method for calculating employee satisfaction regarding a call, for example, there is a method of obtaining a reply by presenting a document to an operator after the call is finished, asking a degree of satisfaction.

  The employee satisfaction is expressed, for example, by a numerical value with a perfect score of 100 points, and the numerical value is obtained by asking the operator how many points the satisfaction degree is when the perfect score is 100 points.

  As a method of calculating the call time, for example, there is a method of measuring the time from the start to the end of the call.

  Here, the operation of the collective attribute calculation unit 121 is illustrated.

  For example, it is assumed that a customer's degree of satisfaction regarding a call is 80 points as a result of a question to the customer regarding a certain call. Further, it is assumed that the employee satisfaction with respect to the call is 70 points as a result of the question to the operator. Further, it is assumed that the call time is measured as 50 seconds. At this time, the collective attribute calculation unit 121 calculates 80 customer satisfaction regarding the call, 70 employee satisfaction regarding the call, and 50 seconds of call time.

  The attribute recording unit 131 records and accumulates the sequential attribute at the end of the call related to each call calculated by the sequential attribute calculation unit 111 and the collective attribute related to the call calculated by the collective attribute calculation unit 121. In other words, the attribute recording unit 131 sequentially accumulates the sequential attribute at the end of the call calculated for each call by the sequential attribute calculation unit 111 and the collective attribute calculated for each call by the collective attribute calculation unit 121, respectively. The past sequential attributes and the past collective attributes are recorded together. That is, the attribute recording unit 131 sequentially accumulates the sequential attributes and the collective attributes at the end of the call for each call, and records them collectively as past sequential attributes and past collective attributes, respectively.

  Here, the operation of the attribute recording unit 131 is exemplified.

  FIG. 5 is a table showing an example of recorded contents recorded and accumulated by the attribute recording unit 131.

  The attribute recording unit 131 records, for example, sequential attributes and collective attributes at the end of the call for 1000 calls made in the past from the call number “0001” to the call number “1000”. Here, for example, when a call between a customer and an operator is performed and the call is ended, the maximum value of the volume is 5.0 as a sequential attribute at the time of the end of the call, and the change of the volume. The maximum value of rate is 4.0, the maximum value of speech rate is 6.0, the maximum value of rate of change of speech rate is 3.0, the cumulative number of detections of “Oi” is 2 times, It is assumed that the cumulative number of times of detection is calculated as 1. Further, it is assumed that the customer satisfaction regarding the call is 80 points, the employee satisfaction regarding the call is 70 points, and the call time is 50 seconds as a collective attribute. The attribute recording unit 131 assigns a call number of “1001” to the call, and calculates the calculation results of the sequential attribute and the collective attribute at the end of the call related to the call. As a collective attribute, it is recorded and accumulated in addition to the recorded contents accumulated so far.

  The relationship calculating unit 141 refers to the sequential attribute and the collective attribute at the end of the call recorded by the attribute recording unit 131, and the relationship between the sequential attribute and the collective attribute at the end of the call accumulated up to the reference time Is calculated. In other words, the relationship calculating unit 141 refers to the past sequential attribute and the past collective attribute recorded in the attribute recording unit 131, and determines the telephone response relationship between the past sequential attribute and the past collective attribute. Calculate for each call. That is, the relationship calculation unit 141 analyzes the telephone response relationship between past sequential attributes and past collective attributes, and calculates this telephone response relationship for each call.

  The relationship between the sequential attribute at the end of the call and the collective attribute (telephone response relationship) is as follows: after each call, the sequential attribute at the end of the call accumulated up to that point is used as the explanatory variable, and the collective attribute is used as the dependent variable. The relationship is calculated by performing multiple regression analysis.

  Here, the operation of the relationship calculation unit 141 will be exemplified.

  FIG. 6 is a diagram showing an example of the relationship between the sequential attribute and the collective attribute at the end of the call calculated by the relationship calculating unit 141.

  For example, it is assumed that, at the start of a certain call, the attribute recording unit 131 stores sequential attributes and collective attributes at the end of the call for 1000 calls as past sequential attributes and past collective attributes, respectively. At the end of the call, the relationship calculation unit 141 performs a multiple regression analysis on 1001 calls including the call, with the sequential attribute at the end of the call as an explanatory variable and the collective attribute as a dependent variable.

  As a result of the multiple regression analysis, the relationship between the sequential attribute and the collective attribute at the end of the call (telephone response relationship) is, for example, the maximum value of the volume, the maximum value of the change rate of the volume, and the speech as the sequential attribute at the end of the call The maximum value of the speed, the maximum value of the change rate of the speech rate, the cumulative number of detections of “Oi”, and the cumulative number of detections of “Oikengeniro” are represented as x1, x2, x3, x4, x5, and x6, respectively. When the customer satisfaction (100 points maximum), employee satisfaction (100 points maximum), and talking time (seconds) as the collective attributes are expressed as y1, y2, and y3, respectively, as follows: Calculated.

y1 = 120.0-2.0.x1-5.0.x3-10.0.x5
y2 = 110.0-3.0.x2-10.0.x4-20.0.x6
y3 = 10.0 + 2.0 · x1 + 1.0 · x2 + 2.0 · x3
+ 2.0 · x4

  The relationship between the sequential attribute at the end of the call and the collective attribute (phone response relationship) is calculated based on the sequential attribute and collective attribute data accumulated at the end of the call at the end of each call. The Therefore, the accuracy of the relational expression is improved as the amount of accumulated data increases. Thereby, more accurate sequential prediction is possible.

  The relationship recording unit 151 records and accumulates the relationship (telephone response relationship) between the sequential attribute and the collective attribute at the end of the call calculated by the relationship calculating unit 141. In other words, the relationship recording unit 151 records the telephone response relationship calculated for each call by the relationship calculation unit 141 as the past telephone response relationship. That is, the relationship recording unit 151 records the telephone response relationship calculated for each call as a past telephone response relationship.

  The prediction presentation unit 161 is based on the relationship between the sequential attribute and the collective attribute at the end of the call recorded by the relationship recording unit 151 (telephone response relationship) and the sequential attribute regarding each call calculated by the sequential attribute calculation unit 111. , Calculate (predict) and present the telephone response results (customer satisfaction, employee satisfaction, and call duration) during the current call. In other words, the prediction presenting unit 161 responds to the sequential attribute during the current call calculated by the sequential attribute calculating unit 111, based on the past telephone response relationship recorded in the relationship recording unit 151, the current call The predicted value of the telephone response result (customer satisfaction, employee satisfaction, and call time) is predicted and presented during the current call. That is, the prediction presenting unit 161 responds to the sequential attribute during the current call calculated by the sequential attribute calculation unit 111, and based on the past telephone response relationship recorded in the relationship recording unit 151, Predicted values of response results (customer satisfaction, employee satisfaction, and call time) are sequentially predicted and presented.

  The presented telephone response results (customer satisfaction, employee satisfaction, and call duration) are the relationship between the sequential attribute and the collective attribute obtained at the end of the call until the start of the call (phone response relationship) As a premise (rule of thumb), the final call response results (customer satisfaction, employee satisfaction, and call time) of the call are calculated sequentially by applying the sequential attributes obtained sequentially for the call. It is a predicted value.

  Here, the operation of the prediction presentation unit 161 will be exemplified.

  Assume that the relationship (telephone response relationship) illustrated in FIG. 6 is obtained as the relationship (telephone response relationship) between the sequential attribute and the collective attribute at the end of the call. That is, as the sequential attribute at the end of the call, the maximum value of the volume, the maximum value of the change rate of the volume, the maximum value of the speech rate, the maximum value of the rate of change of the speech rate, the cumulative number of detections of “Oi”, and “ The cumulative detection count of “Niro” is expressed as x1, x2, x3, x4, x5, and x6, respectively, and customer satisfaction (100 points maximum), employee satisfaction (100 points maximum), When the call time (second) is expressed as y1, y2, and y3, respectively, it is assumed that there is the following relationship between them.

y1 = 120.0-2.0.x1-5.0.x3-10.0.x5
y2 = 110.0-3.0.x2-10.0.x4-20.0.x6
y3 = 10.0 + 2.0 · x1 + 1.0 · x2 + 2.0 · x3
+ 2.0 · x4

  Further, it is assumed that the sequential attributes illustrated in FIG. 4 are obtained. That is, the maximum value of the volume, the maximum value of the change rate of the volume, the maximum value of the speaking rate, the maximum value of the changing rate of the speaking rate, the cumulative number of detections of `` Ooi '', and the cumulative number of detections of `` Iikengeniro '' 2 seconds after the start of the call, 3.0, 2.0 / 4.0, 4.0, 1.0 / second, 1 time, 0 times, 3 seconds after the start of the call, 3.0, 2.0 / second, 4 0.0, 1.0 once per second, 0 times, 4 seconds after the start of the call, 3.0, 2.0 per second, 6.0, 3.0 per second, 2 times, 0 times, 5 after the start of the call After the second, 5.0, 4.0, 6.0 per second, 3.0 per second, 2 times, 1 time, and thereafter, these values do not change until the end of the call, 5.0, per second 4.0, 6.0, 3.0 / second, 2 times, 1 time.

  At this time, the final customer satisfaction, employee satisfaction, and call duration prediction values of the call are 84.0, 94.0, and 28.0 seconds after the start of the call, respectively. 34.0 seconds after the start of the call, 84.0, 94.0, 28.0 seconds, 4 seconds after the start of the call, 64.0, 74.0, 36.0 seconds, 5 seconds after the start of the call, 60. It is calculated as 60.0, 48.0, 42.0 seconds without changing until 0, 48.0, 42.0 seconds, and after the end of the call.

  Next, the operation of the telephone response result prediction apparatus according to the present embodiment will be described with reference to FIG. FIG. 7 is an explanatory diagram showing the main operation procedure of the present embodiment in a flowchart.

  The telephone response result prediction apparatus 100 starts to operate when a call between a customer and an operator is started. First, the prediction presenting unit 161 refers to the relationship (telephone response relationship) between the sequential attribute and the collective attribute at the end of the call recorded by the relationship recording unit 151 (step S11).

  Next, the sequential attribute calculation unit 111 sequentially analyzes the call voice of the call, for example, at intervals of 1 second, and calculates a sequential attribute related to the call (step S12).

  Next, the prediction presenting unit 161 relates the relationship between the sequential attribute at the end of the call referenced in step S11 and the collective attribute (telephone response relationship), and the sequential attribute related to each call calculated by the sequential attribute calculation unit 111 in step S12. Thus, the telephone response result (customer satisfaction, employee satisfaction, and call time) during the current call is calculated (predicted), for example, at intervals of one second (step S13).

  Next, it is determined whether the call between the customer and the operator is continued, but has ended. If the call continues, steps S12 and S13 are repeated. If the call has ended, the process proceeds to the next procedure (procedure S14).

  Next, the attribute recording unit 131 records the sequential attribute at the end of the call calculated by the sequential attribute calculation unit 111 (step S15).

  Next, the collective attribute calculation unit 121 calculates collective attributes related to individual call voices (step S16).

  Next, the attribute recording unit 131 records the collective attribute calculated by the collective attribute calculation unit 121 (step S17).

  Next, the relationship calculating unit 141 refers to the sequential attribute and batch attribute at the end of the call recorded by the attribute recording unit 131, and the sequential attribute and batch attribute at the end of the call accumulated up to the reference time. (Step S18).

  The present invention has been described above with reference to the embodiments, but the present invention is not limited to the above embodiments. Various changes that can be understood by those skilled in the art can be made to the configuration and details of the present invention within the scope of the present invention.

  For example, the collective attribute may be composed of a satisfaction level that includes one or both of a customer satisfaction level and an employee satisfaction level. In this case, a satisfaction relationship acquisition apparatus, method, and program thereof can be provided by a combination of the sequential attribute calculation unit 111, the collective attribute calculation unit 121, and the relationship calculation unit 141. Instead, a satisfaction degree prediction apparatus, method, and program thereof can be provided by a combination of the sequential attribute calculation unit 111, the collective attribute calculation unit 121, the relationship calculation unit 141, and the prediction presentation unit 161. Alternatively, the collective attribute may be configured only from the call time. In this case, a time relationship acquisition apparatus, method, and program thereof can be provided by a combination of the sequential attribute calculation unit 111, the collective attribute calculation unit 121, and the relationship calculation unit 141. Instead, a call duration prediction apparatus, method, and program thereof can be provided by a combination of the sequential attribute calculation unit 111, the collective attribute calculation unit 121, the relationship calculation unit 141, and the prediction presentation unit 161.

  The present invention is based on the premise (rule of thumb) of the relationship between the sequential attribute at the end of the call accumulated until the start of the call and the collective attribute (telephone response relationship). By measuring and calculating the sequential attribute, it is possible to calculate and present the predicted value of the telephone response results (customer satisfaction, employee satisfaction, and call time) sequentially. In order to quickly improve customer satisfaction, improve employee satisfaction, or prevent any increase in call time by referring to the prediction results presented during the telephone response It can be applied to any use.

DESCRIPTION OF SYMBOLS 100 Telephone response result prediction apparatus 111 Sequential attribute calculation part (sequential attribute calculation means)
121 Collective attribute calculation unit (collective attribute calculation means)
131 Attribute recording unit (attribute recording means)
141 Relationship Calculation Unit (Relation Calculation Unit)
151 Relationship recording unit (relation recording means)
161 Prediction presentation unit (prediction presentation means)

Claims (33)

A sequential attribute calculation unit that sequentially measures call voice in the middle of each call and calculates a sequential attribute from the measured amount for each call;
A collective attribute calculation unit for calculating a call response result for each call as a collective attribute for each call;
The sequential attribute calculated for each call by the sequential attribute calculation unit and the collective attribute calculated for each call by the collective attribute calculation unit are sequentially accumulated, respectively. An attribute recording unit that collectively records as a collective attribute;
A relationship calculation unit that calculates a telephone response relationship between the past sequential attribute and the past batch attribute for each call with reference to the past sequential attribute and the past batch attribute recorded in the attribute recording unit. When,
A relationship recording unit that records the telephone response relationship calculated for each call by the relationship calculation unit as a past telephone response relationship;
In response to the sequential attribute during the current call calculated by the sequential attribute calculation unit, based on the past telephone response relationship recorded in the relationship recording unit, the predicted value of the telephone response result during the current call is calculated. A prediction presentation unit that predicts and presents the call during the current call;
A telephone response result prediction apparatus characterized by comprising:   The telephone response result prediction apparatus according to claim 1, wherein the telephone response result includes one or a combination of customer satisfaction related to the call, employee satisfaction related to the call, and call duration of the call. .   The telephone response result prediction apparatus according to claim 1, wherein the sequential attribute calculation unit includes a sequential measurement unit that sequentially measures call voice during a call.   The sequential measurement means includes a volume measurement means for measuring the volume of the call voice, an utterance speed measurement means for measuring an utterance speed of the call voice, and a speech morpheme analysis for analyzing a voice morpheme of the call voice and detecting a word. 4. The telephone response result prediction apparatus according to claim 3, comprising any one of the means or a combination thereof.   The sequential attribute is an attribute calculated from the sound volume measured by the sound volume measuring means, such as a maximum sound volume value, an average sound volume value, a ratio between the maximum sound volume value and the average value, a maximum sound volume change rate, 5. The telephone response result prediction apparatus according to claim 4, which includes any one of an average value of volume change rate and a combination thereof, or a combination thereof.   The sequential attribute is an attribute calculated from the utterance speed measured by the utterance speed measuring means, as a maximum value of the utterance speed, an average value of the utterance speed, a ratio between the maximum value of the utterance speed and the average value, an utterance speed The telephone response result prediction apparatus according to claim 4, comprising any one of a maximum value of change rate and an average value of change rate of speech rate or a combination thereof.   The telephone response result prediction apparatus according to claim 4, wherein the sequential attribute includes an accumulated value of the number of specific words as an attribute calculated from the words detected by the speech morphological analysis unit.   The said sequential attribute calculation part mixes the input voice from the customer side, and the input voice from the operator side, acquires the said call voice, and calculates the said sequential attribute from the said call voice. The telephone response result prediction apparatus as described in any one.   The said sequential attribute calculation part isolate | separates the input audio | voice from a customer side, and the input audio | voice from an operator side, acquires the said call audio | voice, and calculates the said sequential attribute from the said call audio | voice. The telephone response result prediction apparatus as described in any one.   The relationship calculation unit calculates a telephone response relationship by performing multiple regression analysis using the past sequential attribute as an explanatory variable and the past batch attribute as a dependent variable. The telephone response result predicting apparatus according to 1. A step of sequentially measuring the call voice in the middle of each call, and calculating a sequential attribute from the measured amount for each call;
Calculating a telephone response result for each call as a collective attribute for each call;
The sequential attribute calculated for each call and the sequential attribute calculated for each call and the collective attribute calculated for each call are sequentially stored in the attribute recording unit, and collectively as a past sequential attribute and a past collective attribute, respectively. Recording in the attribute recording unit;
Calculating a telephone response relationship between the past sequential attribute and the past batch attribute for each call with reference to the past sequential attribute and the past batch attribute recorded in the attribute recording unit;
Recording the telephone response relationship calculated for each call in a relationship recording unit as a past telephone response relationship;
In response to the calculated sequential attribute during the current call, based on the past telephone response relationship recorded in the relationship recording unit, the predicted value of the telephone response result during the current call is determined during the current call. Predicting and presenting,
Phone response result prediction method including   The telephone response result prediction method according to claim 11, wherein the telephone response result includes any one of or a combination of customer satisfaction regarding the call, employee satisfaction regarding the call, and call duration of the call. .   The telephone response result prediction method according to claim 11 or 12, wherein the step of calculating the sequential attribute includes a step of sequentially measuring a call voice during a call.   The step of sequentially measuring includes any one of a step of measuring a volume of the call voice, a step of detecting an utterance speed of the call voice, and a step of detecting a word by analyzing a voice morpheme of the call voice or The telephone answer result prediction method according to claim 13, comprising the combination thereof.   The sequential attribute is an attribute calculated from the volume measured in the volume measurement step, as a maximum value of volume, an average value of volume, a ratio between the maximum value and average value of volume, a maximum value of change rate of volume, The telephone response result prediction method according to claim 14, further comprising: one of the average values of sound volume change rates and a combination thereof.   The sequential attribute is an attribute calculated from the utterance speed measured in the utterance speed measurement step, as a maximum value of the utterance speed, an average value of the utterance speed, a ratio between the maximum value of the utterance speed and the average value, an utterance speed The telephone response result prediction method according to claim 14, comprising any one of a maximum value of change rate and an average value of change rate of speech rate or a combination thereof.   The telephone response result prediction method according to claim 14, wherein the sequential attribute includes a cumulative value of a specific word as an attribute calculated from the word detected by the speech morphological analysis step.   The step of calculating the sequential attribute obtains the call voice by mixing the input voice from the customer side and the input voice from the operator side, and calculates the sequential attribute from the call voice. The telephone response result prediction method according to any one of the above.   The step of calculating the sequential attribute obtains the call voice by separating the input voice from the customer side and the input voice from the operator side, and calculates the sequential attribute from the call voice. The telephone response result prediction method according to any one of the above.   The step of calculating the telephone response relationship performs a multiple regression analysis using the past sequential attribute as an explanatory variable and the past collective attribute as a dependent variable to calculate the telephone response relationship. The telephone response result prediction method according to any one of the above. A procedure for sequentially measuring the call voice in the middle of each call, and calculating the attribute sequentially for each call from the measured amount,
A procedure for calculating the telephone response result for each call as a collective attribute for each call,
The sequential attribute calculated for each call and the batch attribute calculated for each call are sequentially stored in the attribute recording unit, and the attributes are collectively collected as a past sequential attribute and a past batch attribute, respectively. A procedure for recording in the recording unit;
A procedure of calculating a telephone response relationship between the past sequential attribute and the past batch attribute for each call with reference to the past sequential attribute and the past batch attribute recorded in the attribute recording unit; ,
A procedure for recording the telephone response relationship calculated for each call in the relationship recording unit as a past telephone response relationship;
In response to the calculated sequential attribute during the current call, a predicted value of the telephone response result during the current call is predicted during the current call based on the past telephone response relationship recorded in the relationship recording unit. And the steps to present
Phone answering result prediction program for causing a computer to execute.   The telephone response result prediction program according to claim 21, wherein the telephone response result includes any one or a combination of customer satisfaction regarding the call, employee satisfaction regarding the call, and call duration of the call. .   The telephone response result prediction program according to claim 21 or 22, wherein the step of calculating the sequential attribute includes a step of sequentially measuring a call voice during a call.   The sequential measurement procedure includes any one of a procedure for measuring a volume of the call voice, a procedure for measuring an utterance speed of the call voice, and a procedure for detecting a word by analyzing a voice morpheme of the call voice or The telephone answer result prediction program according to claim 23, comprising the combination thereof.   The sequential attribute includes, as an attribute calculated from the volume measured by the procedure for measuring the volume of the call voice, a maximum value of the volume, an average value of the volume, a ratio between the maximum value and the average value of the volume, and a change in the volume. The telephone response result prediction program according to claim 24, including any one of a maximum value of rate and an average value of rate of change of volume, or a combination thereof.   The sequential attribute is an attribute calculated from an utterance speed measured by a procedure for measuring an utterance speed of the call voice, and is a maximum value of an utterance speed, an average value of an utterance speed, an average value of an utterance speed, and an average value thereof. 25. The telephone response result prediction program according to claim 24, comprising any one of a ratio, a maximum value of a rate of change in speech rate, and an average value of a rate of change in speech rate, or a combination thereof.   25. The telephone response result prediction program according to claim 24, wherein the sequential attribute includes a cumulative value of the number of specific words as an attribute calculated from a word detected by a procedure of analyzing a voice morpheme of the call voice.   28. The procedure for calculating the sequential attribute is obtained by mixing the input voice from the customer side and the input voice from the operator side to obtain the call voice, and calculating the sequential attribute from the call voice. The telephone response result prediction program according to any one of the above.   28. The procedure for calculating the sequential attribute obtains the call voice by separating an input voice from a customer side and an input voice from an operator side, and calculates the sequential attribute from the call voice. The telephone response result prediction program according to any one of the above.   30. The procedure for calculating the telephone response relationship includes performing a multiple regression analysis using the past sequential attribute as an explanatory variable and the past collective attribute as a dependent variable to calculate the telephone response relationship. The telephone response result prediction program according to any one of the above. A sequential attribute calculation unit that sequentially analyzes the call voice of the call while the customer and the operator are talking, and calculates a sequential attribute for each call;
A batch attribute calculation unit that calculates a telephone response result for each call voice as a batch attribute for each call after the call ends;
An attribute recording unit that sequentially accumulates the sequential attribute and the collective attribute at the end of a call for each call, and records them collectively as a past sequential attribute and a past collective attribute, respectively;
Analyzing a telephone response relationship between the past sequential attribute and the past collective attribute, and calculating a relationship of the telephone response for each call;
A relation recording unit that records the telephone response relation calculated for each call as a past telephone response relation;
In response to the sequential attribute during the current call calculated by the sequential attribute calculation unit, based on the past telephone response relationship recorded in the relationship recording unit, the predicted value of the telephone response result during the current call is calculated. A prediction presentation unit that sequentially predicts and presents;
A telephone response result prediction apparatus characterized by comprising: A step of sequentially analyzing the call voice of the call while the customer and the operator are calling, and calculating a sequential attribute for each call;
Calculating a telephone response result for each call voice after the call ends as a collective attribute for each call;
Sequentially storing the sequential attribute and the collective attribute at the end of the call in the attribute recording unit for each call, and recording them in the attribute record collectively as a past sequential attribute and a past collective attribute, respectively;
Analyzing a telephone response relationship between the past sequential attribute and the past batch attribute, and calculating the telephone response relationship for each call;
Recording the telephone response relationship calculated for each call in a relationship recording unit as a past telephone response relationship;
In response to the calculated sequential attribute during the current call, the predicted value of the telephone response result during the current call is sequentially predicted and presented based on the past telephone response relationship recorded in the relationship recording unit. Steps,
A method of predicting a telephone response result, comprising: A procedure for sequentially analyzing the call voice of the call while the customer and the operator are calling, and calculating the attribute for each call,
A procedure for calculating a telephone response result for each call voice after the call ends as a collective attribute for each call;
The sequential attribute and the batch attribute at the end of the call are sequentially accumulated in the attribute recording unit for each call, and recorded in the attribute record as a past sequential attribute and a past batch attribute, respectively,
Analyzing the telephone response relationship between the past sequential attribute and the past batch attribute, and calculating the telephone response relationship for each call;
A procedure for recording the telephone response relationship calculated for each call in the relationship recording unit as a past telephone response relationship;
In response to the calculated sequential attribute during the current call, the predicted value of the telephone response result during the current call is sequentially predicted and presented based on the past telephone response relationship recorded in the relationship recording unit. Procedure and
Phone answering result prediction program for causing a computer to execute.

Images