JP2005318101A - Telephone answering diagnostic device and telephone answering evaluating program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a telephone answering diagnostic device that is adaptive to a variety of expressions impartially making the wide range of creators objects to be diagnosed. <P>SOLUTION: The telephone answering diagnostic device that evaluates telephone answer of a creator who answers a customer's inquiry to diagnose the answering skill of the creator, has a text retrieving means of retrieving a text corresponding to a predetermined text for decision making from the recognized sentence of a speech of telephone answering; a grouping means of grouping texts corresponding to predetermined grouping conditions from corresponding texts; and an answer evaluating means of evaluating telephone answering by matching the pattern of grouped texts with a predetermined expression pattern. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a telephone reception diagnostic apparatus and a telephone reception evaluation program for evaluating a reception skill of a creator in a call center and diagnosing a reception skill.

At call centers that accept phone inquiries from customers, the impression of the creator who responds to the phone is directly linked to the impression of the entire company, so it is essential to regularly investigate and improve the creator's response skills. It has become. For this reason, at call centers, advanced or specialized personnel can monitor the creator's response in real time or later from the recording, and feed back the diagnosis results to the creator to improve the response skills of each creator. ing.
The items to be evaluated in the monitoring range from greetings at the beginning of reception, language usage, motivation, exchange according to the other party, response according to the contents, greeting before cutting, and so on. For this reason, the time required for monitoring one telephone is almost the same as the actual time of the telephone reception.
In addition, monitoring targets cannot be set for all telephones handled by a creator due to cost and time constraints, and are limited to a part sampled at a certain frequency for each creator (for example, non-patent literature). 1).

Japan Management Association Consulting Co., Ltd., “Call Center / Help Desk with Illustration”, 1st Edition, Japan Management Association Management Center, September 2002, p. 160-165

However, if the creator's response skills are evaluated in the form of a sampling survey, the number of samples in the sampling survey should be statistically required due to the large number of incoming calls and the lack of evaluators for the creator. Is difficult. Therefore, there is a problem that the evaluation of the corresponding creator is greatly affected by the quality of the response that is monitored.
It is also difficult for a plurality of evaluators to monitor to score each creator based on the same evaluation standard. In this way, since both the side to be evaluated and the side to be evaluated are humans, it is conceivable that the evaluation is influenced by the mutual human relationship.

  An object of the present invention is to provide a telephone answering diagnosis apparatus and a telephone answering evaluation program which can deal with various expressions fairly with a wide range of creators as diagnosis targets.

  The telephone answering diagnosis apparatus according to the present invention is a telephone answering diagnosis apparatus that evaluates a creator's telephone response to a customer's telephone inquiry and diagnoses the creator's reception skill. Text search means for searching for a text corresponding to the above, grouping means for grouping texts corresponding to a predetermined grouping condition from the corresponding text, a grouped text pattern and a predetermined expression pattern A telephone response evaluation unit that evaluates the telephone response by matching with the telephone response evaluation means.

The effect of the telephone answering diagnosis apparatus according to the present invention is that a sufficient number of samples can be secured for the sampling investigation, and all the creators can be evaluated with the same evaluation standard. By improving the evaluation accuracy in this way, it is possible to achieve a significant improvement in customer service.
Further, since the number of evaluators can be reduced, it is possible to reduce creator training costs and labor costs.

  FIG. 1 is a diagram showing the configuration of a telephone answering diagnosis apparatus according to an embodiment of the present invention. FIG. 2 is a functional block diagram of the telephone reception evaluation apparatus of the telephone reception diagnostic apparatus. FIG. 3 is a diagram illustrating a data structure of the speech recognition result table. FIG. 4 is a diagram illustrating a data structure of the determination text table. FIG. 5 is a diagram illustrating a data structure of the determination text sort table. FIG. 6 shows the data structure of the grouping condition table. FIG. 7 is a diagram illustrating the data structure of the group table. FIG. 8 shows the data structure of the expression pattern table.

The telephone answering / diagnosis apparatus 1 of the present invention is installed in a call center that responds to customer inquiries by telephone. The answering at the call center is performed while the creator operates the telephone answering system 2. The telephone reception system 2 includes a telephone reception system 3 that receives a customer's telephone call and registers a business type corresponding to the inquiry contents, and a microphone 4 that the creator uses for reception. The creator's answering voice is collected by the microphone 4 and input to the telephone answering / diagnosis apparatus 1 as an object to be diagnosed. In this way, the creator name, job type, and answering voice are input to the telephone answering diagnosis apparatus 1 for each answering.
The business type is a type of inquiry from a customer, and includes, for example, contract change, abolition, extension inconsequential, etc., and is selected and input by a button operation when a telephone is received by the telephone reception system 3 .
The telephone reception diagnostic apparatus 1 processes this input to evaluate and diagnose the reception of the creator, and outputs an evaluation result and a diagnosis result.

The telephone answering / diagnosis device 1 includes a voice recognition device 7 that converts an input answering voice into a recognized sentence by large vocabulary continuous voice recognition (dictation), a creator name and a job type input from the telephone reception system 3, and a voice recognition device. 7 has a recognition result storage device 8 for storing a speech recognition result consisting of a recognition sentence inputted from 7.
Furthermore, the telephone reception diagnostic device 1 has an evaluation frequency setting device 9 that reads a voice recognition result from the recognition result storage device 8 at a predetermined evaluation frequency and sends it to the telephone reception evaluation device 12.
Furthermore, the telephone reception diagnostic apparatus 1 includes an evaluation pattern setting apparatus 10 that stores a determination text table 22, a grouping condition table 23, and an expression pattern table 24.
Furthermore, the telephone reception diagnostic apparatus 1 includes a processing result storage device 11 that stores a determination text sort table 25 and a group table 26.

  Further, as shown in FIG. 2, the telephone answering diagnosis apparatus 1 includes a voice recognition result reading means 31 that reads a voice recognition result input from the evaluation frequency setting device 9, and a job type that is obtained by searching for a job type from the voice recognition result. The search means 32, the determination text reading means 33 that reads the determination text from the determination text table 22 based on the job type, searches the text corresponding to the determination text from the recognized sentence, creates a determination text sort, and determines the determination text sort table 25, the text search means 34 to be stored, the grouping condition reading means 35 for reading the grouping condition from the grouping condition table 23 based on the business type, and the corresponding texts are grouped based on the grouping condition from the judgment text sort, and the group table 26, the grouping means 36 stored in the Having a expression pattern reading means 37, telephoning evaluating device 12 having the answering evaluation means 38 by comparing the expression pattern and grouped text to create an evaluation result to read the expression pattern from the can expression pattern table 24.

Further, the telephone reception diagnostic apparatus 1 includes an evaluation result storage device 13 that stores an evaluation result table 27 and an evaluation diagnosis table 28.
Furthermore, the telephone reception diagnostic apparatus 1 has a reception diagnostic apparatus 14 for diagnosing a creator's reception skill using a diagnostic rule with reference to the evaluation result, and a diagnostic rule storage apparatus 15 for storing the diagnostic rule. The diagnosis rule is made up of diagnosis threshold values and is stored as a diagnosis threshold value table 29.
Furthermore, the telephone reception diagnostic apparatus 1 includes a diagnostic result display output device 16 that displays a diagnostic result or prints the diagnostic result.

Next, each apparatus which comprises the telephone reception diagnostic apparatus 1 is demonstrated.
The speech recognition device 7 converts the response speech into a recognition sentence that is a text sentence so that the vocabulary can be searched by dividing each response by dictation.
The recognition result storage device 8 stores a speech recognition result table 21 for each index for which a speech recognition result including a creator name, a job type, and a recognized sentence is received. For example, as shown in the example in Fig. 3, the creator name is "Toshiaki Ishii", the business type is "Change Contract", and the recognition sentence is "Thank you for waiting. The price is 780 yen for 30 amps ... ".
The evaluation frequency setting device 9 has a preset frequency for extracting and evaluating the voice recognition result for each creator or for each business type, and extracts the voice recognition result from the recognition result storage device 8 according to the frequency and evaluates the telephone response evaluation device. 12 to send.
Moreover, the evaluation frequency setting device 9 can change the frequency of evaluation based on the diagnosis result of the response diagnosis device 14. For example, when the result of diagnosis of the response skill of a certain creator is poor and it is necessary to evaluate the response of the creator more frequently, the frequency of extracting the voice recognition result is increased. Conversely, it is possible to reduce the frequency of extracting creators with excellent response skills.

The determination text is composed of text necessary for reception or text that is not preferable to use. The texts are grouped for each element according to the strength of the relation, further grouped for each determination item, and grouped for each business type, and stored as a determination text table 22 as shown in FIG.
For example, determination items for contract change of business type include [greeting], [basic charge], [attendance request], [power outage PR], [prohibition], [mouth]. “Greeting”, “Company name”, “Department name”, and “Nori” are listed as elements related to “greeting” in the judgment items. In addition, “Good morning”, “Thank you”, “Omakase” are listed as the “greeting” text of the element, “Tokyo Electric Power” as the text of the “company name” of the element, and “Customer Center” as the text of the “Department name” of the element. ",""Nori","There","Toshi","Is", and "I can't accept".
In addition, as an element of the “basic charge” of the determination item, “basic charge”, “ampere 1”, “ampere 2”, “charge 1”, and “charge 2” are listed. Text is set in these elements. For example, the text “Amp1” is composed of “10 amps”, “15 amps”, “20 amps”, “30 amps”,..., “60 amps”.
Similarly, elements and texts are also mentioned for the judgment items [attendance request] and [power failure PR].
In addition, “mouth” is cited as an element related to the “mouth” of the judgment item, and “no”, “ah”, “e”, “so”, “no”, “no” as the text of the element “mouth” “Shape” and “Dayo” are mentioned. Similarly, elements and texts are listed for the “prohibited phrase” judgment item.

  Judgment text sort is a text that matches the text for judgment, searches from the beginning of the recognized sentence, extracts the element to which the text belongs, the number of characters from the top and the number of characters of the text each time it matches the text for judgment. As shown in FIG.

The grouping condition consists of three types of conditions. The first is a condition for determining whether two or more texts are related. Even if two texts appear in the response, whether or not the two texts appear in relation to each other depends on the position of the recognized sentence where the two texts appear. For example, the beginning and end of the response are not considered related text.
The second is a condition for determining whether or not to consider the text as a group depending on whether or not the text appears in the recognized sentence. For example, when the text regarding the greeting appears at the beginning or end of the reception, it is determined that the text is worth evaluating.
The third is a condition for determining whether or not to consider a group based on the number of appearances of the text. For example, what is regarded as a mustache is a determination condition that is regarded as a group when a text that matches the determination text appears a predetermined number of times or more in a recognized sentence.
This grouping condition is predetermined for each business type and each determination item, and is stored as a grouping condition table 23 as shown in FIG. In FIG. 6, an element name surrounded by “” means a group of text included in the element.

Text that satisfies the grouping condition is considered to be a predetermined group. Note that AND represents a logical product, and OR represents a logical sum.
For example, in the grouping condition related to [basic charge] of the determination item, text existing within 20 characters is regarded as a group. From the judgment text sort table of FIG. 6 (40 amps) and no. 7 (50 amps), NO. 9 (basic charge) and No. 10 (60 amps) and no. 11 (1560 yen), no. 23 (30) and no. 24 (40) can be regarded as a group. Each of them has a group number. 5, no. 6, no. Number 7

Further, there are four grouping conditions regarding the “greeting” determination item.
As the first grouping condition, “greetings” within 25 characters from the beginning of the recognized sentence are set as a1 group. From the judgment text sort table of FIG. 1 (Please wait) can be regarded as a1 group.
The second grouping condition is that “greeting” and “company name” within 20 characters thereafter are grouped as a2. From the judgment text sort table of FIG. 1 (Please wait) and No. 2 (TEPCO) can be regarded as the a2 group.
As a third grouping condition, “company name” and “department name” within 20 characters thereafter are grouped as a3. From the judgment text sort table of FIG. 2 (TEPCO) and no. 3 (customer center) can be regarded as a3 group.
As a fourth grouping condition, “group name”, a character string of one or more characters within 25 characters thereafter, and “name” are grouped as a4. From the judgment text sort table of FIG. 3 (customer center) and no. Since the character string “Ishii” is included between 4 and there, it can be regarded as a4 group.

In addition, there are three grouping conditions for the determination item “mouth”.
The first grouping condition is set to the group of m1 when any one of the text groups of the “mouth” of the element does not appear in the recognition sentence.
The second grouping condition is set to the group m2 when any one of the text groups of the word “mouth” in the recognition sentence appears from 1 to 5 times.
The third grouping condition is set to the group m3 when any of the text groups of the element “mouth” appears in the recognition sentence six times or more. From the judgment text sort table of FIG. 5, since “mouth” appears six times, it can be regarded as the m3 group.
The grouping result thus performed is stored as a group table 26 as shown in FIG. 7 in association with the group condition, group number, and text sort number.
Similarly, there are also three sectioning conditions regarding the “prohibited phrase” of the determination item.

If there is a group that matches the expression pattern among the grouped groups, it is determined that the determination item includes contents worthy of evaluation. The expression pattern is set for each determination item, and is stored as an expression pattern table 24 as shown in FIG.
For example, four patterns are set for the expression pattern of the “basic charge” of the determination item. When one of the four patterns is matched, it is determined that the creator is correctly responding to the basic fee.
The expression pattern of β1 is {(ampere 1) OR (ampere 2)} AND {(charge 1) OR (charge 2)}. For example, when “30 amperes is 780 yen” is received, it matches the expression pattern of β1.
The expression pattern of β2 is {basic charge} AND {(charge 1) OR (charge 2)}. For example, when “basic charge is changed from 520 yen to 780 yen”, it matches the expression pattern of β2.
The expression pattern of β3 is {basic charge} AND {(ampere 1) OR (ampere 2)} AND {(charge 1) OR (charge 2)}. For example, when responding, “At 30 amps, the basic charge will be 780 yen”, it matches the expression pattern of β3.
As can be seen from the judgment item column of the group table in FIG. 6 matches the expression pattern of β2.

In addition, one expression pattern is set for the determination item [attendance request].
It is “Witness” AND “Request”. For example, it matches when you answer “I'm asking you to be there.”
In addition, six patterns are set as the expression pattern of the determination item [power failure PR]. Then, when it matches any one of the six patterns, the creator determines that the power failure has been accurately performed.
The expression pattern of δ1 is (“construction”) AND (“power failure 1” OR “power failure 2”). For example, when responding “The breaker replacement work involves a power failure”, it matches the expression pattern of δ1.
The expression pattern of δ2 is (“construction”) AND (“power failure 3” AND “power failure 4”). For example, when responding to “breaker replacement stops electricity,” it matches the expression pattern of δ2.
The expression pattern of δ3 is (“time”) AND (“power failure 1” OR “power failure 2”). For example, when responding to “There will be a power outage for about 10 minutes”, it matches the expression pattern of δ3.
The expression pattern of δ4 is (“time”) AND (“power failure 3” AND “power failure 4”). For example, when responding “Because electricity stops for about 10 minutes”, it matches the expression pattern of δ4.
The expression pattern of δ5 is (“construction” AND “time”) AND (“power failure 1” OR “power failure 2”). For example, when responding to “I will have a power outage for about 15 minutes during replacement”, it matches the expression pattern of δ5.
The expression pattern of δ6 is (“construction” AND “time”) AND (“power failure 3” AND “power failure 4”). For example, when responding “Because electricity stops during work”, it matches the expression pattern of δ6.

In addition, when the expression pattern of [Greeting] of the judgment item is set as one pattern and a group that matches this pattern is included in the recognition sentence, the greeting according to the creator's situation is performed. judge. This expression pattern is an α pattern in which all of the groups a1, a2, a3, and a4 are expressed.
In addition, three patterns are set as the expression pattern of [prohibition phrase] of the determination item, and the evaluation is different for each pattern. The ξ1 pattern corresponding to the group k1, the ξ2 pattern corresponding to the group k2, and the ξ3 pattern corresponding to the group k3.
In addition, three patterns are set for the expression pattern of the determination item “mouth”, and the evaluation differs for each pattern. A ζ1 pattern corresponding to the group m1, a ζ2 pattern corresponding to the group m2, and a ζ3 pattern corresponding to the group m3.

Next, response evaluation will be described.
The response evaluation is performed by performing a pattern comparison process on the grouped text and the expression pattern.
For example, as shown in the group table of FIG. There are three of 5-7. When this group is subjected to pattern comparison processing with the expression patterns β1 to β4 shown in FIG. The group of 6 matches the expression pattern β3.
In addition, regarding the “greeting” determination item, the group No. 1-No. 4 and the expression pattern α are subjected to pattern comparison processing. As a result, group No. 1 to 4 and the expression pattern α are matched.
In addition, regarding the [attendance application] judgment item, the group No. 8 and the expression pattern γ are subjected to pattern comparison processing. As a result, group No. 8 matches the expression pattern γ.
Similarly, the determination items [power failure PR], [prohibition], and [mouthpiece] are also subjected to pattern comparison processing to determine the presence or absence of matching.
The evaluation result thus determined is stored in the evaluation result storage device 13 as an evaluation result table 27 as shown in FIG. In FIG. 9, “◯” indicates that it has been determined that the determination item is accurately handled, and “x” indicates that it has been determined that the determination item has not been correctly handled.

The telephone reception evaluation device 12 is configured by a computer having a CPU, a RAM, a ROM, and an interface circuit.
The ROM stores a program consisting of steps read by the CPU and processed by the CPU accordingly.
The program consists of the following steps: FIG. 10 is a flowchart of a program for controlling arithmetic processing in the CPU of the telephone answering / evaluating apparatus 12.
Step 101 reads the speech recognition result of the response to be evaluated extracted by the evaluation frequency setting device 9 from the speech recognition result table 21.
In step 102, a job type is searched and obtained from the read voice recognition result.
In step 103, the determination text is read from the determination text table 22 based on the retrieved business type.
Step 104 sorts the corresponding text from the recognized sentence of the voice recognition result using the determination text.
A step 105 reads the grouping condition from the grouping condition table 23 based on the business type.
Step 106 groups the text that satisfies the grouping condition from the sorted text.
Step 107 reads an expression pattern from the expression pattern table 24 based on the business type.
In step 108, the group and the expression pattern are subjected to pattern comparison processing to evaluate the response.
Step 109 summarizes the evaluation results of each reception.

  The evaluation result storage device 13 stores the evaluation results of each response collected by the telephone response evaluation device 12 as an evaluation result table 24.

Next, the response diagnosis performed based on the response evaluation result will be described.
FIG. 11 is a diagram illustrating a data structure of the threshold value table for diagnosis. FIG. 12 is a diagram illustrating a data structure of the evaluation diagnosis table.
In the diagnostic rule storage device 15, the diagnostic threshold values are stored together as a diagnostic threshold value table 29 as shown in FIG. The threshold value for diagnosis is a “bad” threshold value as a second reference value that is judged to be bad when the execution rate is lower than this, and a first reference value that is judged to be good when the execution rate is higher than this. A “good” threshold is set for each judgment item. When the implementation rate exceeds the “bad” threshold and is less than the “good” threshold, it is judged as the standard.
The response diagnosis device 14 reads the evaluation result from the evaluation result table 27 and obtains the execution rate for each creator, business type, and determination item, as shown in FIG. The implementation rate is obtained with the total response number for each corresponding business type as a parameter, and the response number determined to correspond accurately as a child number. The implementation rate can take a value between 0% and 100%.
Next, the diagnostic threshold value is read from the diagnostic rule storage device 15. Each execution rate is diagnosed as “good”, “standard”, and “bad” based on the diagnostic threshold. As shown in FIG. 12, the diagnosis results are collected as an evaluation diagnosis table 28 and stored in the evaluation result storage device 13.
In addition, when the diagnosis result is “bad”, the reception diagnosis device 14 changes the evaluation frequency of the creator and transmits the evaluation frequency to the evaluation frequency setting device 9.

Next, the output of the diagnosis result will be described. FIG. 13 is a diagram illustrating a state in which the result of the response diagnosis is displayed.
The evaluation result display output device 16 reads the diagnosis result from the evaluation diagnosis table 28, displays the response diagnosis result for each creator on the display as shown in FIG. 13, and if necessary, prints it as a hard copy from the printer.

Since such telephone answering / diagnostic devices store the answering voices of all creators as recognition sentences, if it is necessary to evaluate all of them, it is possible to perform all the evaluations.
In addition, since the expression pattern is preset for each business type and the recognition sentence is evaluated based on the preset expression pattern, it is possible to evaluate based on the same evaluation standard.
In addition, since the corresponding grouped text and expression pattern are evaluated by pattern comparison processing, it is possible to flexibly evaluate the response of creators performed in various expression patterns.

In this way, the number of samples to be evaluated is statistically sufficiently secured, and the evaluation is performed fairly based on the same evaluation standard, so that the evaluation accuracy is improved and the response skill of the creator is greatly improved.
Further, by applying such a telephone answering / diagnosis device to answering diagnosis, the number of evaluators can be greatly reduced, so that it is possible to reduce training costs and labor costs.
In addition, since the evaluator who floated can focus on developing creators with low response skills, the customer satisfaction of the entire call center can be raised.

  Note that the business type, determination item, element, and text group taken up in the above-described embodiment are examples for explanation, and may be set so as to match the business condition of the call center.

It is a figure which shows the structure of the telephone reception diagnostic apparatus concerning embodiment of this invention. It is a functional block diagram of the telephone reception evaluation apparatus of the telephone reception diagnostic apparatus of FIG. It is a figure which shows the data structure of the speech recognition result table stored in the recognition result storage apparatus of FIG. It is a figure which shows the data structure of the text table for determination stored in the evaluation pattern setting apparatus of FIG. It is a figure which shows the data structure of the determination text sort table stored in the processing result storage apparatus of FIG. It is a figure which shows the data structure of the grouping condition table stored in the evaluation pattern setting apparatus of FIG. It is a figure which shows the data structure of the group table stored in the processing result storage apparatus of FIG. It is a figure which shows the data structure of the expression pattern table stored in the evaluation pattern setting apparatus of FIG. It is a figure which shows the data structure of the evaluation result table stored in the evaluation result storage apparatus of FIG. It is a flowchart of the program which controls the arithmetic processing performed in CPU of a telephone reception evaluation apparatus. It is a figure which shows the data structure of the threshold value table for diagnosis stored in the diagnostic rule storage apparatus of FIG. It is a figure which shows the data structure of the evaluation diagnostic table stored in the evaluation result storage apparatus of FIG. It is a figure which shows a mode that the result of a reception diagnosis is displayed.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 Telephone reception diagnostic apparatus, 2 telephone reception system, 3 telephone reception system, 4 microphone, 7 voice recognition apparatus, 8 recognition result storage apparatus, 9 evaluation frequency setting apparatus, 10 evaluation pattern setting apparatus, 11 processing result storage apparatus, 12 telephone Response evaluation device, 13 Evaluation result storage device, 14 Response diagnosis device, 15 Diagnosis rule storage device, 16 Evaluation result display output device, 21 Speech recognition result table, 22 Judgment text table, 23 Grouping condition table, 24 Expression pattern table , 25 determination text sort table, 26 group table, 27 evaluation result table, 28 evaluation diagnosis table, 29 diagnosis threshold value table, 31 speech recognition result reading means, 32 business type search means, 33 determination text reading means, 34 text search means, 35 guru Flop conditions reading unit, 36 grouping means 37 expression pattern reading unit, 38 answering evaluation unit.

Claims (7)

In a telephone answering diagnosis device that evaluates a creator's telephone response to a telephone inquiry from a customer and diagnoses the creator's response skill,
A text search means for searching for a text matching a predetermined text for determination from a voice recognition sentence of a telephone reception;
Grouping means for grouping texts that meet a predetermined grouping condition from the matching texts as a group;
A response evaluation means for evaluating a telephone response based on the presence or absence of matching between the grouped text pattern and a predetermined expression pattern;
A telephone answering / diagnosis apparatus comprising a telephone answering / evaluating apparatus. A recognition result storage device in which the speech recognition text of the telephone reception is stored separately for each type of business of telephone reception;
The telephone answering / diagnosis apparatus according to claim 1, wherein the determination text, the grouping condition, and the expression pattern are predetermined for each business type.   3. The telephone answering / diagnosis apparatus according to claim 1, wherein the determination text is at least one of text essential for telephone answering, text unsuitable for telephone answering, and text of a mustache.   The grouping condition is that text included in a part of the recognition sentence having a predetermined length among two or more texts that match the determination text, and the text at a predetermined position among texts that match the determination text. It is a condition that at least one of texts that appear in a predetermined number of times over the entire sentence of the recognized sentence among the text included in the recognized sentence part or the text that matches the determination text is a condition. The telephone reception diagnostic apparatus as described in any one of Claims 1 thru | or 3.   A response diagnosis device that determines that the response skill is good when the percentage of goodness among the plurality of evaluations of the telephone response is equal to or higher than a predetermined first reference value, and that the response skill is poor when it is equal to or lower than a second reference value The telephone answering / diagnosis device according to claim 1, characterized in that   6. The apparatus according to claim 1, further comprising an evaluation frequency setting device that extracts a recognition sentence at a predetermined frequency from a plurality of speech recognition sentences of the telephone reception and supplies the extracted sentence to the telephone reception evaluation apparatus. Telephone response diagnostic device to be described. Reading a speech recognition result of a response to be evaluated from a speech recognition result table, and searching for a text that matches a predetermined determination text from a recognition sentence of the speech recognition result;
A step of grouping texts that satisfy a predetermined grouping condition from the matching texts as a group;
Evaluating the quality of the telephone response based on the presence or absence of the grouped text pattern matching a predetermined expression pattern;
A telephone answering evaluation program for causing a telephone answering evaluation apparatus to execute a telephone answering evaluation.

Images