JP2009267469A - Automation method of call center and call center system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a call center system which changes content of a response to a customer according to customer data. <P>SOLUTION: A call center 10 performs a questionnaire survey by interpreting and executing a questionnaire survey execution file 3 described using a PHP (Hypertext Preprocessor). A questionnaire survey execution file 3 describes codes to read customer data from a customer table 6 prepared in an SQL (Structured Query Language) server 17, and dynamically generate a Voice XML (Extensible Markup Language) document used for the questionnaire survey according to the customer data. A call center server 10 stores answer data obtained by the questionnaire survey in an answer data file 5. An analysis server 12 analyses a structure of the answer data file 5 using a schema 2, and collects and analyzes the answer data by a procedure based on a template 7. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a technical field of a call center that accumulates answer data such as questionnaire surveys by making a voice call with a customer.

  In order to build CRM (CRM: Customer Relationship Management), companies often use call centers when conducting customer satisfaction surveys on products and campaigns for new products. Conventionally, call centers have a form that uses an automatic voice response device (IVR: Interactive Voice Response) and a form in which an operator responds to a customer. In either form, the call center has response data obtained from the customer. Is stored, and the stored customer response data is analyzed.

  A call center outsourcing company handles a large number of call center operations, so it is desirable to be able to automate the process of obtaining response data from users and analyzing the response data. However, there are various issues in automating call center operations. is there.

  The first challenge when automating call center operations is the automation of answer data acquisition. In a call center in which the operator corresponds to the customer, the operator interacts with the customer according to a predetermined scenario, and the operator determines the answer data from the customer's answer, so that the answer data input operation becomes manual.

  In order to eliminate the task of inputting the answer by the operator, it may be possible to record the voice of the customer's answer as in the invention disclosed in Patent Document 1, but it is recorded when analyzing the effect of the call center. Processing of the customer's voice is required.

  On the other hand, in a call center using an automatic voice response device, as disclosed in Patent Document 2, the automatic voice response device (advertisement CTI server) performs voice guidance according to the advertisement identification code received from the telephone (customer terminal). Is transmitted to the customer terminal, and numerical values and symbols transmitted from the customer terminal are automatically stored in the memory as response data using DTMF (Dual-Tone Multi-Frequency).

  However, in the call center using the automatic voice response device, the customer answers the voice guidance transmitted from the server by using DTMF, so that it is not possible to realize a call center in a form of interacting with the customer.

  In particular, CRM emphasizes the connection of existing customers rather than the acquisition of new customers. Therefore, in the call center related to CRM, the purchase history of customers who have voice conversations in the form of voice conversations with existing customers. There is a request from the company to perform one-to-one call center operations according to customer data, but these inventions automatically implement call center operations according to customer data such as the purchase history of customers who have voice conversations. Is not listed.

The second challenge when automating call center operations is the automation of work to analyze response data. As an invention for automating the analysis of the answer data obtained in the call center business, a questionnaire analysis support system for automating questionnaire counting as disclosed in Patent Document 3 can be used. Is an invention that focuses on the method for displaying the results of questionnaire counting, and does not describe specific contents for automating the counting of questionnaires.
JP 2002-44260 A JP 2007-174354 A JP 2003-288461 A

  Therefore, in order to solve the various problems described above, the present invention can dynamically change the contents of the customer response according to the customer data of the customer who has a voice conversation in a call center that has a voice conversation with the customer. It is another object of the present invention to provide a call center automation method and a call center system capable of automating the analysis of answer data obtained from customers at a call center.

  A first invention for solving the above-described problem is a method for automating call center operations, wherein a server (call center server) installed in a call center is based on a specified table in a database connected to the call center server. The call center server acquires the reference data used in the call center business, dynamically generates the VoiceXML document used in the call center business using the reference data, and executes the generated VoiceXML document. Step a for transmitting a question message described in the VoiceXML document to the voice terminal device by voice, recognizing the voice received from the voice terminal device to obtain answer data, and writing the answer data in the script language Define the structure of the answer data file In order to analyze the answer data obtained in the call center process, and a call center process including the step b of storing the answer data obtained from the voice terminal device in the answer data file in a format according to the schema A server (analysis server) to be installed includes an analysis step of analyzing the structure of the answer data file by referring to the schema and totaling / analyzing the answer data included in the specified answer data file; This is an automated method for call center operations.

  Further, the second invention is the call center business automation method according to the first invention, wherein in the step a of the call center process, the call center server describes a code for dynamically generating the VoiceXML document. The reference data is obtained and the VoiceXML document is dynamically generated by interpreting and executing a program written in a predetermined programming language. This is a method for automating call center operations.

  Furthermore, the third invention is the call center business automation method according to the second invention, wherein the database is provided with a customer table storing customer data as a target of the call center business as the table. In the program interpreted and executed by the call center server in the step a, a code for obtaining at least the customer data associated with the voice terminal device for voice communication from the customer table of the database as the reference data. A call center business automation method characterized by being described.

  Further, a fourth invention is the call center business automation method according to the third invention, wherein in the step a of the call center process, the program executed by the call center server in the step a is referred to as the reference. The call center business automation method is characterized in that a code for using the customer data acquired as data as the answer data is described.

  Furthermore, a fifth invention is the call center business automation method according to any one of the third invention and the fourth invention, wherein the call center server is the step a in the step a of the call center process. The program to be executed is a call center business automation method characterized in that a code for changing the question message included in the VoiceXML document is described in accordance with the customer data acquired as the reference data.

  Further, a sixth invention is the call center business automation method according to any one of the third invention to the fifth invention, wherein in the step a of the call center process, the call center server in the step a. In the program executed by, a code that is included in the VoiceXML document and changes an attribute value related to the voice of the question message according to the customer data acquired as the reference data is described. This is an automated method for call center operations.

  Furthermore, a seventh invention is the call center business automation method according to any one of the first to sixth inventions, wherein the database uses the voice used as the table in the VoiceXML document. A speech recognition dictionary table storing dictionary data used in a recognition dictionary is provided. In the step a of the call center process, the program executed by the call center server in the step a includes the reference data as the reference data, Code for dynamically generating the speech recognition dictionary used when acquiring the dictionary data from a speech recognition dictionary table and obtaining the answer data from speech transmitted from the speech terminal device is described. This is an automated method for call center operations.

  Furthermore, an eighth invention is the call center automation method according to any one of the first to seventh inventions, wherein the answer data included in the answer data file is aggregated and analyzed. A template in which a procedure is described is prepared, and the analysis server aggregates and analyzes the response data included in the specified response data file according to the process indicated by the specified template. It is a business automation method.

   Furthermore, the ninth invention is a call center system comprising a call center server that automatically executes a call center service and an analysis server that automatically analyzes a result of the call center service, wherein the call center server includes the call center server and the call center server. VoiceXML document generation means for acquiring reference data used in a call center service from a specified table in a connected database, and dynamically generating a VoiceXML document used in the call center service using the reference data; By interpreting and executing the generated VoiceXML document, a question message described in the VoiceXML document is transmitted to the voice terminal device by voice, and voice data received from the voice terminal device is recognized to obtain answer data. Fruit And an answer stored in the answer data file in a format according to a schema that defines a structure of an answer data file that is described in a script language and that stores the answer data. Data storage means, and the analysis server analyzes the structure of the answer data file by referring to the schema, and aggregates and analyzes the answer data included in the specified answer data file It is a call center system characterized by comprising.

  Further, a tenth invention is the call center system according to the ninth invention, wherein the VoiceXML document generating means of the call center server describes a code for dynamically generating the VoiceXML document, and the reference data is The reference data is obtained from the database and the VoiceXML document is dynamically generated by interpreting and executing a program described in a predetermined programming language having variables to be substituted. It is a call center system.

  Furthermore, the eleventh invention is the call center system according to the tenth invention, wherein the database is provided with the customer table in which the customer data to be subjected to call center operations is stored as the table, The program executed by the VoiceXML document generation unit of the call center server describes, as the reference data, a code for obtaining at least customer data associated with the voice terminal device for voice communication from the customer table of the database. It is a call center system characterized by that.

  Furthermore, a twelfth invention is the call center system according to the eleventh invention, wherein the program executed by the VoiceXML document means of the call center server includes the customer data acquired as the reference data as the answer data. The call center system is characterized in that a code to be used is described.

  Further, a thirteenth invention is the call center system according to the eleventh invention or the twelfth invention, wherein the program executed by the VoiceXML document generating means of the call center server includes the reference data acquired as the reference data. The call center system is characterized in that a code for changing the question message included in the VoiceXML document is described according to customer data.

  The fourteenth invention is the call center system according to any one of the eleventh to thirteenth inventions, wherein the program executed by the VoiceXML document means of the call center server includes the reference data. In accordance with the customer data acquired as described above, a code that is included in the VoiceXML document and changes an attribute value related to the voice of the question message is described.

  Furthermore, the fifteenth invention is the call center system according to any one of the ninth to fourteenth inventions, wherein the database includes a speech recognition dictionary used in the VoiceXML document as the table. A speech recognition dictionary table storing dictionary data to be used is provided, and the dictionary data is acquired from the speech recognition dictionary table as the reference data in the program executed by the VoiceXML document generation unit of the call center server. In the call center system, a code for dynamically generating the voice recognition dictionary used when obtaining the answer data from the voice transmitted from the voice terminal device is described.

  Furthermore, the sixteenth invention is the call center system according to any one of the ninth to fifteenth inventions, wherein a processing procedure for counting and analyzing the answer data included in the answer data file is provided. In the call center system, the described template is prepared, and the analysis server aggregates and analyzes the response data included in the specified response data file according to the process indicated by the specified template. is there.

  According to the first invention described above, by using the VoiceXML document, it is possible to automate call center operations such as questionnaire surveys while voice conversation between the customer using the voice terminal device and the call center server is possible. . Alternatively, if the structure of the answer data file storing the answer data obtained from the customer is defined in advance in the schema, the structure of the answer data file is common regardless of the call center business to be performed. In order to tabulate the answer data files, it is not necessary to prepare a dedicated program for each answer data file.

  In addition, according to the first aspect of the present invention, the VoiceXML document used in the call center service is dynamically generated using the reference data stored in the table provided in the database connected to the call center server. In this way, it is possible to change the contents to be implemented in the call center business.

  As a technique for dynamically generating the VoiceXML document used in the call center business, as in the second invention, a code for dynamically generating the VoiceXML document is described, and a variable to which the reference data is substituted is described. It is preferable to use the program described in a predetermined programming language.

  If the customer data associated with the voice terminal device that performs voice communication with the call center server is acquired from the database as the reference data as in the third aspect of the invention, the contents of the response to the customer according to the customer data Can be dynamically changed, and OnetoOne call center operations can be realized.

  For example, as in the fourth aspect of the invention, if a code that uses the customer data obtained from the database as the answer data is described in the program that is interpreted and executed by the call center server, the customer data can be used in call center operations. Since it is not necessary to execute a transaction to acquire the call center, it is possible to reduce the time for call center operations for one customer.

  Further, as in a fifth aspect of the invention, in the program, a code for changing an attribute related to the voice of the question message is written according to the customer data obtained from the database, so that the voice can be heard according to the customer. Will be able to change. Here, the attribute relating to the voice of the question message means the speed and volume of the voice.

  Further, as in the sixth aspect of the invention, in the program, a code for changing the question message transmitted by voice to the voice terminal device according to the customer data obtained from the database is described. Accordingly, the question message can be changed.

  In addition, as in the seventh aspect of the invention, if the dictionary data that is the contents of the voice recognition dictionary used in the VoiceXML document is acquired from the database as the reference data, the voice recognition dictionary used in the VoiceXML document Can be dynamically generated, and it is not necessary to prepare the speech recognition dictionary in advance.

  In addition, according to the eighth aspect of the present invention, if the template in which the processing procedure for counting and analyzing the response data included in the response data file is prepared is prepared, the analysis server is configured to use the template. It is necessary to prepare the dedicated program for each answer data file to analyze the answer data file so that the answer data can be automatically aggregated and analyzed. Disappear.

  According to the ninth to sixteenth aspects, the same effects as those of the first to eighth aspects described above can be obtained.

  As described above, according to the present invention described above, in the call center that interacts with the customer, the contents of the response to the customer can be dynamically changed according to the customer data of the customer, and further obtained from the customer at the call center. It is possible to provide a call center automation method and a call center system capable of automating the analysis of response data.

  From here, the call center system and call center automation method according to the present invention will be described in detail with reference to the drawings. FIG. 1 is a diagram illustrating the configuration of the call center system 1.

  The call center system 1 illustrated in FIG. 1 is a system that conducts a questionnaire survey related to CRM (CRM: Customer Relationship Management) as a call center service. The call center system 1 includes a voice terminal device 13 used by a customer and a VoiceXML document. A call center server 10 having a function of dynamically generating and automatically executing a call center operation using a VoiceXML document, and a DWH server 11 (DWH: DataWareHouser) in which answer data obtained from a customer by the call center server 10 is stored And an analysis server 12 that aggregates and analyzes response data obtained from the customer and outputs a report describing the questionnaire survey results, and SQL in which customer data such as customer attributes (for example, gender and age) is stored Server 17 and VoIP gateway server 1 At least composed of a voice terminal device 13 and the call center server 10 the customer operates, via the VoIP gateway server 16, PSTN 14: is connected (PSTN Public Switched Telephone Networks).

  FIG. 2 is a flowchart showing the procedure of the call center business automation method executed by the call center system 1 shown in FIG. 1. In the call center system 1, the call center server 10 sends the contents of the questionnaire survey to the voice terminal device 13. A call center step (S1) for transmitting and storing the answer data obtained from the voice terminal device 13 in the DWH server 11, and an analysis step (S2) in which the analysis server 12 aggregates and analyzes the answer data obtained in the questionnaire survey. Is executed.

  The device that operates in the call center process (S1) is equipped with a voice terminal device 13 operated by the customer and a function for dynamically generating a VoiceXML document for use in a questionnaire survey, and a questionnaire survey is automatically performed using the generated VoiceXML document. A call center server 10, an SQL server 17 that is a database server in which customer data used in a questionnaire survey is stored, and a DWH server 11 in which a questionnaire survey result is stored.

  The SQL server 17 is provided with a customer table 6. The customer table 6 stores customer data used when DM (DM: Direct Mail) is sent to the customer. The call center server 10 performs the questionnaire survey execution file 3 in which the contents of the questionnaire survey are described, the schema 2 that defines the structure of the answer data file 5 that stores the answer data obtained from the voice terminal device 13, and the questionnaire survey. A telephone number list 4 in which customer telephone numbers are listed is stored.

  The questionnaire survey execution file 3 stored in the call center server 10 uses PHP (Hypertext Preprocessor), which is one of programming languages, based on the contents of customer data to be surveyed. Code for dynamically generating a VoiceXML document to be executed is described.

  The call center server 10 dynamically generates a VoiceXML document by interpreting and executing the PHP code described in the questionnaire survey execution file 3, and uses the voice terminal device 13 by executing the generated VoiceXML document. Conduct questionnaire surveys automatically with voice dialogue with customers. The programming language describing the questionnaire survey execution file 3 may be a programming language other than PHP, such as Java (registered trademark) or C language.

  When the call center server 10 starts executing the questionnaire survey, the answer data file 5 for storing the questionnaire survey result is provided in the DWH server 11 and the dynamically generated VoiceXML document is used to execute the questionnaire survey for one customer. Each time, the answer data obtained from the voice terminal device 13 operated by the customer is stored in the answer data file 5 of the DWH server 11.

  The answer data file 5 stored in the DWH server 11 is described in XML (eXtensible Markup Language), which is a script language, and the structure of the answer data file 5 is common regardless of the questionnaire survey to be executed. , Defined by the schema 2 stored in the call center server 10. This schema 2 is stored in the call center server 10 and the analysis server 12.

  The call center server 10 can automate interactive call center operations by using VoiceXML for the questionnaire survey. In addition, when changing the contents of the questionnaire survey, the call center server 10 uses the questionnaire survey execution file 3 described using PHP. Since it is sufficient to rewrite the contents, the work of changing the contents of the questionnaire survey can be reduced.

  If the PHP code for acquiring the customer data stored in the customer table 6 provided in the SQL server 17 is described in the questionnaire survey execution file 3, the customer data obtained from the SQL server 17 is used as the response data. The VoiceXML document can be dynamically changed according to the customer data obtained from the SQL server 17 or used.

  The apparatus that operates in the analysis step (S2) is the analysis server 12. When the analysis server 12 receives an instruction to totalize / analyze the answer data file 5 obtained by the questionnaire, the answer data is referred to the schema 2 Comprehend the structure of the file 5 and automatically calculate and analyze the response data according to the template 7 that describes the processing procedure for counting and analyzing the response data contained in the response data file 5, and determine the results of the questionnaire survey The report 15 is output in the format (e-mail or print).

  Since the structure of the answer data file 5 for storing the answer data is determined in advance, the structure of the answer data file 5 is the same even if the order of the question items executed in the questionnaire survey is different. The server 12 can automatically interpret the structure of the answer data file 5 by referring to the schema 2.

  If the template 7 in which the procedure for counting and analyzing the answer data is described is prepared, the analysis server 12 automatically calculates the answer data by referring to the template 7. Can be implemented.

  From here, the voice terminal device 13, the call center server 10, the SQL server 17, and the DWH server 11 which constitute the call center system 1 illustrated in FIG. 1 will be described in detail. FIG. 3 is a block diagram of the call center system 1 shown in FIG.

  As shown in FIG. 3, the voice terminal device 13 used by the customer is a terminal provided with a voice interface 130 for voice conversation between the call center server 10 and the customer. In FIG. 1, the voice terminal device used by the customer. 13 is illustrated as a telephone, but the voice terminal device 13 used by the customer may be a personal computer in which a voice browser is installed as the voice interface 130.

  As shown in FIG. 3, the SQL server 17 is provided with a customer table 6 which is a table for storing customer data. FIG. 4 is a diagram for explaining the customer table 6 provided in the SQL server 17. As shown in FIG. 4, the customer table 6 provided in the SQL server 17 is composed of records and columns, and column names for identifying the columns are given to the respective columns.

  In FIG. 4, the customer table 6 stores, as columns, a column “code” for storing a customer code, which is a serial number assigned to the customer, a column “telephone” for storing the customer's telephone number, and the gender of the customer. The column “gender”, the column “age” for storing the age of the customer, and the column “purchase” for storing the purchase history of the customer are provided at least.

  The DWH server 11 is provided with a DWH, and this DWH stores an answer data file 5 in which answer data obtained in the questionnaire survey is stored. The structure of the answer data file 5 is stored in the call center server 10. Defined by schema 2. The answer data file 5 and the schema 2 will be described later.

  From here, the call center server 10 will be described. The call center server 10 is realized by mounting a program for functioning as the call center server 10 according to the present invention on a general-purpose Web server including a CPU, a RAM, and a large-capacity data storage device.

  As shown in FIG. 3, the call center server 10 reads out a natural language (in this case, Japanese language) described in a VoiceXML document by voice as a program for realizing the VoiceXML execution means according to the present invention. As a program for realizing a VoiceXML document generating means according to the present invention, comprising a speech synthesis engine 103, a speech recognition engine 102 for recognizing customer pronunciation (speech), and a VoiceXML interpreter 101 for interpreting and executing a VoiceXML document. A Web server program 100 having a function of interpreting and executing a PHP code and automatically executing a questionnaire survey is provided. Program codes of these programs are stored in the data storage device 104 of the call center server 10.

  As shown in FIG. 3, the data storage device 104 of the call center server 10 includes a questionnaire survey execution file 3 used for the questionnaire survey, and a telephone number list 4 listing the telephone numbers of the customers who conduct the questionnaire survey. The schema 2 defining the structure of the answer data file 5 stored in the DWH server 11 is stored.

  In FIG. 3, the questionnaire survey execution file 3 and the telephone number list 4 stored in the call center server 10 are one, but actually, the questionnaire survey execution file 3 and the telephone number list 4 are stored for each questionnaire survey. The When the plurality of questionnaire survey execution files 3 and the telephone number list 4 are stored in the call center server 10, the file names of the questionnaire survey execution file 3 and the telephone number list 4 are described using the questionnaire survey identification data. Since the structure of the answer data file 5 is common regardless of the questionnaire survey, the call center server 10 stores only one schema 2.

  FIG. 5 is a diagram for explaining the questionnaire survey execution file 3 stored in the call center server 10. As described above, the questionnaire survey execution file 3 is described using PHP, which is one of programming languages. The Web server program 100 of the call center server 10 is included in the questionnaire survey execution file 3, interprets and executes the code enclosed by “<? Php” and “?>” As PHP code, and other codes are VoiceXML. Is output as a code.

  In the questionnaire survey execution file 3 illustrated in FIG. 5, a code 31 that defines the contents of the questionnaire survey is described between the VoiceXML form elements after the code 30 for creating the prolog element of the VoiceXML document. The attribute value of the ID attribute described in the start tag of the VoiceXML form element 3a is questionnaire survey identification data, and the questionnaire survey corresponding to the questionnaire survey execution file 3 can be identified from this ID attribute.

  Further, in FIG. 5, a code 31 for defining a questionnaire survey includes a code 32 for acquiring customer data of a customer to be surveyed and a VoiceXML document dynamically based on the customer data of a customer who conducts the questionnaire survey. The generated code 33 is included.

  FIG. 6 is a diagram for explaining a code 30 for creating a prolog element of a VoiceXML document. In this code, a code 30a for defining a character code output by PHP and a prolog element for defining a VoiceXML standard are output. A code 30b and a VoiceXML code 30c for outputting a message of a questionnaire survey are described.

  FIG. 7 is a diagram for explaining a code 32 for acquiring customer data of a customer to be surveyed. A code 32 for acquiring customer data from the SQL server 17 is included in the questionnaire survey execution file 3 in order to dynamically generate a VoiceXML document based on customer data for which the questionnaire survey is performed.

  In the code shown in FIG. 7, first, a PHP code 32a connected to the SQL server 17 for acquiring customer data is described, and then, using an SQL statement suitable for the specification of the SQL server 17, an SQL server is used. A PHP code 32b for acquiring customer data for a questionnaire survey is described from 17 tables (here, customer table 6).

  In FIG. 7, the record for acquiring customer data is specified using the telephone number (here, “TEL # No”) of the voice terminal device 13 that made the phone call. Get the value of the column “gender” indicating the customer's age, the column “age” indicating the customer's age, the value of the column “purchase” indicating the customer's purchase history, and assign the value of the column “gender” to the variable “customer # gender” The value of the column “age” is assigned to the variable “customer #age”, and the value of the column “purchase” is assigned to the variable “customer #purchase”.

  In FIG. 7, a code 32c for obtaining customer data using VoiceXML is described in order to cope with the case where customer data cannot be obtained from the SQL server 17. This code reads a file with the name "VoiceXML # uset # attribute.xml" in which a VoiceXML document for obtaining customer data is stored, and outputs the contents of this VoiceXML document using the PHP echo function. The PHP code to be described is described.

  FIG. 8 is a diagram for explaining a VoiceXML document 34 that acquires customer data from the voice terminal device 13. FIG. 8A shows a VoiceXML document 34 for acquiring customer data from the voice terminal device 13, and a VoiceXML field element 34 a is an input item of VoiceXML. The name attribute of the field element 34a indicates the variable name of the variable to which the answer data obtained from the customer is substituted. The element data of the prompt element 34b which is a child element of the field element 34a describes a question message for the customer. .

  The element data (in this case, the question message) of the prompt element 34b is converted into voice by the voice synthesis engine 103 of the call center server 10 and transmitted to the voice terminal device 13, and the call center server 10 waits for voice input from the customer. Become.

  The grammar element 34c is an element for designating the speech recognition dictionary 34g in which the utterance obtained from the customer and the answer data to be substituted into the variable indicated by the name attribute of the field element 34a are associated with each other. The grammar element 34c is used for the grammar element 34c. A recognition dictionary 34g is designated.

  FIG. 8B shows an example of a speech recognition dictionary 34g used when inquiring about a gender to a customer. In this speech recognition dictionary 34g, the speech synthesis engine 103 utters a question message “Tell me your gender”. The customer's pronunciation that becomes valid after conversion to, and answer data corresponding to the customer's pronunciation are described.

  For example, when the customer's pronunciation (speech) obtained by the speech recognition engine 102 recognizing the customer's pronunciation (speech) is “dan” or “dansei”, a variable “customer # gender” that substitutes the customer's gender. "Male" is substituted as the answer data.

  Note that the element name of the element storing the answer data in the schema 2 is used as the variable name to which the answer data obtained from the customer is substituted, that is, the variable name described in the attribute value of the name attribute of the field element 34a.

  Further, in FIG. 8A, an option element 34d is used so that the customer can reply by voice or tone signal. Furthermore, VoiceXML illustrated in FIG. 8A uses a filled element 34e and an if element 34f to send a reply to the customer according to the answer data obtained from the customer. According to FIG. 8, when inquiring about the gender of the customer, the call center server 10 notifies the voice terminal device 13 of the voice “Please tell me your gender” and the customer answers his gender (here, “ Male ”), the voice terminal device 13 is notified of the voice“ I'm male ”and a questionnaire survey is conducted while an interactive dialogue is made between the call center server 10 and the customer.

  Note that the VoiceXML document 34 described with reference to FIG. 8A actually includes a code for acquiring the customer's age and purchase history, as well as the customer's gender.

  FIG. 9 is a diagram for explaining a code 33 for dynamically generating a VoiceXML document according to customer data for which a questionnaire survey is performed. The code shown in FIG. 9 includes a code for changing the attribute of the “probody” element that controls the voice output of the VoiceXML document according to the age of the customer in order to change the voice output of the VoiceXML document according to the age of the customer. 33a is described using the PHP if syntax.

  In FIG. 9, when the variable “customer #age” into which the age of the customer is substituted is “65” or more, the value of the attribute “rate” of the “prosbody” element that determines the speed of the audio output is set to “slow”. The attribute “volume” of the body element that determines the volume of the audio output is set to “loud”.

  Also, in the code after else, when the variable “customer #age” into which the age of the customer is substituted is “65” or less, the value of the attribute “rate” of the “prosbody” element that determines the speed of audio output is “medium”. The attribute “volume” of the body element that determines the sound output volume is set to “medium”.

  The customer data can be used not only to control the voice output of the VoiceXML document but also to control the execution of the question message. In FIG. 9, when the value of the variable “customer #purchase” indicating the purchase history of the customer is “true” (with purchase experience), the customer is asked about the purchase reason. When the code # 01 "is executed and the value of the variable" customer #purchase "acquired from the SQL server 17 is" false "(no purchase experience), the contents of" Question # 01 "are not executed, A code 36 of “Question # 02” for asking the impression of the product is executed.

  FIG. 10 is a diagram for explaining the code 35 of “Question # 01” for asking the customer about the reason for purchase. FIG. 10A is a code 35 of “Question # 01” for asking the customer for the reason for purchase. The code of “Question # 01” for asking the customer for the reason for purchase in order to ask the customer for the reason for purchase according to the customer data. Is described using PHP, and in the PHP code 35a illustrated in FIG. 10A, when the condition of the if syntax (customer # purchase = true) is satisfied, the text described below the echo is the VoiceXML document 35b. Is output as

  This VoiceXML document 35b includes a field element 35c for asking the customer about the reason for purchase and obtaining the reason for purchase as answer data from the user. Between the prompt elements 35d that are child elements of the field element 35c, attributes are included. A body element 35e having “rate” and an attribute “volume” is described, and the values of these attributes are determined from the age of the customer.

  The VoiceXML document 35b output by the Web server program 100 interpreting the PHP code 35 of FIG. 10A is interpreted by the VoiceXML interpreter 101, and processing for asking the customer about the reason for purchase is executed. Note that the answer data obtained as the reason for purchase is defined by the speech recognition dictionary 35f illustrated in FIG.

  FIG. 11 is a diagram for explaining a code 36 of “Question # 02” for inquiring an impression on a product. FIG. 11A shows a “Question # 02” code 36 for inquiring an impression of a product. The contents of “Question # 02” executed regardless of the contents of customer data are described using a VoiceXML code 36a. The The answer data obtained at this time is defined by the speech recognition dictionary 36b in FIG.

  From here, functions realized by software installed in the call center server 10 will be described. FIG. 12 is a diagram for explaining a data flow in the call center server 10. When receiving the instruction to conduct the questionnaire survey, the Web server program 100 of the call center server 10 is activated, selects the questionnaire survey execution file 3 and the telephone number list 4 specified by the identification data of the designated questionnaire survey, and sets the telephone number. When telephones are made in the order listed in the list 4 and a call is made with the voice terminal device 13, the questionnaire survey is started by interpreting and executing the questionnaire survey execution file 3.

  As described above, the questionnaire survey execution file 3 is described using PHP, and the Web server program 100 has a function of a PHP parser that interprets and executes the syntax of PHP. The Web server program 100 delivers the VoiceXML document output by interpreting the PHP syntax described in the questionnaire survey execution file 3 to the VoiceXML interpreter 101.

  The VoiceXML interpreter 101 of the call center server 10 is software having a function of interpreting and executing the grammar and meaning of the VoiceXML document. When the voice is notified to the customer, that is, as the element data of the prompt element in the VoiceXML file. When the voice message is transmitted to the voice terminal device 13 by voice, the VoiceXML interpreter 101 uses the voice synthesis engine 102 to convert the message described in the text into voice, and the converted voice is sent to the voice terminal device 13. Send to.

  In addition, when obtaining answer data by voice from the customer, that is, after sending the question message described in the prompt element of the field element, the VoiceXML interpreter 101 causes the voice recognition engine 103 to recognize the customer's pronunciation (voice). Get customer pronunciation (text).

  The VoiceXML interpreter 101 obtains answer data corresponding to the pronunciation (text) of the customer by referring to the voice recognition dictionary specified by the grammar element, and the name of the field element that is the parent element of the executed prompt element. The answer data obtained from the customer is substituted into the variable value of the variable having the attribute value of the attribute as the variable name.

  When the VoiceXML interpreter 101 executes all of the VoiceXML documents output by interpreting the questionnaire survey execution file 3, the Web server program 100 executes the variable names and their variable values described in the VoiceXML file, that is, the field element. In accordance with the schema 2 read from the DWH server 11, the variable name and its variable value described in the attribute value of the name attribute and the variable name and its variable value used in the PHP syntax are sent to one customer. The contents of the executed questionnaire survey are added to the answer data file 5.

  Then, the Web server program 100 makes a call to the next telephone number described in the telephone number list 4 read from the DWH server 11, continuously conducts the questionnaire survey, and calls the telephone numbers described in the telephone number list 4. When the questionnaire survey is executed for all the numbers, the Web server program 100 ends the questionnaire survey.

  FIG. 13 is a diagram for explaining the answer data file 5 stored in the DWH server 11. The answer data file 5 is described in XML, and the structure of the answer data file 5 is defined in advance by the schema 2 regardless of the questionnaire survey.

  The answer data file 5 includes a form element 5b describing the contents of the questionnaire survey conducted in addition to the prolog element 5a defined such as the VoiceXML standard. The contents of the questionnaire survey conducted for one person are described in a transaction element 5c which is a child element of the form element 5b, and the transaction element 5c has a customer element 5d and an answer element 5e which are child elements. Before conducting the questionnaire survey, the number of transaction elements 5c described in the answer data file 5 is zero. Each time a questionnaire survey is conducted for one customer, one transaction element 5c is added to the answer data file 5 To be added.

  In FIG. 13, a child element of the customer element 5d includes a gender element 5f indicating the customer's age, an age element 5g indicating the customer's age, and a purchase element 5h indicating the customer's purchase history. If customer data is stored in the customer table 6 of the SQL server 17, the element value of the child element of the customer element 5d is the customer data obtained from the customer table 6.

  Further, as a matter of course, other child elements not shown in FIG. 13, for example, child elements for storing customer data obtained from customers in all questionnaire surveys executed by the call center server 10, such as membership numbers and addresses, are stored. It is included in the customer element 5d.

  In addition, as an element of the answer element 5e, an ans_motion element 5i indicating the reason for purchasing the product and an an_impression element 5j indicating the impression of the product are described. Of course, other child elements such as products of other companies are described. All the child elements for storing the answer data obtained from the customer in the questionnaire survey executed in the call center server 10 such as comparison with the answer element 5e are included in the answer element 5e.

  In the form element of the answer data file 5, transaction elements are described for the number of people who executed the questionnaire survey. The child elements of the customer element 5 d and the answer element 5 e having data are determined by the contents of the questionnaire survey.

  For example, as described with reference to FIGS. 9 and 10, only the customer who has purchased the product is asked a question as to the reason for purchasing the product, so that it corresponds to the customer who has not purchased the product. In the transaction element 5c, the ans_motion element 5j is an empty element.

  FIG. 14 is a view for explaining the tree structure of the answer data file 5 shown in FIG. As described above, the root element of the answer data file 5 is a form element 5b, and the form element 5b has one or more transaction elements 5c. The transaction element 5c has a customer element 5d and an answer element 5e, and the customer element 5d and the answer element 5e each have child elements.

  In the schema 2 stored in the call center server 10 and defining the structure of the answer data file 5, the tree structure shown in FIG. 14 is described in the XML schema language. FIG. 15 is a diagram for explaining the schema 2 that defines the structure of the answer data file 5.

  In the schema 2 that defines the structure of the answer data file 5, an xsd: element element 2b that defines each element of the answer data file 5 is described in addition to the prolog element 2a that defines the standard of the schema language.

  One element of the answer data file 5 is defined by one xsd: element element 2b, and the element name used in the answer data file 5 is described in the attribute value of the name attribute of the xsd: element element 2b.

  The transaction element 5c of the answer data file 5 illustrated in FIG. 5 is defined by the xsd: element element 2b having “transaction” as the attribute value of the name attribute, and the number of appearances is set to unlimited by the maxoccurs attribute.

  Furthermore, a customer element 5d and an answer element 5e of the answer data file 5 are defined between the xsd: element elements 2b having “transaction” as the attribute value of the name attribute. In addition, a child element of the customer element 5d (for example, a customer_gender element 5f) is defined between the xsd: element element 2b that defines the customer element 5d, and a child of the answer element 5e between the xsd: element element 2b that defines the answer element 5e. An element (ans_motion element 5i) is defined.

  The child element of the customer element 5d and answer element 5e cannot be defined so as to cover the questionnaire surveys related to all categories, but the category of the questionnaire survey conducted by the call center server 10, for example, the questionnaire survey related to products And if it is limited to the questionnaire survey for the purpose of CSR (CSR: Corporate Social Responsibility), it is possible to define the child elements of the customer element 5d and the answer element 5e because the questions to the customer are determined. The child element of the customer element 5d and the answer element 5e is used for storing answer data obtained from the customer.

  Next, the analysis server 12 having a function for verifying the effectiveness of a questionnaire will be described. FIG. 16 is a block diagram of the analysis server 12.

  As illustrated in FIG. 16, the analysis server 12 includes a template database 122 in which a template 7 describing a scenario for verifying the effect of a questionnaire is stored, and an answer data file 5 stored in the DWH server 11. The analysis program 120 that operates when totaling / analyzing data is included, and the logistic regression analysis is performed with the analysis module 121 used when performing clustering analysis as the analysis module 121 that totals / analyzes the answer data file 5 And an analysis module 121 used when performing association analysis.

  The template 7 in which the scenario for summing up / analyzing the answer data file 5 is described is preferably created for each questionnaire survey category and stored in the template database 122. This template 7 is used when counting / analyzing the answer data file 5 as the answer / data file 5 used for counting / analysis and the data indicating the answer data included in the answer data file 5 and as a scenario for counting / analysis. The analysis module 121 to be executed and the execution order thereof are described, and are determined according to the report 15 that outputs this scenario.

  For example, when three output results A to C are output to the report of the analysis server 12, the processing content for obtaining each of the output results A to C is described in the template 7. The output result A is an analysis result when a subset of customers who make similar answers in several answer elements 5e included in the response data file 5 is created, and an association analysis is performed for each subset. The template 7 includes data for specifying the answer data file 5 used for analysis (for example, identification data for questionnaire survey) and data for specifying child elements of the answer element 5e to be subjected to clustering analysis (here, The element name is input, and the analysis module 121 used in the analysis and its order (here, the analysis module for clustering analysis is executed and then the analysis module for association analysis is executed) are described in the template 7.

  At this time, the analysis program 120 of the analysis server 12 refers to the schema 2 according to the template 7, analyzes the structure of the response data file 5, and uses the analysis module 121 for clustering analysis from the specified response data file 5. Clustering analysis is performed on the child elements of the specified answer element 5e, and the result of association analysis using the analysis module of association analysis is output to the report 15 as an output result A for each subset obtained by the clustering analysis.

  As described above, if the structure of the answer data file 5 for storing the answer data obtained from the customer in the questionnaire survey is defined in advance in the schema 2, the structure of the answer data file 5 stored in the DWH server 11 becomes common. Therefore, it is not necessary to prepare a program for analyzing the answer data file 5 for each answer data file 5 in order to total the answer data files 5 stored in the DWH server 11.

  If the template 7 in which the processing procedure for collecting and analyzing the answer data included in the answer data file 5 is described is prepared, the work of analyzing the answer data file 5 stored in the DWH can be automated.

  The present invention is not limited to the embodiments described so far, and various modifications and changes can be made. For example, in the above-described embodiment, the voice recognition dictionary used in VoiceXML is defined in advance, and the voice recognition dictionary used in the grammar element is specified. However, the dictionary data used in the voice recognition dictionary is stored in the SQL server 17. A stored dictionary table can be provided, and a voice recognition dictionary can be dynamically created using this dictionary table.

  FIG. 17 is a diagram for explaining the contents of dynamically creating a speech recognition dictionary using the SQL server 17, and FIG. 17A is an example of PHP code for dynamically creating a speech recognition dictionary. 17 (b) is an example of a table provided in the SQL server 17, and FIG. 17 (c) is a diagram for explaining a speech recognition dictionary generated by executing the PHP code of FIG. 17 (a). .

  As shown in FIG. 17C, the dynamically generated speech recognition dictionary includes a grammar element 37a, a rule element 37b, a One-of element 37c, and the contents of the speech recognition dictionary that specify the speech recognition dictionary. Is composed of an item element 37d in which is described.

  The customer's voice or the like is defined by the attribute of the token element 37e described in the item element 37d. The customer's voice (for example, “1”) is defined by the value of the sym attribute of the token element 37e, and a variable name (here, “customer” is assigned to the answer data obtained from the customer's voice in the salt attribute of the token element 37e. #impression ") is defined, and answer data (for example," good ") to be substituted into a variable with the value attribute is defined.

  As illustrated in FIG. 17A, the PHP code 37f for generating the speech recognition dictionary of FIG. 17C uses a pq_query function to store data used in the item element 37d. (Here, “Tbl_impression”) is read, and a code for outputting the item element 37d of the speech recognition dictionary using the echo function is described.

  In the dictionary table 37g illustrated in FIG. 17B, a column “onsei” and a column “answer” are provided, and three records are provided. The data read from the column “onsei” is used as the value of the sym attribute of the token element 37e described in the item element 37d, and the data read from the column “answer” is the token element described in the item element 37d. Used for the value of the value attribute 37e.

  In the present embodiment, the call center server 10 performs a questionnaire survey using the telephone number list 4. However, a free dial for the questionnaire survey is displayed on the Web page published by the call center server 10, and free calls are made. You may make it carry out a questionnaire survey with respect to the customer who called the dial. When using the free dial, the call center server 10 acquires customer data from the SQL server 17 by using the telephone number of the voice terminal device 13 that has called the free dial.

  In addition, the questionnaire survey execution file 3, the answer data file 5, and the schema 2 described in the present embodiment can be arbitrarily set. As a matter of course, the structure of the answer data file 5, that is, the contents of the schema 2 is the same as that of the call center system. Determined according to the contents of the questionnaire survey.

The figure explaining the structure of a call center system. The flowchart which showed the procedure of the automation method of call center business. The block diagram of a call center system. The figure explaining the customer table provided in the SQL server. The figure explaining a questionnaire investigation execution file. The figure explaining the code which creates the prolog element of VoiceXML document. The figure explaining the code which acquires the customer data of a customer. The figure explaining the VoiceXML document which acquires customer data from a voice terminal device. The figure explaining the code | cord | chord for dynamically producing | generating a VoiceXML document. The figure explaining the code of “Question # 01” which asks the customer the reason for purchase. The figure explaining the code of “Question # 02” which asks the impression regarding the commodity. The figure explaining the data flow in a call center server. The figure explaining an answer data file. The figure explaining the tree structure of an answer data file. The figure explaining the schema which defines the structure of an answer data file. The block diagram of an analysis server. The figure explaining the content which creates a speech recognition dictionary dynamically using a SQL server.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Call center system 10 Call center server 100 Web server program 101 VoiceXML interpreter 102 Speech recognition engine 103 Speech synthesis engine 11 DWH server 12 Analysis server 13 Voice terminal device 14 PSTN
15 Report 16 VoIP Gateway Server 17 SQL Server 2 Schema 3 Questionnaire Survey Execution File 4 Telephone Number List 5 Response Data File 6 Customer Table 7 Template

Claims (16)

  A method for automating call center operations, wherein a server installed in a call center (call center server) acquires reference data used in call center operations from a specified table in a database connected to the call center server, Using the reference data, a VoiceXML document to be used in a call center service is dynamically generated, and by executing the generated VoiceXML document, the call center server utters a question message described in the VoiceXML document by voice. Step a for recognizing speech received from the voice terminal device and obtaining response data transmitted to the terminal device, in accordance with a schema that is described in a script language and that defines the structure of the response data file that stores the response data In the form, said voice A call center process including the step b of storing the answer data obtained from the terminal device in the answer data file, and a server (analysis server) installed for analyzing the answer data obtained in the call center process, Call center service automation method characterized by including an analysis step of analyzing the structure of the answer data file by referring to the schema and totaling / analyzing the answer data included in the specified answer data file .   The call center business automation method according to claim 1, wherein in the step a of the call center process, the call center server describes a code for dynamically generating the VoiceXML document, and the reference data is substituted. The call center business is characterized in that the reference data is acquired from the database and the VoiceXML document is dynamically generated by interpreting and executing a program described in a predetermined programming language. Automated method.   3. The call center business automation method according to claim 2, wherein the database includes a customer table storing customer data as a target of the call center business as the table, and the call center server in step a includes: The interpreting / executing program describes, as the reference data, a code for obtaining at least the customer data associated with the voice terminal device for voice communication from the customer table of the database. Automating call center operations.   4. The call center business automation method according to claim 3, wherein in the step a of the call center process, the program executed by the call center server in the step a includes the customer data acquired as the reference data. A call center task automation method characterized in that a code used as answer data is described.   5. The call center business automation method according to claim 3, wherein a program executed by the call center server in the step a in the step a of the call center process includes the reference data as the reference data. A call center business automation method, wherein a code for changing the question message included in the VoiceXML document is described according to the acquired customer data.   The call center business automation method according to any one of claims 3 to 5, wherein in the step a of the call center process, the program executed by the call center server in the step a includes the reference data. A call center task automation method, wherein a code that is included in the VoiceXML document and changes an attribute value related to the voice of the question message is described according to the customer data acquired as follows.   7. The call center business automation method according to claim 1, wherein dictionary data used in a voice recognition dictionary used in the VoiceXML document is stored in the database as the table. In the step a of the call center process, the program executed by the call center server in the step a acquires the dictionary data from the speech recognition dictionary table as the reference data. And a code for dynamically generating the voice recognition dictionary to be used when obtaining the answer data from the voice transmitted from the voice terminal device.   The call center automation method according to any one of claims 1 to 7, wherein a template in which a procedure of a process for counting and analyzing the answer data included in the answer data file is described is prepared, The call server task automation method, wherein the analysis server aggregates and analyzes the answer data included in the designated answer data file in accordance with the process indicated by the designated template.   A call center system comprising a call center server for automatically executing call center operations and an analysis server for automatically analyzing the results of call center operations, wherein the call center server is designated in a database connected to the call center server. The reference data used in the call center business is obtained from the table, and the VoiceXML document generation means for dynamically generating the VoiceXML document used in the call center business using the reference data, and interprets the generated VoiceXML document. VoiceXML execution means for transmitting the question message described in the VoiceXML document to the voice terminal device by executing the voice message, recognizing the voice received from the voice terminal device, and obtaining answer data, and a script Answer data storage means for storing the answer data obtained from the voice terminal device in the answer data file in a format according to a schema that defines the structure of the answer data file that is described in words and stores the answer data The analysis server includes an analysis unit that analyzes the structure of the answer data file by referring to the schema and totals / analyzes the answer data included in the specified answer data file. Call center system, characterized by that.   10. The call center system according to claim 9, wherein the VoiceXML document generating means of the call center server has a variable in which a code for dynamically generating the VoiceXML document is described, a variable to which the reference data is substituted, and a predetermined value. A call center system characterized in that the reference data is acquired from the database and the VoiceXML document is dynamically generated by interpreting and executing a program described in the above programming language.   11. The call center system according to claim 10, wherein the database includes the customer table storing the customer data as a target of a call center service as the table, and the VoiceXML document generation unit of the call center server. The call center system includes a code for obtaining at least customer data associated with the voice terminal device that performs voice communication from the customer table of the database as the reference data. .   12. The call center system according to claim 11, wherein the program executed by the VoiceXML document means of the call center server describes a code that uses the customer data acquired as the reference data as the answer data. Call center system characterized by that.   13. The call center system according to claim 11, wherein the program executed by the VoiceXML document generation unit of the call center server is included in the VoiceXML document according to the customer data acquired as the reference data. A call center system in which a code for changing the question message is described.   The call center system according to any one of claims 11 to 13, wherein the program executed by the VoiceXML document unit of the call center server includes the VoiceXML according to the customer data acquired as the reference data. A call center system characterized in that a code included in a document and describing an attribute value related to the voice of the question message is described.   15. The call center system according to claim 9, wherein dictionary data used in a voice recognition dictionary used in the VoiceXML document is stored in the database as the table. A recognition dictionary table is provided, and the program executed by the VoiceXML document generation means of the call center server acquires the dictionary data from the speech recognition dictionary table as the reference data and is transmitted from the voice terminal device A call center system in which a code for dynamically generating the speech recognition dictionary used when obtaining the answer data from speech is described. The call center system according to any one of claims 9 to 15, wherein a template is prepared in which a procedure for a process of aggregating and analyzing the answer data included in the answer data file is prepared, and the analysis is performed. The server totals and analyzes the answer data included in the designated answer data file in accordance with the process indicated by the designated template.

Images