TWI765463B - Verification system, method and computer-readable medium for customer service - Google Patents

Abstract

The invention discloses a verification system, method and computer-readable medium for customer service. A phone number verification module retrieves a numeric string of a phone number from a phone number string input by a customer to determine consistency between the phone number and an incoming phone number, and asks whether the customer is a registrant of the phone number. Then, when the customer is the registrant of the phone number, a name verification module judges consistency between a name string input by the customer and the name of the registrant of the phone number according to a similarity confidence score of the name, wherein the name verification module compares a name phonetic string of the name string with a corresponding name standard answer phonetic string to calculate the similarity confidence score of the name.

Description

Authentication system, method and computer readable medium for customer service

The present invention relates to an authentication technology, and more particularly, to an authentication system, method and computer-readable medium for customer service.

At present, when a customer requests service from the customer service staff by calling the customer service phone number or online customer service voice, the customer service staff often cannot automatically identify the customer's identity, because the phone number used by the customer is often not a pre-bound phone number, and the customer's voice usually needs to be used at this time. To confirm the identity of the customer, for example, the customer needs to verbally state the relevant account information of the customer account to the customer service staff.

In a prior art, a method for authenticating customer identity among customer service personnel is proposed, including: sending an account alias code associated with a customer account to a client associated with a service; and receiving feedback information from the terminal, so as to use the feedback information and account alias code according to the feedback information Authenticate client identity.

However, this existing technology cannot certify multiple verification items (such as phone number, certificate number, name, date of birth or address, etc.) in advance, nor can it be developed according to the data characteristics required by the verification item during verification. Appropriate normalization methods and data conversion methods are also unable to use phonetic symbols and tones removal methods for verification items (such as names or addresses), and it is impossible to establish verification. The mechanism of similarity confidence score and threshold value of items.

Therefore, how to provide a set of authentication technology for customer service to solve the above-mentioned problems in the prior art has become a major issue for those skilled in the art.

The present invention provides an authentication system, method and computer-readable medium for customer service, which is used to authenticate multiple authentication items (such as telephone numbers, license numbers, names, birthdays or addresses, etc.) Develop appropriate normalization methods and data conversion methods for the data characteristics required by the project during verification, or use phonetic symbols and tones removal methods for verification projects (such as names or addresses), or establish a verification method. The mechanism of similarity confidence score and threshold value of certificate projects.

The verification system for customer service of the present invention includes: a phone number verification module, which extracts a digital string of a telephone number from a telephone number string input by a customer, so that the telephone number verification module can determine the digits of the telephone number. The consistency between the string and the incoming line's phone number, and then the phone number verification module asks whether the customer is the registrant of the phone number; and the name verification module is based on the phone number verification module to ask the customer's phone number. When the registrant of the phone number is the registrant, the name verification module determines the consistency between the name string input by the customer and the name of the registrant of the phone number based on the similarity confidence score of the name. The name verification module is a Compare the phonetic character string of the name string entered by the customer with the phonetic character string of the corresponding standard answer of the name to calculate the similarity confidence score of the name.

The verification method for customer service of the present invention includes: the phone number verification module extracts the phone number character string from the phone number string input by the customer, so that the phone number verification module determines the phone number character string The consistency between the serial number and the incoming line phone number, and then The phone number verification module asks whether the customer is the registrant of the phone number; and when the phone number verification module asks the customer to be the registrant of the phone number, the name verification module judges based on the similarity confidence score of the name The consistency between the name string input by the customer and the name of the registrant of the phone number, among which, the name verification module compares the phonetic string of the name string input by the customer with the corresponding phonetic name standard answer phonetic The phonetic comparison of the character string is performed to calculate the similarity confidence score of the name.

The computer-readable medium of the present invention is applied to a computing device or a computer, and stores instructions for executing the aforementioned authentication method for customer service.

In order to make the above-mentioned features and advantages of the present invention more obvious and easy to understand, the following embodiments are given and described in detail with the accompanying drawings. Additional features and advantages of the present invention will be set forth in part in the description that follows, and in part will be apparent from the description, or may be learned by practice of the invention. It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory and are not intended to limit the scope of the invention as claimed.

1: Authentication system for customer service

10: Phone number verification module

20: Certificate number verification module

30:Name verification module

40: Birthday Verification Module

50:Address verification module

60:Database

70: Comparison table of redundant words / redundant words and punctuation marks

80: Comparison table of similar phonetic phonetics

A: Customer

B: customer service staff

S: Authentication method for customer service

S1: Phone Number Verification

S2: Certificate number verification

S3: Name Verification

S4: Birthday verification

S5: Address Verification

S11 to S18: Steps

S21 to S26: Steps

S31 to S38: Steps

S41 to S46: Steps

S51 to S58: Steps

FIG. 1 is a schematic diagram of the architecture of the authentication system for customer service according to the present invention;

2 is a schematic flowchart of the authentication method for customer service of the present invention;

3 is a schematic flow chart of a phone number verification module (phone number verification) in the verification system and method for customer service of the present invention;

4 is a schematic flowchart of the verification procedure of the certificate number verification module (certificate number verification) in the verification system and method for customer service of the present invention;

5 is a schematic flowchart of the verification procedure of the name verification module (name verification) in the verification system and method for customer service of the present invention;

6 is a schematic flowchart of the verification procedure of the birthday verification module (birthday verification) in the verification system and method for customer service of the present invention; and

FIG. 7 is a schematic flowchart of the verification procedure of the address verification module (address verification) in the verification system and method for customer service of the present invention.

The embodiments of the present invention are described below with specific specific embodiments. Those skilled in the art can understand other advantages and effects of the present invention from the contents disclosed in this specification, and can also be implemented by other different specific equivalent embodiments. or use.

1 is a schematic diagram of the structure of a verification system 1 for customer service according to the present invention, and FIG. 2 is a schematic flowchart of a verification method S for customer service according to the present invention, and the verification system 1 or the verification method S for customer service It can be composed according to functional structure, data characteristics or service flow.

As shown in FIG. 1, the authentication system 1 for customer service may (sequentially) include a phone number verification module 10, a certificate number verification module 20, a name verification module 30, a birthday verification module 40, and an address verification module 50 and other five verification modules. As shown in FIG. 2 , the verification method S for customer service may (sequentially) include five verification procedures including phone number verification S1, certificate number verification S2, name verification S3, birthday verification S4, and address verification S5. The telephone number verification module 10 may be a telephone number verification device (chip/circuit), a telephone number verification software (program), etc., and the certificate number verification module 20 may be a certificate number verification device (chip/circuit), a certificate number verification software ( program), etc., the name verification module 30 can be a name verifier (chip/circuit), a name verification software (program), etc., and the birthday verification module 40 can be a birthday verifier (chip/circuit) chip/circuit), birthday verification software (program), etc. The address verification module 50 may be an address verifier (chip/circuit), address verification software (program), and the like. In addition, the front-end system (not shown in the figure) described below can be a server (such as a front-end server), a host (such as a front-end host), a front-end platform (software), or an application (App), etc. The electronic device (not shown in the figure) ) can be a telephone, a mobile phone (such as a smart phone), a smart watch, a computer (such as a tablet computer), and the like.

The following will describe the content of each verification module and each verification program:

1. Phone number verification module 10 (phone number verification S1): Customer A can go to the front-end system in advance. When the front-end system determines that the incoming customer A needs to be authenticated, it will make the authentication program of customer A enter the phone number Verify the module 10, and the phone number of customer A can be a local phone number or a mobile phone number or the like. The guiding tone of the phone number verification module 10 will request the customer A to input (speak) the phone number string through the electronic device's voice (speak) or press the button, so as to extract the phone number string that meets the rules from the phone number string input by the customer A. The numeric string of the phone number, and ask if Customer A is the registrant of the phone number? If the customer A is the registrant of the telephone number, the telephone number verification module 10 directs the verification procedure of the customer A to the certificate number verification module 20 . On the other hand, if customer A is not the registrant of the telephone number, the telephone number verification module 10 ends the verification procedure of customer A to lead out of the verification system 1 for customer service, because the subsequent four verification modules 20/ 30/40/50 will check the personal information with the registrant of the phone number, so the registrant of the phone number can only proceed.

2. Certificate number verification module 20 (certificate number verification S2): After the phone number of customer A is verified by the phone number verification module 10, the verification process of customer A will enter the certificate number verification module 20, and the customer The ID number of A can be the identity card number or the unified number (such as the company's unified number). The certificate number verification module 20 will request the customer A to input (speak) through the voice of the electronic device or press The card number string is input in the key mode, and the certificate number digital string that conforms to the rules is extracted from the certificate number string input by customer A, and then the certificate number verification module 20 determines whether the retrieved certificate number digital string is the same as the one. Is the information of the registrant's ID number of the correct phone number the same (that is, the consistency of the two information)? If the two data are inconsistent, the certificate number verification module 20 guides the verification procedure of customer A to customer service staff B, so that customer service staff B can query and confirm customer A's certificate number. On the contrary, if the data of the two are consistent, the certificate number verification module 20 guides the verification procedure of the client A to the name verification module 30 .

3. Name verification module 30 (name verification S3): After the certificate number of client A is verified by the certificate number verification module 20, the verification process of client A will enter the name verification module 30, and the name of client A It can be a name or a company name, etc. The name verification module 30 will request the customer A to input (speak) the name string through the voice of the electronic device, so that the name verification module 30 obtains from the name string input (spoken) by the customer A through the voice recognition technology The character string of the voice recognition result of the name, and the character string of the voice recognition result of the name is subjected to character string preprocessing and data conversion by the name verification module 30, and then the name verification module 30 is based on the similarity score of the name similarity confidence value ( The mechanism of Confidence Measure Rate) and Threshold Rate (Threshold Rate) determines whether the name string entered (spoken) by customer A through the voice of the electronic device is the name of the registrant of the correct phone number (correct name), and then The authentication procedure of client A is directed to the birthday verification module 40 .

4. Birthday verification module 40 (birthday number verification S4): After the name of customer A is verified by the name verification module 30, the verification process of customer A will enter the birthday verification module 40, and the birthday of customer A can be For the birthday date, etc. The birthday verification module 40 will request the customer A to input (speak) the birthday date string through the voice method of the electronic device or input the date string of the birthday by pressing the button, so as to The birthday verification module 40 extracts a numeric string of the birthday date that conforms to the rules from the birthday date string input by the customer A, and converts the year-specific part of the numeric string of the birthday date to There are two types of the Western New Year and/or the Republic of China. The birthday verification module 40 determines whether the converted birthday date is the birthday of the registrant of the correct telephone number, and then guides the verification process of Client A to Address verification module 50 .

5. Address verification module 50 (address number verification S5 ): After the birthday of customer A is verified by the birthday verification module 40 , the verification procedure of customer A will enter the address verification module 50 . The address verification module 50 will request the customer A to input (speak) the registered address string through the voice of the electronic device, so that the address verification module 50 can input (speak) the address word from the customer A through the voice recognition technology. The word string of the speech recognition result of the address is obtained from the string, and the address verification module 50 performs string preprocessing and data conversion on the word string of the speech recognition result of the address, and then converts the address standard answer string (correct address string) Expand all possible combinations (address combination strings), and then the address verification module 50 uses the mechanism of the similarity confidence value score and threshold value of the addresses to determine whether the address string input (spoken) by customer A is the correct registered address .

Therefore, after the above five verification modules (five verification procedures) complete the verification or verification, the verified or verified information can be provided to the customer service personnel B for acceptance, so as to use the service process or modular design To reduce the average handling time (Average Handling Time; AHT) of customer service agent B's certification. That is, in order to achieve voice-based (automation) service and reduce the average processing time (AHT) of manual verification of customer service personnel B, and to meet the needs of fast verification services, the present invention introduces an automated verification system and method, and uses human The machine-to-machine collaboration method assists customer service staff B to verify first, which improves the service efficiency of customer service staff B and reduces the average processing time (AHT) of customer service staff B.

The present invention can carry out the data characteristic development and rule establishment of each verification item (such as telephone number, certificate number, name, date of birth or address), and develop corresponding normalization methods and data conversion methods according to the characteristics of different verification items , and can also set the similarity confidence score and threshold value of each verification item to judge whether it is a valid result and the identity of the registrant of the phone number.

The present invention provides a modular or automated verification system and method, and introduces a normalized method or a data conversion method, and adds a judging mechanism for the similarity confidence score and threshold value of each verification item, so as to improve the verification accuracy. Usability and correctness, thereby applied to customer service (eg, customer service staff).

In order to make the operation of the customer A more flexible, the present invention not only provides the voice input of the electronic device, but also provides the dual-tone multi-frequency (DTMF) button input, that is, the dual-track input function (ie Voice or key input), so that customer A can choose the phone number verification module 10 (phone number verification S1), certificate number verification module 20 (certificate number verification S2), birthday verification module 40 (birthday verification S4) , the verification can be performed through the voice or keystroke input of the electronic device. For example, in a crowded or noisy place, customer A does not want to input (speak) his verification content through the voice of the electronic device or worry about affecting the recognition result, so he can choose to input through the button of the electronic device for verification. .

3 is a schematic flowchart of the verification procedure of the telephone number verification module 10 (telephone number verification S1 ) in the verification system and method for customer service of the present invention, and is described with reference to FIGS. 1 and 2 .

As shown in step S11 of FIG. 3 , the phone number verification module 10 obtains the character string of the phone number recognition result (eg, the voice recognition result), and the phone number can be the local telephone number or the mobile phone number of the customer A. For example, the phone number verification module 10 can identify techniques for phone numbers The language model (Language Model; LM) in the recognition technology (such as speech recognition technology) increases the weight of the appearance of numbers and narrows the combination range of the word groups in the language model (LM), so that the language in the recognition technology (such as speech recognition technology) The model (LM) has a higher probability of appearing the combination of the digit string of the phone number.

As shown in step S12 of FIG. 3 , the phone number verification module 10 normalizes the data format of the phone number of customer A. For example, the phone number verification module 10 can delete the non-digital words in the character string of the phone number recognition result (such as the speech recognition result) and the part of the digits in the reserved phone number string, and then put the phone number in the character string. Parts of numbers are converted to Arabic numeral format.

As shown in step S13 of FIG. 3 , the phone number verification module 10 retrieves the numeric string of the phone number. For example, the phone number verification module 10 can extract 9 to 10 digits after the beginning of "0" for the digit string (a string of Arabic numerals) of the phone number.

As shown in step S14 of FIG. 3 , the phone number verification module 10 determines the rule of the digit string of the phone number. For example, the phone number verification module 10 can determine the consistency between the digital string of the captured phone number and the incoming phone number (ie, whether the two are consistent).

As shown in step S15 of FIG. 3 , the phone number verification module 10 determines whether the incoming phone number is a valid door number? For example, the phone number verification module 10 can be linked to the database 60 (eg, the back-end database), so as to determine whether the incoming phone number is a valid door number through the comparison of the database 60 (eg, the back-end database) ? If no (the incoming phone number is not a valid door number), then proceed to step S16 in FIG. 3 , and the phone number verification module 10 guides the customer service personnel B to inquire about the incoming phone number. confirm. On the contrary, if it is (the incoming telephone number is a valid door number), then proceed to step S17 of FIG. 3 .

As shown in step S17 of FIG. 3 , the phone number verification module 10 asks whether it is the registrant of the phone number. For example, the phone number verification module 10 may ask whether Customer A is the registrant of the phone number? If not (customer A is not the registrant of the telephone number), then proceed to step S18 in FIG. 3 , and the telephone number verification module 10 ends the verification procedure of customer A to guide him to leave the verification system 1 for customer service. The four verification modules 20/30/40/50 will check personal information with the registrant of the phone number, so the registrant who needs the phone number can proceed. On the contrary, if (customer A is the registrant of the telephone number), the process of the telephone number verification module 10 is ended, and the process of the certificate number verification module 20 (certificate number verification S2 ) shown in FIG. 4 is performed.

FIG. 4 is a schematic flowchart of the verification procedure of the card number verification module 20 (card number verification S2 ) in the verification system 1 and method for customer service of the present invention, and is described with reference to FIGS. 1 and 2 .

As shown in step S21 of FIG. 4 , the card number verification module 20 obtains the string of the identification result of the card number (such as the voice recognition result), and the card number can be the ID card number of the customer A or a unified number (such as the company's unified number) )Wait. For example, the certificate number verification module 20 can increase the weight of digit occurrences and narrow the combination range of the word groups in the language model (LM) for the language model (LM) in the identification technology of the certificate number (such as the speech recognition technology), so as to The language model (LM) in the recognition technology (such as speech recognition technology) has a higher probability to appear the combination of the digital string of the certificate number.

As shown in step S22 of FIG. 4 , the certificate number verification module 20 normalizes the data format of the customer A's certificate number. For example, the certificate number verification module 20 can delete the non-digital words in the character string of the identification result (such as the speech recognition result) of the certificate number and the number part in the character string of the reserved certificate number, and then replace the character string of the certificate number in the character string of the certificate number. Parts of numbers are converted to Arabic numeral format.

As shown in step S23 of FIG. 4 , the certificate number verification module 20 retrieves the digital string of the certificate number. For example, the certificate number verification module 20 can extract the digits of multiple numbers (eg, nine numbers) for the number string after the beginning of the number string "1" or "2" of the certificate number, and the numbers of the nine numbers are Numbers can represent the ID card number as the ID card number. Alternatively, the certificate number verification module 20 may extract the digits of all numbers (eg, eight numbers) for the number string of the certificate number, and the numbers of the eight numbers may represent that the certificate number is a unified number (eg, a company's unified number).

As shown in step S24 of FIG. 4 , the certificate number verification module 20 determines the rules of the digital string of the certificate number. For example, the certificate number verification module 20 can determine whether the certificate number is a number obtained by extracting multiple numbers or all numbers (eg, nine numbers or eight numbers), so as to obtain the numbers from multiple numbers or all numbers (eg, nine numbers or eight numbers) The last number (such as the last four numbers) or the digits of all numbers required to obtain authentication among the digits of the number), the digits of the last number can represent the identity card number, and the digits of all numbers It can be represented by a unified number (such as the company's unified number).

As shown in step S25 of FIG. 4 , the certificate number verification module 20 determines whether the certificate number is consistent with the registration data? For example, the phone number verification module 10 can be linked to the database 60 (eg, the back-end database), so as to determine whether the card number is a correct phone number (door number) through the comparison of the database 60 (eg, the back-end database). The certificate number (registration number) of the registrant with the number)? If no (the certificate number is not the certificate number of the registrant with the correct telephone number), then proceed to step 26 in FIG. 4 , and the telephone number verification module 10 guides the customer service personnel B, so that the customer service personnel B can verify the certificate number. Query confirmation. On the contrary, if (the certificate number is the certificate number of the registrant of the correct telephone number), the process of the certificate number verification module 20 is ended, and the process of the name verification module 30 (name verification S3) shown in FIG. 5 is performed. .

FIG. 5 is a schematic flowchart of the verification procedure of the name verification module 30 (name verification S3 ) in the verification system 1 and method for customer service of the present invention, and is described with reference to FIGS. 1 and 2 .

As shown in step S31 of FIG. 5 , the name verification module 30 obtains the character string of the voice recognition result of the name, and the name can be the name of the customer A or the company name or the like. For example, the name verification module 30 can use the speech recognition technology to obtain the name string input (speak) by the customer A through the electronic device, and the speech recognition technology can be Automatic Speech Recognition (ASR) technology.

As shown in step S32 of FIG. 5 , the name verification module 30 removes redundant words/words in the name string. For example, the name verification module 30 can match the name string input (speaked) by the voice method of the electronic device by the customer A with the superfluous words/superfluous words and punctuation mark comparison table 70, so as to remove the superfluous words/superfluous words in the name string word and keep the rest of the name string.

As shown in step S33 of FIG. 5 , the name verification module 30 converts the name string into a toneless name phonetic string. For example, the name verification module 30 can convert the name string processed in the aforementioned step S32 into a name phonetic string, so as to remove the tones and reserve the phonetic part of the name phonetic string.

As shown in step S34 of FIG. 5 , the name verification module 30 obtains the name standard answer string from the database 60 . For example, the name verification module 30 can be linked to the database 60 (eg, a backend database) to obtain the name answer string from the database 60, and the name answer string is the correct phone number of the registrant himself name string.

As shown in step S35 of FIG. 5 , the name verification module 30 normalizes the data format of the name standard answer string. For example, the name verification module 30 can cooperate with redundant words/words and punctuation The number comparison table 70 is used to remove redundant words/words in the name standard answer string, add punctuation and notation symbols to the name standard answer string, and retain the remaining name standard answer string.

As shown in step S36 of FIG. 5 , the name verification module 30 converts the name standard answer string into a toneless name standard answer phonetic string. For example, the name verification module 30 can convert the name standard answer string into the name standard answer phonetic string, so as to remove the tone and retain the phonetic part of the name standard answer phonetic string.

As shown in step S37 of FIG. 5 , the name verification module 30 calculates the similarity confidence score of the name. For example, the name verification module 30 can phoneticize the name phonetic character string (phonetic symbol string) of the name string input (spoken) by the customer A through the electronic device's voice method and the corresponding phonetic character string of the standard answer of the name The comparison is used to calculate the similarity confidence score of the name, which is beneficial to solve the problem of Chinese homophones. In addition, the name verification module 30 can also cooperate with the similar phonetic phonetic comparison table 80 to process the phonetic phonetics of some similar phonetics in the name phonetic string or the name standard answer phonetic string.

As shown in step S38 of FIG. 5 , the name verification module 30 determines the threshold value of the name. For example, the name verification module 30 can calculate (comparison) the name phonetic character string (phonetic symbol string) and the corresponding name standard answer phonetic character string according to the similarity confidence value score of the name obtained in the aforementioned step S37. For the similarity result, the name verification module 30 determines whether the similarity confidence score of the obtained name passes the set threshold value according to the similarity result between the name phonetic character string and the name standard answer phonetic character string. If (the similarity confidence value score of the name passes the threshold), the name verification module 30 compares the name phonetic character string (phonetic symbol string) of the name string input (spoken) by customer A with the phonetic characters of the name standard answer. The two strings are judged to be (considered) consistent, and end The flow of the name verification module 30 is used to perform the flow of the birthday verification module 40 (birthday verification S4 ) shown in FIG. 6 .

FIG. 6 is a schematic flowchart of the verification procedure of the birthday verification module 40 (birthday verification S4 ) in the verification system 1 and method for customer service of the present invention, and is described with reference to FIGS. 1 and 2 .

As shown in step S41 of FIG. 6 , the birthday verification module 40 obtains the character string of the recognition result (eg, the speech recognition result) of the birthday date. For example, the birthday verification module 40 can increase the weight of digit occurrence and narrow the combination range of the word group in the language model (LM) for the language model (LM) in the recognition technology (such as the speech recognition technology) of the birthday date. In order to make the language model (LM) in the recognition technology (such as speech recognition technology) have a higher probability to appear the combination of the digital string of the birthday date.

As shown in step S42 of FIG. 6 , the birthday verification module 40 converts the numeric string of the birthday date into the Western Year and/or the Republic of China Year. For example, the birthday verification module 40 can convert the obtained digital string of the date of birth to the format of the year of the Western year and/or the year of the Republic of China, so as to obtain the date of birth of the two types of the year of the Western Yuan and/or the Republic of China A numeric string in the format of day representation.

As shown in step S43 of FIG. 6 , the birthday verification module 40 obtains the birthday date standard answer string from the database 60 . For example, the birthday verification module 40 can be linked to the database 60 (eg, a backend database) to obtain the standard answer string of the date of birth from the database 60, and the standard answer string of the date of birth is the correct phone number The birthday number string of the registrant of the number.

As shown in step S44 of FIG. 6 , the birthday verification module 40 normalizes the data format of the birthday date standard answer string. For example, the birthday verification module 40 can cooperate with the superfluous words/superfluous words and punctuation mark comparison table 70 to remove superfluous words/superfluous words in the standard answer string of birthday date, Punctuation and notation symbols are added to the standard answer string of date, year, month, and date, and the remaining standard answer strings of date, date, and date are reserved.

As shown in step S45 of FIG. 6 , the birthday verification module 40 calculates the similarity confidence score of the birthday date. For example, the birthday verification module 40 can compare the combination of the birthday number string input by the customer A and the birthday number string of the standard answer to calculate the similarity confidence score of the date of birth.

As shown in step S46 of FIG. 6 , the birthday verification module 40 determines the threshold value of the birthday date. For example, the birthday verification module 40 can calculate (comparison) the similarity result between the birthday digital string and the birthday digital string of the standard answer according to the similarity confidence score of the birthday date obtained in the aforementioned step S45, to obtain The birthday verification module 40 determines whether the obtained similarity confidence score of the birthday date passes the set threshold value according to the similarity result between the birthday number string and the birthday number string of the standard answer. If (the similarity confidence score of the birthday date passes the threshold), the birthday verification module 40 determines (considered) the combination of the birthday number string input by customer A and the birthday number string of the standard answer. , and end the process of the birthday verification module 40 to perform the process of the address verification module 50 (address verification S5 ) shown in FIG. 7 .

FIG. 7 is a schematic flowchart of the verification procedure of the address verification module 50 (address verification S5 ) in the verification system 1 and method for customer service of the present invention, and is described with reference to FIGS. 1 and 2 .

As shown in step S51 of FIG. 7 , the address verification module 50 obtains the character string of the speech recognition result of the address. For example, the address verification module 50 can use the speech recognition technology to obtain the address string input (spoken) by the customer A through the voice of the electronic device, and the speech recognition technology can be automatic speech recognition (ASR) technology.

As shown in step S52 of FIG. 7 , the address verification module 50 removes redundant words/words in the address string. For example, the address verification module 50 can match the address string entered (speaked) by customer A through the voice of the electronic device with the redundant word/redundant word and punctuation mark comparison table 70 to remove redundant words/redundant words in the address string word and the remaining address strings are reserved.

As shown in step S53 of FIG. 7 , the address verification module 50 converts the address string into a toneless address phonetic string. For example, the address verification module 50 can convert the address string processed in the aforementioned step S52 into an address phonetic string, so as to remove the tones and reserve the phonetic part of the address phonetic string, which is beneficial to Solve the problem of related homonyms in addresses such as township, li, road, etc.

As shown in step S54 of FIG. 7 , the address verification module 50 obtains the address standard answer string from the database 60 . For example, the address verification module 50 can be linked to the database 60 (eg, a backend database) to obtain the address standard answer string from the database 60, and the address standard answer string is the correct address of the registrant of the telephone number. string.

As shown in step S55 of FIG. 7 , the address verification module 50 expands all possible combinations (address combination strings) of the address standard answer string. For example, the address verification module 50 can list all possible address combination strings for the address canonical answer string (correct address string) obtained from the database 60 (eg, the backend database), because this verification item There may be a lack or inconsistency between the address entered (spoken) by Customer A and the actual registered address, so all possible address combinations must be listed for subsequent comparison and confirmation.

As shown in step S56 of FIG. 7 , the address verification module 50 converts the address standard answer string into a toneless address standard answer phonetic string. For example, the address verification module 50 may All possible address combination strings are converted into address standard answer phonetic strings respectively, so as to remove the tones from the address standard answer phonetic string and retain the phonetic part of the address standard answer phonetic string.

As shown in step S57 of FIG. 7 , the address verification module 50 calculates the similarity confidence score of the addresses. For example, the address verification module 50 can perform phonetic comparison between the address phonetic character string (phonetic symbol string) of the address string inputted (speaked) by customer A through the voice method of the electronic device and the phonetic character string of the address standard answer. Calculate the similarity confidence score of the address, which is beneficial to solve the problem of homophones in the address. In addition, the address verification module 50 can also cooperate with the similar phonetic phonetic comparison table 80 to process the phonetic phonetics of some similar phonemes in the address phonetic string or the phonetic string of the address standard answer.

As shown in step S58 of FIG. 7 , the address verification module 50 determines the threshold value of the address. For example, the address verification module 50 can calculate (comparison) the similarity result between the address phonetic character string and the address standard answer phonetic character string according to the similarity confidence value score of the address obtained in the aforementioned step S57, and according to the method of the present invention The set address needs to review the rules of the unit (such as road, number, building) to judge whether the similarity confidence score of the obtained address passes the set threshold value? If (the similarity confidence score of the address passes the threshold and passes the rule), the address verification module 50 compares the address phonetic string (phonetic symbol string) of the address string input (spoken) by the customer A with the address standard The two answer phonetic strings are determined to be (considered) consistent, and the process of the address verification module 50 ends. In this way, the verification process of the five verification modules is completed, and the verification process of customer A is guided to the corresponding service or customer service personnel B to accept the service.

Embodiments of the authentication system and method for customer service of the present invention are described below. As shown in FIG. 1 to FIG. 7 , in this embodiment, customer A enters the verification system 1 for customer service to enter each verification module in sequence, for example, enter the phone number verification module 10, the certificate number verification module in sequence Module 20, Name verification module 30, birthday verification module 40 and address verification module 50, and describe the processing flow or processing method of each verification module.

As shown in Figure 3, after customer A enters the line, the verification process of customer A will first enter the phone number verification module 10 (phone number verification S1), and the phone number verification module 10 will request customer A through the voice method of the electronic device Input (speak) or press key to input the phone number (such as local phone number or mobile phone number) to be serviced. For example, if customer A chooses to input (speak) a phone number (such as a mobile phone number) through the voice of the electronic device, the phone number verification module 10 will generate the phone number through the voice recognition technology. Phone 40956123456".

Next, the phone number verification module 10 will normalize the data format of the character string of the phone number recognition result (eg, the speech recognition result), so as to delete the non-numeric characters in the character string of the phone number recognition result (eg, the speech recognition result) Words and reserved numbers in the phone number string, and then convert the numbers in the phone number string into Arabic numerals. For example, if the character string of the speech recognition result of the phone number is "my phone number 40956123456", the phone number verification module 10 will delete the non-digits in the character string of the speech recognition result of the phone number The word "my phone" and the reserved number part "40956123456" to convert the number part "40956123456" into an Arabic numeral format "40956123456", and then extract the number string of 9 to 10 numbers after the beginning of "0" from the numeric string (a string of Arabic numerals) of the phone number to become the phone number "0956123456".

Then, the phone number verification module 10 will determine whether the retrieved phone number "0956123456" and the incoming phone number are the same (ie, the consistency of the two)? If (the two are consistent), the phone number verification module 10 can further determine whether the incoming phone number is Valid door number (phone number)? If it is (the incoming phone number is a valid door number), the phone number verification module 10 can then ask whether Customer A is the registrant of the phone number? If it is otherwise (the customer A is the registrant of the telephone number), the process of the telephone number verification module 10 is ended, and the process of the certificate number verification module 20 (certificate number verification) shown in FIG. 4 is performed.

As shown in FIG. 4 , the verification process of customer A enters the certificate number verification module 20 (certificate number verification S2 ), and the certificate number verification module 20 will request customer A to input (speak) or press a button through the electronic device. Enter the card number string, and the card number (card number string) can be an ID card number or a unified number (company unified number). For example, if customer A chooses to input (speak) a card number (such as an ID card number) through the voice of the electronic device, the card number verification module 20 will generate the voice recognition result of the card number through the voice recognition technology. The ID number is 122233456."

Next, the card number verification module 20 will normalize the data format of the character string of the identification result of the card number (such as the voice recognition result), so as to delete the non-numeric characters in the string of the identification result of the card number (such as the voice recognition result). The word and the number part in the string of reserved card numbers are converted into Arabic numerals. For example, the character string of the speech recognition result of the ID card number is "My ID number is 122233456", the ID verification module 20 will delete the non-digits in the character string of the speech recognition result of the ID card number The word "my ID number is" and the reserved part of the number "122233456" are used to convert the part of the number into the format of Arabic numerals "122233456", and then target the number of the card number. After the string (a string of Arabic numerals) is determined to be the digits of multiple numbers (such as nine numbers), the certificate number verification module 20 obtains the last number (such as the last four numbers) or all numbers required for verification the numbers.

Then, the certificate number verification module 20 can use the database 60 (such as a back-end database) to compare whether the last numbers (such as the last four numbers) or all numbers of the certificate number required for verification are the same. The last number of the registrant's card number or the digits of all the numbers that are the same as the correct phone number (door number), for example, the card number is a unified number, the digits of all numbers must be exactly the same (consistent). If they are the same, the process of the certificate number verification module 20 is ended, and the process of the name verification module 30 (name verification S3 ) shown in FIG. 5 is performed.

As shown in FIG. 5 , the authentication process of customer A enters the name verification module 30 (name verification S3 ), and the name verification module 30 will request customer A to input (speak) the name string through the voice of the electronic device, and the name (Name string) can be a name or a company name, etc. For example, customer A inputs (says) a name string (such as a name string) as "my name is Hong Yuxiang" through the voice of the electronic device, the name verification module 30 will remove the redundant word "my name" in the name string Yes" and keep the remaining name strings (such as name strings) "Hong Yuxiang".

Next, the name verification module 30 can convert the name string (such as the name string) into the name phonetic string "Factory ㄨ ㄥ ^ˊ ㄩ ^{ˇ ㄒ}一 ㄤ ^ˊ ", so as to remove the tone and retain the name phonetic string. The phonetic part of the phonetic string of the name is "Factory ㄨ ㄥ ㄩ ㄒ 一 ㄤ". The name verification module 30 can obtain the name standard answer string from the database 60 (eg, the backend database), and the name standard answer string is the name string of the registrant with the correct telephone number. For example, if the name standard answer string (correct result) is "Hong Yuxiang", the name verification module 30 can remove the tones and reserve the name standard from the name standard answer phonetic string "Factory ㄨ ㄥ ^ˊ ㄩ ^ˊ ㄒ一ㄤ ^ˊ " Answer the phonetic part of the phonetic string "Factory ㄨ ㄥ ㄩ ㄒ 1 ㄤ", and then calculate the similarity score of the name.

For example, the calculation formula of the similarity confidence score of each verification item (such as name or address) is the following formula [1] and formula [2].

門檻值 [3] Similarity Confidence Score

Threshold

In the above formula [1] and formula [2], the parameters of the calculation formula for the similarity confidence score of each verification item (such as name or address) are explained as follows: Word _all is the number of words in the standard answer, and Word _cor is the number of words in the standard answer. The correct number of words in the identification result (that is, the number of words entered by Customer A that is consistent with the standard answer; cor means the abbreviation of correct), bopomofo _juin is the correct proportion of the initial/final phonetic part of each word, and Word _ins is the identification The resulting extra words (ins stands for the abbreviation of insertion), juin _all is the number of phonetic characters that are split into phonetics to remove tones, and juin _cor is the phonetic number of characters that are split into phonetics to remove tones and the standard answer. The number of the same phonetic number of the phonetic phonetics when the characters are split into phonetics to remove the tones, and n is a positive integer.

In the above formula [3], the passing criteria of each verification item (such as phone number, card number, name, birthday or address) is that the similarity confidence score of each verification item must be greater than or equal to the threshold value to pass certified.

The name verification module 30 can also perform corresponding processing for ㄓㄔㄕ (warped tongue) and ㄗㄘㄙ (non-warped tongue) with the same pronunciation in phonetic, so as to solve the problem when customer A says (pronounces) the name , there are homonyms or spelling errors due to the same pronunciation. In addition, the name verification module 30 can use the processing method of phonetic symbols and the processing mechanism of grouping with characteristics of initials/finals, such as initials "ㄅㄆㄇㄈㄉㄊㄋㄌㄍㄎㄏㄐㄑㄒㄓㄔㄕㄖㄗㄘㄙ ” are grouped into the same group, the intermediate sound “一ㄨㄩ” is grouped into the same group, the finals “ㄚㄛㄜㄝㄞㄟㄠㄡㄢㄣㄤㄥㄦ” are grouped into the same group, this processing method or processing mechanism is suitable for the name In an infinite number of combinations of applications, in order to improve the identification accuracy and system availability.

From this embodiment, although "yu" and "yu" are not the same, they are both "ㄩ" when converted to Zhuyin and the tones are removed, so this embodiment is passed. The name verification module 30 uses the method of removing the tone, mainly to deal with the phenomenon of voice change when the pronunciation is a continuous character string and the speech speed is relatively fast, which will affect the words that appear in the voice recognition, such as "in "The tone is two (ㄩ ^ˊ ) and the "yu" tone is three (ㄩ ^ˇ) , which in turn affects the verification result of the name, so the name verification module 30 uses the method of removing the tone to process or solve the voice change. The impact of the phenomenon is conducive to improving the verification reliability.

As shown in FIG. 6 , the verification process of customer A enters the birthday verification module 40 (birthday verification S4 ), and the birthday verification module 40 will request customer A to input (speak) the birthday year through the voice method of the electronic device (speak) or by pressing the button. month day string. For example, if customer A chooses to input (speak) the birthday date string as "781016" through the voice of the electronic device, the birthday verification module 40 will generate the birthday date, date, and voice recognition result string through the voice recognition technology. After obtaining the numeric string of the date of birth, the birthday verification module 40 determines the length of the numeric string of the date of birth, and for the "year" of the date of birth, the format of the year of the first Western year and/or the year of the Republic of China is performed. Conversion, so you will get a format number string with two representations of the birthday date as "781016" and/or "19891016".

Next, the birthday verification module 40 will be linked to the database 60 (eg, a back-end database) to obtain a standard answer string of the date of birth (eg, the registrant's birthday digits of the correct phone number) from the database 60 string), remove the non-numeric words in the birthday date standard answer string, add punctuation and annotation symbols to the birthday date standard answer string, and keep the remaining birthday date standard answer string. Then, the birthday verification module 40 can calculate the similarity confidence value score of the birthday date to determine whether the obtained similarity confidence value score of the birthday date passes the set The threshold value is set, and when the similarity confidence score of the birthday date passes the threshold value, the verification program of the customer A enters the process of the address verification module 50 .

As shown in FIG. 7 , the verification process of client A enters the address verification module 50 (address verification S5 ), and the address verification module 50 requests client A to input (speak) the address string through the voice of the electronic device. For example, customer A inputs (says) the address string as "My address is No. 270, Fuguo Road, Taoyuan" through the voice of the electronic device, and the address verification module 50 will remove redundant words/redundancies in the address string" My home address is ", to keep the remaining address string "No. 270, Fuguo Road, Taoyuan". Next, the address verification module 50 can convert the address string into an address phonetic string, so as to remove the tones from the address phonetic string and reserve the phonetic part of the address phonetic string, such as "ㄊㄠㄩㄢㄕㄈㄨㄍ ㄨㄛㄌㄨㄌ一ㄤㄅㄞㄑ一ㄕㄏㄠ”.

Then, the address verification module 50 will be linked to the database 60 (such as a back-end database) to obtain the address standard answer string (correct address string) from the database 60, such as "270 Fuguo Road, Taoyuan City. Sixth Floor, No. 6". The address verification module 50 will expand the address standard answer string (correct address string) into all possible combinations (address combination strings), such as "6th Floor, No. 276, Fuguo Road, Taoyuan City" and "Fuguo, Taoyuan City" There are many possible combinations (address combination strings) such as the sixth floor, No. 276, Road. The address verification module 50 can remove the tones from the results of all possible combinations (address combination strings) and retain the phonetic parts of all possible combinations (address combination strings), such as "ㄊㄠㄩㄢㄕㄈㄨㄍ ㄨㄛㄌㄨㄌ一ㄤㄅㄞㄑone "Wait. After that, the address verification module 50 calculates the similarity confidence value score of the address, and then judges whether the obtained similarity confidence value score of the address passes the set threshold value.

In addition, the address verification module 50 can set the rules of the necessary address verification units (eg road, number, building), so as to extract the address necessary verification unit (eg road, number, building) from the address standard answer string. If the address verification module 50 determines that the similarity confidence score of the address passes the threshold, and the address verification module 50 recognizes the address string as a result of failing (complies with) one of the rules of the address necessary verification unit, it means that the address The verification of the address by the verification module 50 fails. On the contrary, if the address verification module 50 judges that the similarity confidence value score of the address passes the threshold value, and the address verification module 50 recognizes the address string as a result of passing (complies with) the rules of the address necessary verification unit, then the address verification module 50 Determines (deems) that the address string entered (speaked) by customer A through the voice of the electronic device and the address standard answer string are (considered) the same, and ends the process of the address verification module 50, and then checks the customer A's The certification procedure is directed to the corresponding service or customer service personnel B to accept the service.

In addition, the present invention also provides a computer-readable medium for a method for customer service, which is applied to a computing device or computer having a processor and/or memory, and stores instructions, which can be processed by the computing device or computer. The computer-readable medium is executed by the computer and/or memory so as to execute the above-mentioned content when the computer-readable medium is executed. For example, the processor can be a microprocessor, a central processing unit (CPU), a graphics processing unit (GPU), etc., and the memory can be a random access memory (RAM), a memory card, a hard disk (cloud/network hard disk) ), etc., but not limited thereto.

In conclusion, the authentication system, method and computer-readable medium for customer service of the present invention have at least the following features, advantages or technical effects.

1. The present invention can perform (automatic) verification on multiple verification items (such as phone number, certificate number, name, birthday or address) in advance, so as to reduce the average processing time (AHT) of customer service personnel, which is beneficial to reduce the number of customer service personnel. repetitive inspection tasks.

2. The name verification module of the present invention can perform phonetic comparison between the phonetic string of the name string input by the customer through the voice of the electronic device and the phonetic string of the corresponding standard answer of the name to calculate the similarity of the name Confidence score, which is helpful for solving the problem of Chinese homophones.

3. The present invention can develop an appropriate normalization method and data conversion method according to the data characteristics required by the verification project during verification.

4. The present invention can use phonetic symbols and remove tones for verification items (such as names or addresses), so as to facilitate the processing of homonyms, similar tones or phonetic tones of verification items (such as names or addresses) and other voice influencing factors to improve system availability and verification reliability.

5. The present invention can establish a mechanism of similarity confidence score and threshold value of verification items, so as to facilitate the judgment of the correctness of verification data and results.

6. The present invention can carry out modular classification according to the characteristics of the verification items, and cooperate with the actual verification process to develop the system service process, so that the service can be more in line with the actual application.

7. The present invention can be applied to the result of input by voice or key press of electronic devices, so as to facilitate (automatically) capture the required information in the string of verification items for data verification, thereby achieving voice-based or automated services.

8. In order to make the customer's operation more flexible, the present invention not only provides voice input, but also provides dual-tone multi-frequency (DTMF) button input, that is, the dual-track input function (ie, voice or button input), so that Customers can choose to verify by phone number verification, certificate number verification, and birthday verification through voice or key input on electronic devices.

The above-mentioned embodiments are illustrative of the principles, features and effects of the present invention, and are not intended to limit the applicable scope of the present invention. Modifications and changes can be made to the above-mentioned embodiments without departing from the spirit and scope of the present invention. Any equivalent changes and modifications made by using the contents disclosed in the present invention should still be covered by the scope of the patent application. Therefore, the protection scope of the present invention should be listed in the scope of the patent application.

1: Authentication system for customer service

10: Phone number verification module

20: Certificate number verification module

30:Name verification module

40: Birthday Verification Module

50:Address verification module

60:Database

70: Comparison table of redundant words / redundant words and punctuation marks

80: Comparison table of similar phonetic phonetics

A: Customer

B: customer service staff

Claims (19)

A verification system for customer service, comprising: a telephone number verification module, which extracts a digital string of a telephone number from a telephone number string input by a customer, so that the telephone number verification module can determine whether the telephone number is valid. The consistency between the number string and the incoming phone number, the phone number verification module asks whether the customer is the registrant of the phone number; and the name verification module is based on the phone number verification module When inquiring that the customer is the registrant of the phone number, the name verification module judges both the name string input by the customer and the name of the registrant of the phone number according to the similarity confidence score of the name The name verification module compares the phonetic character string of the name string input by the customer with the phonetic character string of the corresponding standard answer of the name, so as to base on the name string The result of the phonetic comparison between the phonetic string of the name and the phonetic string of the standard answer of the name calculates the similarity confidence score of the name. The authentication system of claim 1, wherein the phone number verification module normalizes the data format of the phone number to delete non-numeric words and retain words of the phone number in the character string of the phone number The part of the number in the string, and then the part of the number in the string of the phone number is converted into the format of Arabic numerals. The verification system according to claim 1, wherein the phone number verification module further increases the weight of digit appearance and narrows the combination range of the word group in the language model for the language model in the identification technology of the phone number, to make the language model appear the combination of the digit string of the phone number. The verification system according to claim 1, wherein the name verification module further calculates the name phonetic string and the name standard answer annotation according to the similarity confidence score of the name The similarity result between the phonetic character strings, so that the name verification module determines whether the similarity confidence score of the name passes the threshold value according to the similarity result between the phonetic character string of the name and the phonetic character string of the standard answer of the name. When the similarity confidence value score of the name passes the threshold value, the name verification module determines that both the name phonetic string of the name string input by the customer and the phonetic string of the name standard answer are consistent . The verification system according to claim 1, wherein the name verification module further matches the name string input by the customer with a comparison table of redundant words/redundant words and punctuation to remove redundant words in the name string The name string is converted into a name phonetic string by the name verification module, and the tones are removed from the name phonetic string and the phonetic pronunciation of the name phonetic string is retained. part. The verification system according to claim 1, wherein the name verification module further cooperates with a similar phonetic phonetic comparison table to process the phonetic phonetics of the phonetic string of the name or part of the phonetic string of the standard answer phonetic of the name. The verification system as described in claim 1 further includes a certificate number verification module, which extracts the certificate number digital string from the certificate number string input by the customer, and then judges the retrieval by the certificate number verification module. Take the consistency between the digital string of the certificate number and the certificate number of the registrant of the phone number. The verification system as described in claim 1 further includes a birthday verification module, which increases the weight of the appearance of numbers and narrows the combination range of the word group in the language model for the language model in the identification technology of birthday date, month, and date. To make the language model appear the combination of the numeric string of the birthday date. The authentication system as described in claim 1 further includes a birthday verification module, which extracts the digital string of the birthday date from the birthday date string input by the customer, and then uses the birthday date The part of the number string for the year is converted to the type of the year of the first year and/or the year of the Republic of China, Then, the birthday verification module determines whether the converted birthday date is the birthday of the registrant of the telephone number. The verification system according to claim 1, further comprises an address verification module, which is used to match the address string input by the customer with a comparison table of redundant words/redundant words and punctuation marks, so as to remove redundant words/ The redundant words and the remaining address string are reserved, and then the address verification module converts the address string into an address phonetic string, and then removes tones from the address phonetic string and retains the phonetic part of the address phonetic string . The verification system according to claim 1, further comprising an address verification module, which uses voice recognition technology to obtain a string of speech recognition results of the address from the address string input by the customer, so that the address verification module can Expand all possible combinations of the address standard answer string, and then the address verification module determines whether the address string input by the customer is a correct registered address using the mechanism of the similarity confidence score and the threshold value of the address. The verification system according to claim 1 further includes an address verification module, which is used to compare the address phonetic string of the address string input by the customer through the voice method with the phonetic string of the address standard answer, so as to obtain Calculate the similarity confidence value score of the address, and then judge whether the similarity confidence value score of the address passes the threshold value according to the rules of the address necessary review unit, so that when the similarity confidence value score of the address passes the threshold value, the The address phonetic string and the address standard answer phonetic string are determined to be consistent. A verification method for customer service, comprising: extracting a digital string of a phone number from a phone number string input by a customer by a phone number verification module, so that the phone number verification module can determine the digits of the phone number The consistency between the character string and the incoming phone number, and then the phone number verification module asks whether the customer is the registrant of the phone number; and When the phone number verification module inquires that the customer is the registrant of the phone number, the name verification module determines the name string input by the customer and the registrant of the phone number according to the similarity confidence score of the name. The consistency between the two names of my own, wherein the name verification module is to compare the phonetic character string of the name string entered by the customer with the phonetic character string of the corresponding standard answer of the name, to obtain The similarity confidence score of the name is calculated according to the result of the phonetic comparison between the name phonetic string of the name string and the phonetic string of the name standard answer. The verification method according to claim 13, further comprising: retrieving the digital string of the certificate number from the certificate number string input by the customer by the certificate number verification module, and then determining the retrieved by the certificate number verification module The number string of the certificate number is consistent with the certificate number of the registrant of the phone number. The verification method according to claim 13, further comprising increasing the weight of the appearance of numbers in the language model of the birthday verification module in the identification technology of the birthday date and restricting the combination range of the word groups in the language model, so as to Makes the language model appear the combination of the numeric strings of the birthday date. The verification method according to claim 13 further comprises: extracting the digital string of the birthday date from the birthday date string input by the customer by the birthday verification module, and then using the date of the birthday date. The part of the numeric string for the year is converted into the type of the first year of the Western year and/or the year of the Republic of China, and then the birthday verification module determines whether the converted birthday date is the birthday of the registrant of the phone number. The verification method according to claim 13, further comprising matching the address string input by the customer with a comparison table of redundant words/redundant words and punctuation marks by the address verification module to remove redundant words/redundant words in the address string word and retain the remaining address string, and then use the address verification module to determine the address The character string is converted into an address phonetic character string, and then the address phonetic character string is removed tones and the phonetic part of the address phonetic character string is reserved. The verification method according to claim 13, further comprising performing phonetic comparison between the address phonetic string of the address string input by the customer and the phonetic string of the address standard answer by the address verification module to calculate the similarity of the addresses Then, according to the rules of the address necessary review unit, it is judged whether the similarity confidence value score of the address passes the threshold value, so that when the similarity confidence value score of the address passes the threshold value, the phonetic character string of the address is combined with the threshold value. The phonetic strings of the standard answer of the address are determined to be the same. A computer-readable medium used in a computing device or a computer and storing instructions for executing the authentication method for customer service as described in any one of claims 13 to 18.

Images