JP2002014695A - Method and device for interactive personal identification - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method and a device for interactive personal identification in which the knowledge of a user is utilized to make the device simple, and convenient and the work load on the user is minimized. SOLUTION: When identification information of the user is inputted to a user information input section 2, a question section 4 generates an attribute value of the user corresponding to the attribute stored in a database 31 as a question by voice and a voice recognition section 6 voice-recognizes the attribute value responded by the user by voice. Then, the attribute value of the user is collated with each of the attribute values of all users for the questions stored in a database 61, a process to compute the evaluation values that represent the degree of certainty is performed with respect to each of the attribute values of the all users, as to the attribute response value of the user for plural attributes, evaluation values for plural attributes are computed and the identification of the user is judged based on the integrated results of the plural evaluation values.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an interactive personal authentication method and apparatus for authenticating a user by using a dialog with the user, and more particularly, to a user using a voice dialog with the user. The present invention relates to an interactive personal authentication method and apparatus for authenticating a user.

[0002]

2. Description of the Related Art A service provider using information and communication technology wants to provide services only to registered valid users. In this case, the service is improperly impersonated as a valid user. Since there is an intruder who wants to receive the service, the service provider needs an identity authentication method for determining whether or not the user requesting the service is the identity.

[0003] The most widely used personal authentication method for authenticating the identity of a user is that the password requested to be input to the user is determined to be that of the user to be authenticated. Only the password input method that permits the use of the system is generally used, but this simple password authentication method is very weak against attacks such as peeping and eavesdropping, and the password can be peeped once by others. There is a danger of being impersonated by this person.

[0004] In recent years, there has been proposed a personal authentication system using physical characteristics of a person, such as fingerprints, voice prints, and facial features, that cannot be impersonated by others. This personal authentication method using physical characteristics is expected to be more difficult to impersonate than the password method, but such fingerprints, voice prints, and face recognition have also been developed with speech synthesis technology and image synthesis technology. As a result, it is becoming possible to copy voiceprints or fingerprints of others, and impersonation by copying is becoming possible.

Further, the above-mentioned password input, voice print,
In the personal authentication method using a fingerprint or the like, the more strict security is assured, the more complicated and difficult the operation of the user becomes. For example, the user is required to input a plurality of times.

In recent years, in order to prevent spoofing of passwords, voiceprints, fingerprints, and the like, an interactive personal authentication method for authenticating the identity of a user by interacting with the user by voice using the brain of the user has been proposed. Have been. Since this interactive personal authentication method uses the brain of the user,
Although it is highly durable against impersonation by others, it is considered that there is a limit because there are many parts that rely on the user's memory, such as the use of passwords.

[0007] Voice input is an input method that is the closest to the user and does not require practice for operation. However, there is no guarantee that voice recognition technology for recognizing input voice can always obtain 100% accuracy. However, the accuracy varies depending on the use environment and use conditions of the speech recognition device. For this reason, in order to accurately grasp the response from the user, it is necessary to present the recognition result to the user and make a correctness check until a correct response is obtained. Is considered to be a burden for the user.

[0008]

Among the above-mentioned conventional personal authentication systems, the password-based personal authentication system is extremely weak against attacks such as peeping and eavesdropping. There is a problem of being impersonated by another person.

In addition, the personal identification method using the physical characteristics of the user, such as fingerprints, voiceprints, and facial features, makes it possible to copy voiceprints and fingerprints of other people with the development of voice synthesis technology and image synthesis technology. There is a problem that impersonation by copying is possible.

[0010] Further, in a personal identification method using a password input, a voice print, a fingerprint, or the like, in order to assure stricter security, if the user is required to input a plurality of times, the operation of the user becomes difficult. There is a problem that it becomes complicated and difficult.

[0011] The interactive personal authentication system using voice communication with the user has a problem because many parts rely on the memory of the user, such as the use of a password.

Further, in the interactive personal authentication system, there is always one voice recognition technology for recognizing voice input from a user.
There is no guarantee that the accuracy of 00% can be obtained, and the accuracy varies depending on the use environment and use conditions of the voice recognition device.
In order to accurately grasp the response from the user, it is necessary to present the recognition result to the user and make a correctness check until a correct response is obtained. There is a problem that the user is burdensome.

The present invention has been made in view of the above,
The purpose is to use the knowledge of the user himself,
An object of the present invention is to provide an interactive personal authentication method and apparatus using an interaction that is considered to be simplest, convenient, and minimal in burden for the user.

[0014]

According to one aspect of the present invention, there is provided an interactive personal authentication method for authenticating a user using a dialog with the user. The attribute value of each user for each of the attributes is stored in the database, and when the identification information of the user to be authenticated is input, the attribute value of the user for the attribute stored in the database is stored. A question is asked to the user, an attribute value that the user has responded to the attribute of the question is received, and the attribute response value of the user is stored in the database as the attribute value of all the users for the attribute of the question. In comparison with each, the attribute response value of the user, an evaluation value representing the likelihood for each of the attribute values of all users is calculated, and the question is asked for each of a plurality of attributes stored in the database. One In the same manner, the respective evaluation values of the attribute response values from the user with respect to the plurality of attributes are similarly calculated, and the identity of the user is determined based on the integration result of the plurality of evaluation values with respect to the calculated plurality of attributes. The point is to do.

According to the first aspect of the present invention, when the identification information of the user is input, the user is inquired about the attribute value of the user for the attribute stored in the database,
Receives the attribute value that the user responded to for the attribute of this question, compares the attribute response value of this user with each of the attribute values of all users for the attribute of the question stored in the database, The process of calculating an evaluation value indicating the likelihood of the attribute response value of each user for each of the attribute values of all users is performed for each of the plurality of attributes stored in the database. In the same manner, the evaluation value of each attribute response value of the user is calculated, and the identity of the user is determined based on the integration result of the plurality of evaluation values for the calculated attributes. Ask multiple knowledge-based questions to prevent spoofing, ensure greater security, and answer knowledge-based questions And from, it does not matter even such as a storage force.

The present invention according to claim 2 provides the invention according to claim 1.
In the invention described above, the question is asked to the user by voice, and the reception of the attribute value responded to by the user by the question is performed by voice-recognizing the voice from the user. I do.

According to the second aspect of the present invention, the question is given to the user by voice, and the user responds to the question by voice, and the voice is recognized by voice. Without feeling burdened and difficult to operate,
It can be carried out.

Furthermore, the present invention according to claim 3 is an interactive personal authentication apparatus for authenticating a user by using a dialog with the user, wherein each user's attribute value for each of a plurality of attributes. And question means for asking the user about the attribute value of the user for the attribute stored in the database when identification information of the user to be authenticated is input, and the attribute of the question Response receiving means for receiving an attribute value responded by the user to the user, and comparing the received attribute response value of the user with each of the attribute values of all users for the attribute of the question stored in the database. Evaluation value calculation means for calculating an evaluation value indicating the likelihood of the attribute response value of the user with respect to each of the attribute values of all users; For each of the attributes, and similarly calculate the respective evaluation values of the attribute response values from the user for the plurality of attributes. Based on the integrated result of the plurality of evaluation values for the calculated plurality of attributes, the user's The gist is to have personality judgment means for judging personality.

According to the third aspect of the present invention, when the identification information of the user is input, the user is inquired about the attribute value of the user for the attribute stored in the database.
Receives the attribute value that the user responded to for the attribute of this question, compares the attribute response value of this user with each of the attribute values of all users for the attribute of the question stored in the database, The process of calculating an evaluation value indicating the likelihood of the attribute response value of each user for each of the attribute values of all users is performed for each of the plurality of attributes stored in the database. In the same manner, the evaluation value of each attribute response value of the user is calculated, and the identity of the user is determined based on the integration result of the plurality of evaluation values for the calculated attributes. Ask multiple knowledge-based questions to prevent spoofing, ensure greater security, and answer knowledge-based questions And from, it does not matter even such as a storage force.

According to a fourth aspect of the present invention, in the third aspect of the present invention, the question means has a voice output means for asking a user by voice, and the response receiving means comprises a It is essential to have a voice recognition means for recognizing a voice responded to by the user.

According to the fourth aspect of the present invention, the question is given to the user by voice, and the user responds to the question by voice, and the voice is recognized by voice. Without feeling burdened and difficult to operate,
It can be carried out.

[0022]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a configuration of an interactive personal authentication device according to an embodiment of the present invention. The interactive personal authentication device 1 shown in FIG. 1 asks a user by voice a question that depends on the knowledge of the user who seems to know only the user, not the memory of the user. Is an interactive personal authentication device that authenticates a user as a true person only when a response obtained by voice from the user is correct, and a user-specific user I that can uniquely identify the user.
It has a user information input unit 2 for inputting D as user information from the user.

Further, the interactive personal authentication apparatus of FIG. 1 recognizes a personal information database 31 storing attribute values of each user for each of a plurality of attributes, and recognizes attribute values of all users for each attribute. It has a recognition target database 61 stored as a target. In addition, the principal information database 3
1 and the storage of data in the recognition target database 61 are performed, for example, at the time of user registration.

The personal information database 31 is, for example, as shown in FIG.
, Information 1, information 2, which is a plurality of pieces of user information for each of a plurality of attributes 1, attribute 2,.
.., And each attribute value of information n is stored. More specifically, as shown in FIG. 4, the personal information database 31 has a plurality of attributes such as "first name and last name", "affiliation",... "Hobbies", and user information. There are user IDs 1, 2, 3,..., N for uniquely identifying users, and the attribute values of each user identified by the user ID are stored for each of a plurality of attributes. For example, for a user with user ID = 1,
Attribute values such as "Yamada Hanako", "Secretary Section" ... "tennis" are stored for the respective attributes of "first name, last name", "affiliation" ... "hobby".

As shown in FIG. 3, for example, the recognition target database 61 stores attributes 1, 2, 3, and 3 to be recognized.
, User information 1, 2, 2,.
.., N, attribute values of all users are stored. More specifically, as shown in FIG. 5, the recognition target database 61 includes a plurality of attributes “first and last name”
"Department" ... User ID1 for each "hobby"
Attribute values of all users identified by 2,..., N are stored. For example, for the attribute “first and last name”,
"Yamada Hanako" as attribute value of all users of 2, ..., n
"Suzukitarou" ... "Kumiko Oomori" is stored.

As described above, the user ID, which is the user information input from the user information input unit 2, is input to the user information extraction unit 3. The user information extracting unit 3 searches the personal information database 31 based on the user ID, extracts and holds attribute values for each of the plurality of attributes corresponding to the user ID, and stores the extracted plurality of attribute values. The attribute is supplied to the question section 4. The question section 4 asks the user by voice via the voice output device 10 a question asking the attribute value for each of the plurality of attributes.

When the user responds by voice to the question asked by the question section 4 by voice, the attribute response value is received by the voice input section 5 and supplied to the voice recognition section 6. The voice recognition unit 6 sets the attribute values of all the users for the attribute question from the recognition target database 61 for the attribute response values from the user as recognition targets, and performs voice recognition using the voice recognition device 11. The speech recognition result is supplied to the recognition result output unit 7.

The recognition result output unit 7 supplies the recognition result of the attribute response value of the user to be authenticated to the recognition result integrating unit 8 as personal identification likelihood. The recognition result integrating unit 8 uses a plurality of recognition results of a plurality of attribute response values of the user with respect to a plurality of attributes of the question asking the attribute values for each of the plurality of attributes in the question unit 4 as personal identification likelihoods. Consolidate,
Update and maintain the personal identification integration degree. As a method of calculating the personal identification likelihood, for example, there is a method of using the value of the identification likelihood itself, or weighting the identification likelihood according to the importance of the question. As an example of a method of integrating personal identification likelihoods, for example, four arithmetic operations using a combination of addition, subtraction, and multiplication are given. It is needless to say that the attribute value for the first attribute is not integrated.

Then, at the point in time when all questions for a plurality of attributes have been completed, the recognition result integrating unit 8 supplies the integrated result of the personal identification likelihood to the personal authentication unit 9. The personal authentication unit 9
The personal identification likelihood from the recognition result integrating unit 8 is compared with a predetermined threshold. If the personal identification likelihood is larger than the threshold, it is determined that the user is a person.

This will be described more specifically. For example, when the user ID “1” is input from the user information input unit 2 as the user information, the user information extraction unit 3 sends a plurality of users corresponding to the user ID “1” from the personal information database 31. The attribute values for the attribute are "Yamada Hanako" for the attribute "First and last name", "Secretary section" for the attribute "Department",
... "Tennis" is extracted for the attribute "hobby" and supplied to the inquiry section 4.

The query section 4 first asks the user the attribute value for the first attribute "first and last name" by voice. The attribute value to which the user responded by voice to the question, for example, “Yamada Hanako” is received by the voice input unit 5 and supplied to the voice recognition unit 6. The voice recognition unit 6 retrieves the attribute value “Yamada Hanako” of the attribute value of all users for the question attribute “first and last name” from the recognition target database 61 for the attribute value responded by the user.
"Suzukitaro" ... "Kumiko Oomori" is set as a recognition target, a voice recognition process is performed, and this voice recognition result is supplied to a recognition result output unit 7.

The recognition result output unit 7 supplies the recognition result of the attribute value for the user to be authenticated to the recognition result integrating unit 8 as personal identification likelihood. As an example of a method of outputting a recognition result, as shown in FIG. 6, the recognition result output unit 7 outputs the first and last names of all the users in the order of the highest likelihood of identification, “Yamada Hanako”, “Yamaguchi Hana”, “Yamaharu Haruko” ... "Kumiko Omori" is output side by side. In the present embodiment, for example, the value of the identification likelihood output as the personal identification likelihood is used as it is. That is, the recognition result output unit 7 outputs the user ID
The identification likelihood 1.8 of the attribute value “Yamadanako” for the attribute “first and last name” of the user having “1” is sent to the recognition result integrating unit 8 as the personal identification likelihood. The recognition result integrating unit 8 holds the transmitted personal identification likelihood 1.8.

As described above, the question for asking the attribute value for the attribute “first and last name” is completed, and the identification identification likelihood for this question is held in the recognition result integrating unit 8. When a question asking a value is similarly made to the user and the likelihood of identification of “secretary section” is similarly obtained as an attribute value to which the user responds to the question attribute “department”, this identification The likelihood is sent to the recognition result integrating unit 8, and integrated into the personal identification likelihood 1.8 held earlier in the recognition result integrating unit 8. As an example of the method of integrating the personal identification likelihood, there is a method of multiplying all likelihoods.

Assuming that a method of multiplying all likelihoods is used as an example of a method of integrating personal identification likelihoods, the recognition result integrating unit 8 holds the identification likelihood of the attribute value “secretary section” first. The calculated identification likelihood 1.8 is multiplied to calculate a new identification likelihood, and this new personal identification likelihood is held.

The above operation is repeated until all the questions are completed. When all the questions are completed, the integrated result of the identification likelihood held in the recognition result integrating unit 8 is supplied to the personal authentication unit 9. . Then, the personal authentication unit 9 compares the integration result of the personal identification likelihood from the recognition result integration unit 8 with a predetermined threshold value, and when the integrated value of the personal identification likelihood is larger than the predetermined threshold value, Is determined, and when it is small, it is determined that the user is not himself.

As described above, in the present embodiment, the operation is the simplest for the user, and the input by the convenient voice is realized, so that the impossibility to the operation of the user can be minimized. Further, in the present embodiment, the ambiguity of the recognition accuracy is covered by inquiring the fact based on the knowledge of the user a plurality of times and comprehensively using the recognition results of all the responses obtained from the user. And the likelihood of identity.

In the above embodiment, a question to the user is made by voice using the voice output device 10, and a voice response from the user is recognized by the voice recognition device 11. The output of such a question and the input of a response to the question are not limited to those that are performed by voice. The means for outputting a question and the means for inputting a response to the question may be other than voice. . In particular, means for inputting a response may be input using a keyboard or the like in addition to voice.

[0038]

As described above, according to the present invention,
Ask the user the attribute value of the user for the attribute stored in the database, receive the attribute value that the user responded to the attribute of this question, and store the attribute response value of this user in the database. Stores in the database the process of calculating the evaluation value indicating the likelihood of the attribute response value of the user for each of the attribute values of all users by comparing with the attribute values of all users for the attribute of the question being asked Is performed for each of the plurality of attributes, and the respective evaluation values of the attribute response values from the user for the plurality of attributes are similarly calculated, and the integrated result of the plurality of evaluation values for the calculated plurality of attributes is obtained. Since the identity of the user is determined based on the user's knowledge, asking multiple questions based on the user's knowledge of the attribute value of the user can prevent impersonation and provide stronger security. It is possible to secure the tee, because it is intended to answer questions based on the knowledge,
There is no question about memory.

Further, according to the present invention, a question is given to the user by voice, and the user responds to the question by voice, and the voice is recognized by voice. You can do it without feeling or feeling difficult.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of an interactive personal authentication device according to an embodiment of the present invention.

FIG. 2 is a diagram showing a configuration of a personal information database used in the interactive personal authentication device shown in FIG.

FIG. 3 is a diagram showing a configuration of a recognition target database used in the interactive personal authentication device shown in FIG. 1;

FIG. 4 is a diagram showing a specific example of the personal information database shown in FIG. 2;

FIG. 5 is a diagram showing a specific example of a recognition target database shown in FIG. 3;

FIG. 6 is a diagram showing an example of a recognition result output by the interactive personal authentication device shown in FIG. 1;

[Explanation of symbols]

 2 User Information Input Unit 3 User Information Extraction Unit 4 Question Unit 5 Voice Input Unit 6 Speech Recognition Unit 7 Recognition Result Output Unit 8 Recognition Result Integration Unit 9 Personal Authentication Unit 10 Voice Output Device 11 Voice Recognition Device 31 Identity Information Database 61 Recognition target database

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification symbol FI Theme coat ゛ (Reference) G10L 3/00 571U

Claims (4)

[Claims] 1. An interactive personal authentication method for authenticating a user by using a dialog with the user, wherein an attribute value of each user for each of a plurality of attributes is stored in a database, and the authentication is performed. When the identification information of the target user is input, the user is queried about the attribute value of the user for the attribute stored in the database, and the attribute value that the user responded to the attribute of the question And comparing the attribute response value of the user with each of the attribute values of all users with respect to the attribute of the question stored in the database to obtain the attribute response value of the user, An evaluation value representing the likelihood for each of the values is calculated, and the question is asked for each of the plurality of attributes stored in the database, and the attribute response values from the user for the plurality of attributes are calculated. Interactive authentication method characterized by similarly calculating an evaluation value of, respectively, to determine the identity of the user based on the integration results of the plurality of evaluation values for a plurality of attributes this calculation. 2. The method according to claim 1, wherein the question is given to the user by voice, and the attribute value received by the user in response to the question is received by voice recognition of the voice from the user. The interactive personal authentication method according to claim 1. 3. An interactive personal authentication device that authenticates a user by using a dialogue with a user, comprising: a database storing attribute values of each user for each of a plurality of attributes; Question means for inquiring the user of the attribute value of the user with respect to the attribute stored in the database when identification information of a certain user is input; and an attribute responding to the attribute of the question by the user. Response receiving means for receiving the attribute value of the user; comparing the received attribute response value of the user with each of the attribute values of all users for the attribute of the question stored in the database; Evaluation value calculating means for calculating an evaluation value representing the likelihood of each of the attribute values of all users; and performing the question for each of the plurality of attributes stored in the database. Calculate the respective evaluation values of the attribute response values from the user for the plurality of attributes in the same manner, and determine the user's identity based on the integration result of the plurality of evaluation values for the calculated attributes. An interactive personal authentication device comprising: a determination unit. 4. The question means has voice output means for asking a user by voice, and the response receiving means voice-recognizes a voice of a user responding to the question. 4. The method according to claim 3, further comprising:
An interactive personal authentication device as described.