JP2012014390A - Sentence generator, sentence generation method, and sentence generation program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reduce the number of key inputs to be required for acquisition of template information before starting authentication service to further reduce an input load and input time of a user in key stroke authentication.SOLUTION: A sentence generator 1 generating a sentence for acquiring the template information in the key stroke authentication includes: storage means 12 for storing a word dictionary in which at least one word is set for every part of speech, and a plurality of sentence structure patterns constituted of a plurality of parts of speech; sentence structure selection means 11 for selecting a sentence structure pattern from the storage means 12; and word selection means 11 for selecting a word from the word dictionary for every part of speech constituting the selected sentence structure pattern to generate a sentence. Each word set for every part of speech in the word dictionary is classified according to an initial character of the word to be registered, the word selection means 11 counts each word selected from the word dictionary for every initial character of the word, and selects a word of the initial character with the least number of counts from the word dictionary.

Description

  The present invention relates to a sentence generation technique for generating a sentence for acquiring template information in keystroke authentication.

  As an authentication technique for authenticating a user who has accessed a computer, keystroke dynamics (hereinafter referred to as “keystroke authentication”) has been studied. Keystroke authentication is one of biometrics (biometric authentication). A person's keystroke (computer keyboard typing) has a pattern unique to that person, and keystroke authentication uses this property for authentication. Keystroke authentication is described in Non-Patent Document 1, for example.

“Keystroke dynamics identification in atypical Japanese sentence input”, Toshiharu Samura, Haruhiko Nishimura, Transactions of the Institute of Systems, Control and Information Engineers, Vol.22, No.4, pp. 145-153, 2009

  In keystroke authentication, before performing an authentication service, it is necessary to previously register the feature amount of each user called template information on the authentication server side. In keystroke authentication, there are a case where an input sentence used for authentication is a fixed character string such as a password and a case where a non-standard free sentence is used.

  When an atypical sentence is used for authentication, the input character string at the time of authentication is an atypical (free) sentence. Therefore, it is necessary to retain keystroke characteristics regarding many characters and consecutive characters in the template information. Therefore, in order to obtain highly accurate template information, there is a problem that the user is required to input a large number of keyboards (thousands of characters or thousands of keys), and the user's input load is large and troublesome.

  That is, in keystroke authentication, authentication cannot be performed until template information is acquired, and the user cannot use the authentication service. Therefore, the user needs to input many keyboards in order to use the authentication service, which is a major obstacle until the actual authentication service is started.

  When the user performs keyboard input to obtain template information, there are a method in which the user freely inputs an arbitrary sentence and a case in which a predetermined sentence presented from the authentication server is input. In the method in which the user freely inputs arbitrary text, the user inputs it without considering what kind of information is necessary for generating the template information. The number of characters is much larger than the number of characters. Further, even when a predetermined sentence is presented from the authentication server, no consideration is given to reducing the user's input load.

  The present invention has been made in view of the above circumstances, and an object of the present invention is to further reduce the number of key inputs necessary for obtaining template information before starting an authentication service in keystroke authentication, and to reduce the input load of the user. And to reduce the input time.

  To achieve the above object, the present invention provides a sentence generation device for generating a sentence for acquiring template information in keystroke authentication, a word dictionary in which at least one word is set for each part of speech, and a plurality of words Storage means for storing a plurality of syntax patterns composed of parts of speech; syntax selection means for selecting any one of the syntax patterns from the storage means; and the word dictionary for each part of speech that constitutes the selected syntax pattern Word selection means for selecting a word from and generating a sentence, each word set for each part of speech in the word dictionary is classified and registered according to the first character of the word, the word selection The means counts each word selected from the word dictionary for each first character of the word, and selects the word of the first character having the smallest count from the word dictionary.

  The present invention is also a sentence generation method for generating a sentence for acquiring template information in keystroke authentication, wherein the computer includes a word dictionary in which at least one word is set for each part of speech, and a plurality of parts of speech. A storage unit that stores a plurality of syntax patterns composed of: a syntax selection step of selecting any one of the syntax patterns from the storage unit; and for each part of speech that constitutes the selected syntax pattern, from the word dictionary A word selection step of selecting a word and generating a sentence; and each word set for each part of speech in the word dictionary is classified and registered according to the first character of the word, and the word selection step Counts each word selected from the word dictionary for each first character of the word, and determines whether the first character with the smallest count is the word dictionary. To choose.

  The present invention is also a sentence generation program for generating a sentence for acquiring template information in keystroke authentication, which is executed by a computer, wherein the computer has a word dictionary in which at least one word is set for each part of speech. And a storage unit that stores a plurality of syntax patterns composed of a plurality of parts of speech, and the computer selects a syntax pattern from the storage unit, and the selected syntax pattern For each part of speech that constitutes a word part, and a word selection step of generating a sentence by selecting a word from the word dictionary, and each word set for each part of speech in the word dictionary is set to the first character of the word The word selection step selects each word selected from the word dictionary for each first character of the word. To count, to select a word count is the lowest first letter from the word dictionary.

  According to the present invention, in keystroke authentication, the number of key inputs necessary for obtaining template information before the start of an authentication service can be further reduced, and the user input load and input time can be further reduced.

1 is an overall configuration diagram of a keystroke authentication system according to an embodiment of the present invention. It is a figure which shows an example of a word dictionary and a syntax pattern. It is a flowchart which shows the production | generation process of an input presentation text. It is a figure which shows an example of the counter of the selected word.

  Hereinafter, embodiments of the present invention will be described.

  FIG. 1 is an overall configuration diagram of a keystroke authentication system according to an embodiment of the present invention.

  The keystroke authentication system of the present embodiment includes an authentication server 1 (sentence generation device) that performs keystroke authentication and a client 3. The authentication server 1 and the client 3 are connected by a network 9 such as the Internet.

  The authentication server 1 according to the present embodiment performs keystroke authentication and generates an input presentation text for allowing a user to input a keyboard (typing) in a registration phase for registering template information before starting the keystroke authentication.

  The illustrated authentication server 1 includes a text generation unit 11, a storage unit 12, a text storage unit 13, a template information registration unit 14, a template information storage unit 15, and an authentication unit 16.

  In order to obtain template information, the sentence generation unit 11 generates an input presentation sentence for allowing the user to perform keyboard input (typing) in the registration phase. The sentence generation unit 11 includes a syntax selection unit, a part of speech acquisition unit, and a word selection unit (not shown).

  The storage unit 12 stores a word dictionary and a plurality of syntax patterns for generating an input presentation sentence. The text storage unit 13 stores the input presentation text generated by the text generation unit 11.

  The template information registration unit 14 registers the keystroke feature amount transmitted from the client 3 in the registration phase in the template information storage unit 15 as template information. The template information is a feature amount indicating a keystroke feature unique to each user who uses the client 3.

  In the authentication phase, the authentication unit 16 compares and collates the keystroke feature quantity received from the client 3 with the template information stored in the template information storage unit 15 to calculate the similarity, and according to the similarity Authentication. For example, when the similarity is equal to or higher than a predetermined threshold, it is determined that the person is the person, and when the similarity is lower than the predetermined threshold, it is determined that the person is an unauthorized third party.

  The client 3 is a terminal used by a user (user) of the keystroke authentication system, and for example, a PC (Personal Computer) can be used. The client 3 shown in the figure receives the input presentation text from the authentication server 1 and outputs it to the output device such as a display, and the feature amount (keystroke information) from the keystrokes keyed by the user according to the input presentation text. And a keystroke information acquisition unit 32 that extracts and transmits to the authentication server 1.

  As the authentication server 1 and the client 3 described above, for example, a computer system including at least a CPU, a memory, and a storage device such as an HDD can be used. In this computer system, each function of each device is realized by the CPU executing a predetermined program loaded on the memory. For example, the functions of the authentication server 1 and the client 3 are executed by the CPU of the authentication server 1 in the case of the program for the authentication server 1 and the CPU of the client 3 in the case of the program for the client 3, respectively. Realized.

  FIG. 2 shows an example of a word dictionary and a syntax pattern stored in the storage unit 12.

  The word dictionary shown in FIG. 2 (a) is a table in which the horizontal axis is the part of speech and the vertical axis is 50 sounds (aiueo ...). That is, in the word dictionary, at least one word corresponding to the part of speech is registered for each part of speech, and each word registered for each part of speech is classified and registered according to the first character of the word. ing. In the example shown in the figure, each word set for each part of speech is registered by being classified into every first 50 characters (aiueo ...).

  FIG. 2B shows an example of a plurality of syntax patterns composed of a plurality of parts of speech.

  In this embodiment, as shown in FIG. 2, generation of input presentation text in Japanese will be described as an example. However, the present invention is not limited to Japanese and generates input presentation text in other languages such as English. It is good to do. For example, when generating an input presentation sentence in English, the word dictionary is classified and registered not for every 50 sounds but for every 26 characters of the alphabet. As for the syntax pattern, the part of speech is set in the word order according to the English grammar.

  Next, an input presentation text generation process performed by the text generation unit 11 of the authentication server 1 according to the present embodiment will be described.

  FIG. 3 is a flowchart showing the process of generating the input presentation text. First, the sentence generation unit 11 (syntax selection unit) selects one of the syntax patterns stored in the storage unit 12 (S1). For example, the sentence generation unit 11 (syntax selection unit) selects from the syntax pattern 1 in order.

  Then, the sentence generation unit 11 (part of speech acquisition unit) acquires the part of speech constituting the syntax pattern selected in S1 (S2). For example, when the syntax pattern selected in S1 is syntax pattern 1, first, “combined words” are acquired.

  Then, the sentence generation unit 11 (word selection unit) selects a word corresponding to the part of speech acquired in S2 from the word dictionary (S3). Here, the sentence generation unit 11 (word selection unit) selects a word from the word dictionary so that each of the 50 sounds is used in a balanced manner as much as possible in the generated input presentation sentence, and the selected word is selected. Store in memory.

  For example, each word selected from the word dictionary is counted for each first character of the word, and the word of the first character with the smallest count is selected from the word dictionary. In this embodiment, each word selected from the word dictionary is counted for every 50 sounds of the first character, and the word of the first character of 50 sounds with the smallest count is selected from the word dictionary.

  Specifically, the authentication server 1 holds a counter for every 50 sounds as shown in FIG. 4 in the storage unit 12, selects a word in S3, and then sets “1” as the value of the leading sound of the selected word. Is added. For example, when “that” of “combined verb” is selected from the word dictionary in S3, “1” is added to the count number of “a”. Note that the initial value of each sound of the counter is “0”.

  The counter shown in FIG. 4 has a minimum count number (minimum value) and a maximum count number (maximum value) for all sounds. The sentence generation unit 11 (word selection unit) updates the count value of the selected word, searches for the count number of all sounds, and updates the minimum count number and the maximum count number to the latest state. .

  Then, when selecting a word in S3, the sentence generation unit 11 (word selection unit) selects from the word dictionary the word having the sound with the smallest count number at the head. When there are a plurality of sounds having the minimum count number, they are selected at random. In addition, in the case of a part of speech in which there is no sound word having the minimum count number, a sound word having a count value obtained by adding “1” to the minimum count value is selected.

  If there is the next part of speech of the syntax pattern selected in S1 (S4: YES), the process returns to S2 to acquire the next part of speech, and the words corresponding to the part of speech are used uniformly for each of the 50 sounds as described above. As described above, a word is selected from the word dictionary and stored in a memory or the like (S3). As described above, when the processes of S2 and S3 are repeated by the number of parts of speech of the syntax pattern selected in S1, and words are selected for all parts of speech (S4: NO), the sentence generator 11 (word selector) The words selected and stored in the memory are stored and output to the sentence storage unit 13 in the selected order (S5).

  For example, when the syntactic pattern 1 shown in FIG. 2 (b) is selected in S1, a sentence such as “That horse runs very fast” is generated and stored as a sentence memory. Stored in the unit 13.

  The sentence generation unit 11 generates a sentence by randomly converting the tense of the selected verb (current, past, future) or stylistic (plain sentence, negative sentence, question sentence, command sentence) or the like. It is good.

  If there is the next syntax pattern (S6: YES), the process returns to S1, and the sentence generation unit 11 (syntax selection unit) selects the next syntax (for example, syntax pattern 2) from the syntax pattern stored in the storage unit 12. ) Is selected (S1). Then, as described above, a word is selected from the word dictionary for each part of speech and a sentence is generated and stored in the sentence storage unit 13 (S2 to S5). When there is no next syntax pattern (S6: NO), the sentence storage unit 13 stores an input presentation sentence composed of a plurality of sentences generated for each syntax pattern, and the process ends.

  After the processing shown in FIG. 3, the text generation unit 11 transmits the input presentation text stored in the text storage unit 13 to the client 3 via the network 9.

  The receiving unit 31 of the client 3 receives the input presentation text transmitted from the authentication server 1 and outputs it to an output device such as a display. The user refers to the input presentation text that has been output, and performs key input (typing) on the input presentation text using the keyboard.

  As a result, the keystroke information acquisition unit 32 of the client 3 acquires data related to keystrokes such as a key press time and a key release time for each input key (or each input character), and presses each key press time ( Time from key press to key release), the time from pressing the first key to pressing the next key in two consecutive keys, the time from releasing the first key to pressing the next key, Extracts keystroke features such as the time from when the first key is released until the next key is released. For the feature amount extraction, for example, as described in Non-Patent Document 1 and the like, the above average and standard deviation of the time are used.

  And the keystroke information acquisition part 32 transmits the extracted feature-value to the authentication server 1 with the user ID of the user who input the key.

  The template information registration unit 14 of the authentication server 1 registers the feature amount transmitted from the client 3 in the template information storage unit 15 together with the user ID as the template information of the user.

  In the authentication phase after registration of template information, the keystroke information acquisition unit 32 of the client 3 acquires data related to keystrokes when a user inputs an atypical (various) sentence using the keyboard. The amount is extracted and transmitted to the authentication server 1 together with the user ID. The authentication unit 16 of the authentication server 1 compares and collates the feature amount transmitted from the client 3 with the feature amount (template information) corresponding to the user ID stored in the template information storage unit 15 to determine the similarity. If the calculated similarity is equal to or greater than a predetermined threshold, it is determined that the user is a valid user, and if the similarity is less than the predetermined threshold, it is determined that the user is an unauthorized user.

  In the present embodiment described above, in keystroke authentication, the number of key inputs required for obtaining template information before the start of the authentication service can be reduced, and the user input load and input time can be further reduced.

  In other words, in the present embodiment, by selecting from the word dictionary the word of the first character that has the smallest count of the first character of each word that has already been selected, an input presentation sentence that causes the characters (each key) to be input uniformly is generated. Thus, the number of key inputs required for obtaining template information can be further reduced.

  Also, when generating input presentation sentences in Japanese, since the consonant-vowel sequence is important, the word dictionary is classified into 50 sounds and words are selected so that the 50 sounds are used evenly. Highly accurate template information can be acquired. Also, by mixing not only hiragana but also kanji in the word dictionary, the user can input shift keys and enter keys.

  When generating an input presentation sentence in English, the word dictionary is classified into 26 alphabets, the 26 alphabets are input uniformly, and a sentence in which a space key between words is inserted is generated. In the case of English sentences, since consonant-vowels are not continuous as in Japanese, words that do not overlap the same alphabet are selected and registered in the word dictionary.

  In addition, this invention is not limited to the said embodiment, Many deformation | transformation are possible within the range of the summary. For example, in the above-described embodiment, the authentication server 1 that performs keystroke authentication generates the input presentation text. However, the server that performs keystroke authentication and the text generation server that generates only the input presentation text (the text generation unit 11). The storage unit 12 and the sentence storage unit 13) may be separated.

  In addition, it is provided with a plurality of word dictionaries according to Japanese, English, user attributes, etc., selecting the optimum word dictionary according to the usage scene, and generating the input presentation sentence using the selected word dictionary Also good. Thereby, improvement of authentication accuracy can be expected.

DESCRIPTION OF SYMBOLS 1 Authentication server 11 Text production | generation part 12 Storage part 13 Text storage part 14 Template information registration part 15 Template information storage part 16 Authentication part 3 Client 31 Reception part 32 Keystroke information acquisition part 9 Network

Claims (6)

A sentence generation device that generates a sentence for acquiring template information in keystroke authentication,
A word dictionary in which at least one word is set for each part of speech, and storage means for storing a plurality of syntax patterns composed of a plurality of parts of speech;
Syntax selecting means for selecting any syntax pattern from the storage means;
Word selection means for selecting a word from the word dictionary and generating a sentence for each part of speech that constitutes the selected syntax pattern;
Each word set for each part of speech in the word dictionary is classified and registered according to the first character of the word,
The word selection unit counts each word selected from the word dictionary for each first character of the word, and selects a word of the first character having the smallest count from the word dictionary. . The sentence generation device according to claim 1,
Each word set for each part of speech in the word dictionary is classified and registered for every 50 sounds of the first character,
The word selection means counts each word selected from the word dictionary for every 50 sounds of the first character, and selects from the word dictionary the word of the first character of the 50 sounds with the smallest count. A sentence generator. A sentence generation method for generating a sentence for acquiring template information in keystroke authentication,
Computer
A word dictionary in which at least one word is set for each part of speech, and a storage unit that stores a plurality of syntax patterns composed of a plurality of parts of speech,
A syntax selection step of selecting any one of the syntax patterns from the storage unit;
For each part of speech that constitutes the selected syntax pattern, select a word from the word dictionary and generate a sentence; and
Each word set for each part of speech in the word dictionary is classified and registered according to the first character of the word,
The word selection step counts each word selected from the word dictionary for each first character of the word, and selects a word with the smallest first character from the word dictionary. . The sentence generation method according to claim 3,
Each word set for each part of speech in the word dictionary is classified and registered for every 50 sounds of the first character,
In the word selection step, each word selected from the word dictionary is counted for every 50 sounds of the first character, and the word of the first character having the smallest count is selected from the word dictionary. Sentence generation method. A sentence generation program for generating a sentence for acquiring template information in keystroke authentication executed by a computer,
The computer
A word dictionary in which at least one word is set for each part of speech, and a storage unit that stores a plurality of syntax patterns composed of a plurality of parts of speech,
In the computer,
A syntax selection step of selecting any one of the syntax patterns from the storage unit; and
Selecting a word from the word dictionary for each part of speech constituting the selected syntax pattern and generating a sentence; and
Each word set for each part of speech in the word dictionary is classified and registered according to the first character of the word,
The word selection step counts each word selected from the word dictionary for each first character of the word, and selects a word of the first character with the smallest count from the word dictionary. . A sentence generation program according to claim 5,
Each word set for each part of speech in the word dictionary is classified and registered for every 50 sounds of the first character,
In the word selection step, each word selected from the word dictionary is counted for every 50 sounds of the first character, and the word of the first character having the smallest count is selected from the word dictionary. Sentence generation program.

Images