Program, determination system and determination method
The AI-based program generates a correlation model to determine if multiple actors performing similar actions are the same person, addressing confusion in application forms by enhancing accuracy and correcting identity verification errors.
Patent Information
- Application Number
- JP2023209704
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2023-12-12
- Publication Date
- 2025-06-24
- Estimated Expiration
- 2043-12-12
Smart Images
Figure 2025093811000001_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a program, a determination system, and a determination method.
Background Art
[0002] In daily life, we fill out various application forms with prescribed items to go through procedures. For example, when moving, we submit transfer notifications, transfer-in notifications, etc. to administrative agencies, and when a child is born, we submit a birth notification (see Non-Patent Document 1), an application form for maternity and child-rearing support benefits, an application form for maternity and childcare lump-sum benefits, etc. to administrative agencies. And depending on the circumstances of the application, there may be cases where multiple types of application forms are submitted at once.
[0003] Among multiple types of application forms submitted to administrative agencies, etc., even if the respective application items have the same content, their names (item names) may be different from each other. "Surname", "Name", and "Your name" are examples of this. Therefore, in a case where multiple types of application forms are submitted at the same time, when one has to be filled in the "Surname" column in one application form and the other has to be filled in the "Your name" column in the other application form, that is, when there is no consistency in the names of the two application forms, the applicant may be confused (and the associated burden) for this reason. Also, in a case where multiple types of application forms are submitted at the same time, when there are the same or same-content item names in both application forms, that is, when the same entry needs to be filled in multiple types of application forms, the applicant is forced to bear a burden for this reason.
Prior Art Documents
Non-Patent Documents
[0004]
Non-Patent Document 1
Summary of the Invention
Problems to be Solved by the Invention
[0005] Incidentally, through their recent research activities, the inventors of the present application have created a program for reducing the above burden by generating a correlation model of the semantic identity between a plurality of similar words using AI (artificial intelligence). Separately from this program, the inventors of the present application have succeeded in generating a correlation model related to an action using AI based on action information obtained by a plurality of persons performing actions such as application, purchase, and other actions. And, by using this correlation model related to the action, it was possible to obtain the knowledge that it is possible to determine with a certain probability whether a certain actor has performed the action in the past.
[0006] The present invention has been created in view of the above problems, and one of its purposes is to provide a program capable of determining that two actors are the same person when they perform the same or similar actions as the actions performed by one actor in the past.
Means for Solving the Problems
[0007] The program according to the first aspect is a program for determining that two actors are the same person when they perform the same or similar actions as the actions performed by one actor in the past, causing a computer to a storage function for storing action information including the content of the action and the time of the action for each actor every time the same or similar action is performed by a plurality of actors including the one actor; a generation function for performing machine learning on the correlation relationship between the plurality of action contents and the times of those actions from the plurality of pieces of action information stored by the storage function as teacher data to generate a correlation model of the action information; a determination function for determining, based on the correlation model, that the one actor is the same person as the actor who performed the same or similar action in the past when the one actor performs the same or similar action; to execute.
[0008] The program according to the second aspect is In the program according to the first aspect, "Based on the correlation model" means that it depends on whether the same similar behavior matches any of the behaviors specified by the correlation model and the similar behaviors of that behavior.
[0009] The program of the third aspect is In the program of the second aspect, In the storage function, the behavior information is stored separately by behavior type, In the generation function, from the plurality of pieces of behavior information stored separately by behavior type by the storage function, the correlation relationship between the plurality of behavior contents and the timing of those behaviors is machine-learned using teacher data, and a correlation model of the behavior information for each behavior type is generated.
[0010] The program of the fourth aspect is In the program of any one of the first to third aspects, The computer is caused to have a reading function for reading the identity document of each actor, and execute The storage function stores the information of the identity document of each actor read by the reading function when each actor performs an action for the first time.
[0011] The program of the fifth aspect is In the program of the fourth aspect, The computer is caused to after a negative judgment is made during the execution of the judgment function, read the identity document of the actor who is the subject of the negative judgment, and have a re-judgment function for judging whether the actor who is the subject of the negative judgment is the same person as the actor who has performed the same similar behavior in the past, and when an affirmative judgment is made during the execution of the re-judgment function, have a correction function for correcting the behavior information of the actor who is the subject of the negative judgment stored by the storage function. and execute.
[0012] The judgment system of the first aspect is A computer having a memory unit and a processing unit, storing a program of any one of the first to third aspects in the memory unit, and processing the program by the processing unit to execute the memory function, the generation function, and the determination function, and An interface that is communicably connected to the computer and acquires the plurality of pieces of behavior information stored by the memory function from each actor, and Comprising.
[0013] The determination system of the second aspect is A computer having a memory unit and a processing unit, storing a program of the fifth aspect in the memory unit, and processing the program by the processing unit to execute the memory function, the generation function, the determination function, the reading function, the re-determination function, and the correction function, and An interface that is communicably connected to the computer and acquires the plurality of pieces of behavior information stored by the memory function from each actor, and Comprising.
[0014] The determination method of the first aspect is A determination method for determining that two actors are the same person when they perform the same or similar actions as the actions performed by one actor in the past, using a program of any one of the first to third aspects, The step of the computer executing the memory function, and The step of the computer executing the generation function, and The step of the computer executing the determination function, and Including.
[0015] The determination method of the second aspect is A determination method for determining that two actors are the same person when they perform the same or similar actions as the actions performed by one actor in the past, using a program of the fifth aspect, The step of the computer executing the memory function, and The step of the computer executing the generation function, and The step of the computer executing the determination function, and The step of the computer executing the reading function; The step of the computer executing the re-judgment function; The step of the computer executing the correction function; It includes.
Effect of the Invention
[0016] According to the programs of the first and second aspects, when an actor performs the same or similar act as an act performed in the past, it is possible to determine whether both actors are the same person.
[0017] According to the program of the third aspect, compared with the mode of storing behavior information without classifying it into behavior types, it is possible to determine with high accuracy whether they are the same person.
[0018] According to the program of the fourth aspect, it is possible to generate a correlation model of behavior information by using the information of the identity document.
[0019] According to the program of the fifth aspect, while enabling correction by an actor who was negatively judged at the time of determining the same person in the past, it is possible to generate a correlation model updated with the information accompanying the correction.
[0020] According to the determination system of the first aspect, when an actor performs the same or similar act as an act performed in the past, it is possible to determine whether both actors are the same person.
[0021] According to the determination system of the second aspect, while enabling correction by an actor who was negatively judged at the time of determining the same person in the past, it is possible to generate a correlation model updated with the information accompanying the correction.
[0022] According to the determination method of the first aspect, when an actor performs the same or similar act as an act performed in the past, it is possible to determine whether both actors are the same person.
[0023] According to the determination method of the second aspect, while enabling a correction by an actor who was negatively determined in the past by the same person, it is possible to generate a correlation model updated with the information accompanying the correction.
Brief Description of Drawings
[0024]
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Modes for Carrying Out the Invention
[0025] Hereinafter, this embodiment and a plurality of modification examples thereof will be described in the order of these descriptions.
[0026] ≪This Embodiment≫ Hereinafter, the configuration and functions, determination method, and effects of the determination system JS of this embodiment will be described in the order of these descriptions.
[0027] <Configuration and Functions of Determination System> As shown in FIG. 1, the determination system JS includes a user interface 10 (hereinafter referred to as UI10), a server 20 (an example of a computer), and a communication network 30 (the Internet is an example thereof). The judgment system JS operates on a server 20 that is communicably connected to the UI10 via a communication network 30, and uses the behavior information of a plurality of actors U obtained by the UI10 to determine whether an actor U1 (actor U1 means any actor, that is, a single actor. The same applies hereinafter) is the same person as an actor who has performed the same or similar behavior in the past. Although the server 20 is taken as an example of a computer, the combination of the server 20 and the UI10, or the hardware constituting the judgment system JS may be regarded as an example of a computer.
[0028] Here, the meaning of the coined terms used in this embodiment will be described. ======================================== "Behavior information" Behavior information is information that includes (1) the personal information of actor U1 and (2) the content of the behavior performed by actor U1 and the time of that behavior. The personal information of actor U1 in (1) refers to, for example, information recorded on a personal information card (not shown) such as name, gender, age, address, blood type, etc. An example of a personal information card is an identity document (not shown) such as a My Number card or an identity certificate. The content of the behavior performed by actor U1 in (2) means the behavior performed by actor U1 in relation to facility F. For example, if facility F is a library, the content of the behavior is borrowing books, borrowing CDs, etc. Also, for example, if facility F is a government office, the content of the behavior is administrative procedures, applications, etc. The time of the behavior in (2) means the time when the target behavior was performed. "Same or similar behavior" Same or similar behavior means a combination of the behavior performed by actor U1 and a behavior similar to the behavior performed by actor U1. Here, a similar behavior refers to a behavior included in the similar range for the same behavior. For example, if the same behavior is borrowing a book, the similar range includes purchasing a book. Also, for example, if the same behavior is purchasing vegetables, the similar range includes purchasing fruits. ========================================
[0029] [[User Interface (UI)]] UI10 is, for example, arranged at the entrance of facility F, and has a function of interacting with actor U who plans to perform an action in facility F to obtain the identity confirmation of actor U and the action information of actor U (see FIGS. 1 and 4). The UI10 shown in FIGS. 1 and 4 is a humanoid robot capable of interacting with actor U, but the form of UI10 is not limited as long as it can exhibit the above functions. It may be a simple display with a touch panel (not shown). The specific operation of UI10 will be described later.
[0030] [[Server and Communication Network]] As described above, server 20 is communicably connected to UI10 via communication network 30 as shown in FIG. 1. Server 20 has a function (see FIG. 3) of executing a storage function, a generation function, a judgment function, a reading function, a re-judgment function, and a correction function (hereinafter referred to as a plurality of functions of the present embodiment) by a program PG (see FIGS. 2 and 3) stored in itself.
[0031] Server 20 has, as shown in FIG. 2, a processing unit 22 (CPU (Central Processing Unit) is an example thereof) and a storage unit 24 (RAM (Random Access Memory) is an example thereof). Processing unit 22 reads the data stored in storage unit 24 and performs arithmetic processing. The data includes, for example, application AP. Application AP is composed of program PG, data file DF, and various libraries LB.
[0032] [[Program]] The program PG is used to determine whether the two actors are the same person when the actor U1 performs the same or similar act as the act performed in the past. And, in order to achieve this purpose, the program PG has a function of causing the server 20 to execute a plurality of functions of the present embodiment described above.
[0033] Hereinafter, a plurality of functions of the present embodiment will be described. Hereinafter, each function constituting the plurality of functions will be described, but each function exhibits an organic technical meaning as a whole by cooperating with each other. Regarding this point, refer to the description of the determination method of the present embodiment described later.
[0034] (Storage function) The storage function is (1) a function of storing behavior information including the behavior content of each actor U and the time of the behavior every time the same or similar act is performed by a plurality of actors U including the actor U1, and (2) a function of storing information of the personal information card (not shown) of each actor U read by the reading function described later when each actor U performs an act for the first time. These pieces of information are stored in the data file DF constituting the storage unit 24.
[0035] (Generation function) The generation function is a function of performing machine learning on the correlation relationship between a plurality of behavior contents and the times of those behaviors from the plurality of behavior information stored by the storage function as teacher data, and generating a correlation model of the behavior information.
[0036] (Judgment function) The judgment function is a function of judging whether the actor U1 is the same person as the actor who performed the same or similar act in the past based on the correlation model when the actor U1 performs the same or similar act as the act performed in the past. Here, "based on the correlation model" means depending on whether the same or similar act described above matches any of the acts specified by the correlation model and similar acts of the act.
[0037] (Reading function) The reading function is a function that reads personal information stored in the personal information card (not shown) of each actor U.
[0038] (Rejudgment function) The rejudgment function is a function that, after a negative judgment is made during the execution of the judgment function, causes the personal information card (not shown) of the actor U1 who is the subject of the negative judgment to be read, and determines whether the actor U1 who is the subject of the negative judgment is the same person as the actor who has committed the same or similar act in the past.
[0039] (Correction function) The correction function is a function that corrects the behavior information of the actor U1 who is the subject of the negative judgment stored by the storage function when an affirmative judgment is made during the execution of the rejudgment function.
[0040] The above is the description of the configuration and functions of the judgment system JS of this embodiment.
[0041] (Judgment method (operation flow SP of judgment system)) Next, the judgment method (operation flow SP of judgment system JS) of this embodiment will be described with reference to FIGS. 3, 4, and 5. The operation flow SP of the judgment system JS is composed of S10 (abbreviation for step 10. The same interpretation applies to those other than S10) to S140 shown in FIG. 3. S10 to S140 proceed by the server 20 executing the above-mentioned multiple functions based on the program PG. Here, in the following description, as an example, the facility F is a library, and the actions that the actor U1 wishes to perform in the future and the actions that the actor U1 has performed in the past are borrowing books.
[0042] [S10] S10 is a step of confirming whether the actor U1 who wishes to enter (enter the library) the facility F (library) has an identity certificate (personal information card). If the result of S10 is an affirmative judgment, the process proceeds to the judgment step of S20. On the contrary, if the result is a negative judgment, the process proceeds to the judgment step of S60. This confirmation is performed by the server 20 that the UI 10 has a conversation with the actor U1 and receives the transmission information from the UI 10, as shown in Q1 and A1 which are examples of the conversation in FIG. 4.
[0043] 〔S20〕 S20 is a step of confirming the truth or falsehood for the actor U1 who made an affirmative determination in S10. In this step, the UI 10 instructs the actor U1 to present the identity certificate, transmits the information of the identity certificate read by the aforementioned reading function to the server 20, and the server 20 confirms the identity with the information stored in the storage unit 24 or an external authentication server (not shown). If the result of S20 is an affirmative determination, proceed to S30. On the contrary, if the result of S20 is a negative determination, proceed to the determination step of S60.
[0044] 〔S30, S40 and S50〕 S30 is a step performed when the identity of the actor U1 can be confirmed in S20, and is a step where the UI 10 permits the actor U1 to enter. S40 is a step of storing in the storage unit 24 the action content (specifically, the information of the borrowed book) and the time of the action (the borrowed date) of the actor U1 who entered the facility F and actually borrowed a book. That is, S40 is a step of executing the storage function. S50 is a step performed after the end of S40, and is a step of adding the action information of the current actor U1 stored in the storage unit 24 in S40 to a plurality of pieces of action information stored in the past, and generating a correlation model of the action information from the added plurality of pieces of action information. That is, S50 is a step of executing the aforementioned generation function. When the actor U1 leaves the facility F, the operation flow SP of the judgment system JS via S30 ends.
[0045] 〔S60 and S90〕 As shown in FIG. 3, S60 is a step that is performed when the result of S10 is a negative determination and when the result of S20 is a negative determination, that is, when the identity verification using the identity certificate fails. In this step, UI10 checks whether the actor U1 has used the facility F in the past. If the result of S60 is an affirmative determination, the process proceeds to the determination step of S70. On the other hand, if the result is a negative determination, the process proceeds to S90. This check is performed by UI10 having a dialogue with the actor U1, as shown in Q2 and A2, which are examples of the dialogue in FIG. 4. However, an affirmative determination of the result of S60 is merely a self-declaration of the actor U1 in terms of the execution of the program PG. Note that when the result of S60 is a negative determination and the process proceeds to S90, the actor U1 is refused entry to the facility F, and the operation flow SP of the determination system ends.
[0046] 〔S70 and S80〕 As shown in FIG. 3, S70 is a step in which, when the result of S60 is an affirmative determination, UI10 has a dialogue with the actor U1 to find out the planned actions for today (the actions planned after entering the facility F from now on) and identify the actions. Q3 and A3 of the dialogue in FIG. 4 are examples of the dialogue performed in this step. Next, when the planned actions of the actor U1 for today are identified by UI10, the process of S80 is performed. In the example of the dialogue in FIG. 4, the name of the actor U1 is identified as "Ako Suzuki", and the planned action for today is identified as "borrowing 'Norwegian Wood' by Haruki Murakami". S80 is a step of determining whether the actor U1 is the same person as the actor who has performed the same or similar actions in the past based on the correlation model when the actor U1 performs the same or similar actions as those performed in the past. That is, S80 is a step of executing the above-described determination function. Here, the actions taken by actor U1 in the past are, as shown in FIG. 5, "borrowing 'Hear the Wind Sing' by Haruki Murakami" (three times ago), "borrowing 'Pinball, 1973' by Haruki Murakami" (two times ago), and "borrowing 'A Wild Sheep Chase' by Haruki Murakami" (last time). The details of these actions are stored in the memory unit 24. Since the planned action identified in S70 for today is "borrowing 'Norwegian Wood' by Haruki Murakami", the server 20 determines that the planned action this time is the same or similar to the actions taken in the past. On the other hand, the correlation model of the action information of multiple actors U is generated by machine learning using the action information of actors U who have borrowed Haruki Murakami's books at facility F in the past as teacher data. The action identified from the correlation model based on the past actions of this actor U1 is, as shown in FIG. 5, "borrowing 'Norwegian Wood' by Haruki Murakami", and the similar actions of this action are either "borrowing 'The End of the World and Hard-Boiled Wonderland' by Haruki Murakami" or "borrowing 'Dance Dance Dance' by Haruki Murakami". In the above case, the server 20 determines that the planned action of actor U1 this time matches one of the same or similar actions derived from the correlation model (see FIGS. 4 and 5), and makes an affirmative determination in S80 (see FIG. 3). In this case, as shown in FIG. 3, the process proceeds to S30, and actor U1 is permitted to enter facility F. On the contrary, if the result of S80 is a negative determination, the process proceeds to S100, which will be described later.
[0047] 〔S100, S110, and S120〕 S100 is a process that is performed when it is not recognized in S80 that actor U1 is a person who has used facility F in the past (when it is determined that actor U1 is not the same person as the person who has performed the same or similar actions using facility F in the past), and in which UI10 rejects actor U1's entry. S110 is a process of storing in the memory unit 24 the action details (specifically, the information of the book planned to be borrowed) and the time (date) of the planned action of actor U1 whose entry to facility F has been rejected. That is, S110 is a process of executing the memory function. S120 is a process performed after the completion of S110. In this process, the behavior information of the current actor U1 stored in the storage unit 24 in S110 is added to a plurality of pieces of behavior information stored in the past, and a correlation model of the behavior information is generated from the plurality of pieces of behavior information after the addition. That is, S120 is a process of executing the above-described generation function. When a period (for example, one month) defined for the actor U1 to leave the facility F elapses, the operation flow SP of the determination system JS when passing through S100 ends.
[0048] 〔S130 and S140〕 S130 is a process of having the actor U1, who was rejected entry at S100 because the result of S80 was a negative determination, visit the facility F again within a period (for example, one month) determined from the day of rejection, and having UI10 read the identification document, and determining that the actor U1 who is the subject of the negative determination in S80 is the same person as an actor who has performed the same or similar behavior in the past. That is, S130 is a process of executing the above-described re-determination function. If the result of S130 is an affirmative determination, the process proceeds to S140, and the behavior information stored in S110 in the past and the correlation model generated in S120 are corrected. That is, S140 is a process of executing the above-described correction function. When S140 is executed, the operation flow SP of the determination system JS ends. On the other hand, if the result of S130 is a negative determination or if S130 is not executed within the above-described determined period, the operation flow SP of the determination system JS ends without executing S140.
[0049] The above is the description of the determination method (operation flow SP of the determination system JS) of the present embodiment.
[0050] <Effect> Next, the effects of the present embodiment will be described.
[0051] 〔The First Effect〕 The program PG of the present embodiment causes the server 20 or the judgment system JS to execute the above-described storage function (see S40 and S110 in FIG. 3), generation function (see S50 and S120 in FIG. 3 and FIG. 5), and judgment function (see S80 in FIG. 3). The judgment system JS machine-learns a correlation model of the behavior information generated by the generation function from a plurality of pieces of behavior information stored by the storage function, using the correlation relationship between the plurality of behavior contents and the timing of those behaviors as teacher data to generate it (see FIG. 5). Also, by the judgment function, when the actor U1 performs the same or similar behavior as the behavior performed in the past, based on the correlation model (depending on whether the above-described same or similar behavior matches any of the behaviors specified by the correlation model and the similar behaviors of that behavior), it is determined whether the actor U1 is the same person as the actor who performed the same or similar behavior in the past (see FIG. 5). Therefore, according to the program PG, judgment system JS, and judgment method of the present embodiment, when the actor U1 performs the same or similar behavior as the behavior performed in the past, it is possible to determine whether the two actors are the same person. Along with this, according to the judgment system JS of the present embodiment, when one actor performs the same or similar behavior as the behavior performed in the past, it is possible to determine whether the two actors are the same person.
[0052] 〔Second effect〕 As shown in FIG. 3, the judgment method of the present embodiment is performed when entering the facility F. However, after the identity verification by the identity certificate could not be performed at the initial stage (see FIG. 3), the steps of executing the above-described storage function (see S40 and S110 in FIG. 3), generation function (see S50 and S120 in FIG. 3 and FIG. 5), and judgment function (see S80 in FIG. 3) are performed. Therefore, it is effective in that the actor U can enter when the actor U does not have an identity certificate and has used the facility F in the past.
[0053] 〔Third effect〕 When the result of S80 is a negative judgment, the program PG of the present embodiment proceeds to S100 in the server 20 or the judgment system JS and rejects the entry of the actor U1. However, at a later date, if the actor U1 can be authenticated by having the identity certificate read on the UI10 in S120, the behavior information stored in S110 in the past and the correlation model generated in S120 are corrected in S130. Therefore, according to the program PG of the present embodiment, it is possible to correct the actor U1 who was negatively judged at the time of determining the same person in the past, and to generate (correct) a correlation model updated with the information accompanying the correction. That is, correct information accompanying human behavior can be fed back to the correlation model.
[0054] The above is the description of the effects of the present embodiment. Also, the above is the description of the present embodiment.
[0055] ≪Multiple Modification Examples≫ As described above, the present invention has been described by taking the above-described embodiment as an example. However, the embodiments included in the technical scope of the present invention are not limited to the above-described embodiments. For example, the following modification examples are also included. Here, in the following multiple modification examples, it should be noted that the names and symbols of the above-described embodiments are applied mutatis mutandis to the same or similar constituent elements described in the above-described embodiments.
[0056] For example, in the present embodiment, it is assumed that dialogue is used when the UI10 acquires behavior information from the actor U1. In this case, there is a possibility that fluctuations in words (similarity of words transmitted inaccurately) may occur in the name of the behavior transmitted from the actor U1. In such a case, in order not to result in a negative judgment during the execution of the judgment function, the fluctuations in words may be corrected with the following correlation model. This correlation model may be generated as an additional function by the generation function. Here, the function performed by the generation function refers to generating a correlation model by machine learning with the correlation relationship based on semantic identity as teacher data regardless of whether the descriptions of each name (word) are the same. Also, the correlation relationship based on semantic identity regardless of whether the descriptions of each name (word) are the same means, for example, (1) when "name", "first name", and "your name" are used as names (words), (2) when "address", "your address", and "your place of residence" are used as names (words), and (3) when "child", "kid", "child", and "your child" are used as names (words), the names (words) in each group have the same content (the same meaning) as each other, so they have the same relationship with each other. In contrast, for example, when "date of birth" and "place of birth" are used as names (words), these item names have different contents (different meanings) from each other, so they have different relationships with each other.
[0057] Also, as described, the determination system JS of the present embodiment includes a UI10, a server 20, and a communication network 30 as shown in FIG. 1. However, if the UI10 can also function as a server 20 as a single computer, the determination system JS may be configured only with the UI10.
[0058] Also, in the present embodiment, the facility F where the UI10 is installed is assumed to be a library as an example. Therefore, the action planned for the facility F may be borrowing books. However, the facility F may be a complex facility that divides the entering actor U into multiple types of action targets and allows entry. Here, as an example of a complex facility, there may be a general hospital having multiple types of medical services (internal medicine, surgery, dermatology, dentistry, pediatrics, etc.), an administrative agency having multiple types of administrative services, a comprehensive university having multiple types of faculties, and other complex facilities. That is, from the perspective of users (actors) of these complex facilities, it can be said that a complex facility has multiple types of actions. Also, in such a case, in the memory function, the action information may be stored separately by action type. Further, in the generation function, from a plurality of action information stored separately by action type by the memory function, the correlation relationship between a plurality of action contents and the timing of those actions may be machine-learned as teacher data to generate a correlation model of the action information for each action type. According to the program of this modification example, it is effective in that it can determine with high accuracy that it is the same person compared to a program that stores action information without classifying it by action type.
[0059] The above is the description of a plurality of modification examples.
Description of Signs
[0060] 10 User Interface (UI) 20 Server 22 Processing Unit 24 Memory Unit 30 Communication Network AP Application DF Data File F Facility JS Judgment System LB Various Libraries PG Program SP Operation Flow of Judgment System (Judgment Method) U Actor U1 Actor (One Actor)
Claims
1. A program for determining whether two actors are the same person when one actor performs the same or similar act as an act performed in the past, the program causing a computer to: have a storage function that stores, for each occurrence of the same or similar act performed by a plurality of actors including the one actor, act information including the content of the act of each actor and the time of the act; a generation function that performs machine learning on the correlation between a plurality of the act contents and the times of those acts from the plurality of pieces of act information stored by the storage function, using the correlation as teacher data, to generate a correlation model of the act information; a determination function that, when the one actor performs the same or similar act, determines, based on the correlation model, whether the one actor is the same person as the actor who performed the same or similar act in the past; and execute the above functions. A program.
2. The phrase "based on the correlation model" means that it depends on whether the same or similar act matches any of the acts specified by the correlation model and similar acts of that act. The program according to claim 1.
3. In the storage function, the act information is stored separately by act type, and in the generation function, machine learning is performed on the correlation between a plurality of the act contents and the times of those acts from the plurality of pieces of act information stored separately by act type by the storage function, using the correlation as teacher data, to generate a correlation model of the act information for each act type. The program according to claim 2.
4. The program causes the computer to have a reading function that reads the identification document of each actor, and the storage function stores the information of the identification document of each actor read by the reading function when each actor performs an act for the first time. The program according to any one of claims 1 to 3.
5. The program causes the computer to have a re-determination function that, after a negative determination is made during the execution of the determination function, reads the identification document of the actor subject to the negative determination to determine whether the actor subject to the negative determination is the same person as the actor who performed the same or similar act in the past, and a correction function that, when an affirmative determination is made during the execution of the re-determination function, corrects the act information of the actor subject to the negative determination stored by the storage function. The program according to claim 4.
6. A computer having a memory unit and a processing unit, storing the program according to any one of claims 1 to 3 in the memory unit, and processing the program by the processing unit to execute the storage function, the generation function, and the determination function, and An interface communicably connected to the computer and acquiring a plurality of pieces of the action information stored by the storage function from each actor, and Comprising A judgment system.
7. A computer having a memory unit and a processing unit, storing the program according to claim 5 in the memory unit, and processing the program by the processing unit to execute the storage function, the generation function, the determination function, the reading function, the re-determination function, and the correction function, and An interface communicably connected to the computer and acquiring a plurality of pieces of the action information stored by the storage function from each actor, and Comprising A judgment system.
8. A judgment method for judging that two actors are the same person when an actor performs an action similar to an action performed in the past using the program according to any one of claims 1 to 3, comprising: A step in which the computer executes the storage function; A step in which the computer executes the generation function; A step in which the computer executes the determination function; Including A judgment method.
9. A judgment method for judging that two actors are the same person when an actor performs an action similar to an action performed in the past using the program according to claim 5, comprising: A step in which the computer executes the storage function; A step in which the computer executes the generation function; A step in which the computer executes the determination function; A step in which the computer executes the reading function; A step in which the computer executes the re-determination function; A step in which the computer executes the correction function; Including A judgment method.