A method, device, equipment and medium for determining an account identification model

Abstract

The embodiment of the present invention discloses a method, device, equipment and medium for determining an account identification model. The method includes: obtaining account data, extracting features of the account data, and obtaining account features; training the account data according to the acquisition time of the account data and the account features through a preset training model to obtain at least two base classifiers; determining the classifier calculation formula corresponding to each base classifier according to a preset forgetting factor based on time variation, and determining the account identification model according to each classifier calculation formula. By running the technical solution provided by the embodiment of the present invention, the problem of identifying zombie accounts based on machine learning models can be solved. However, since zombie accounts have a strong timeliness and machine learning models are usually relatively fixed, it is easy to cause the detection accuracy of accounts to gradually decrease over time, thereby achieving the effect of improving the accuracy of account identification.

Description

Technical Field

[0001] Embodiments of the present invention relate to computer technology, and more particularly to a method, device, equipment and medium for determining an account identification model. Background Art

[0002] With the development of society, social networks have changed people's lives, but with it comes the proliferation of zombie accounts. In the field of social networks, zombie accounts usually refer to personal accounts with low activity or fake accounts operated in batches by robots, so these low-value accounts are filtered out.

[0003] In the prior art, zombie accounts are usually identified based on machine learning models. However, since zombie accounts have a strong timeliness and machine learning models are usually relatively fixed, the detection accuracy of accounts may gradually decrease over time. Summary of the invention

[0004] Embodiments of the present invention provide a method, device, equipment and medium for determining an account identification model to improve the accuracy of account identification.

[0005] In a first aspect, an embodiment of the present invention provides a method for determining an account identification model, the method comprising:

[0006] Acquire account data, and extract features of the account data to obtain account features;

[0007] By using a preset training model, the account data is trained according to the acquisition time of the account data and the account features to obtain at least two base classifiers;

[0008] The classifier calculation formula corresponding to each of the base classifiers is determined according to a preset forgetting factor based on time variation, and the account recognition model is determined according to each of the classifier calculation formulas.

[0009] In a second aspect, an embodiment of the present invention further provides an account identification model determination device, the module comprising:

[0010] An account feature extraction module is used to obtain account data and extract features from the account data to obtain account features;

[0011] A base classifier acquisition module, configured to train the account data according to the acquisition time of the account data and the account characteristics through a preset training model to obtain at least two base classifiers;

[0012] The account recognition model determination module is used to determine the classifier calculation formula corresponding to each base classifier according to a preset forgetting factor based on time changes, and determine the account recognition model according to each classifier calculation formula.

[0013] In a third aspect, an embodiment of the present invention further provides an electronic device, the electronic device comprising:

[0014] one or more processors;

[0015] a storage device for storing one or more programs,

[0016] When the one or more programs are executed by the one or more processors, the one or more processors implement the account identification model determination method as described above.

[0017] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the account identification model determination method as described above.

[0018] The embodiment of the present invention obtains account data and extracts features of the account data to obtain account features; trains the account data according to the acquisition time of the account data and the account features through a preset training model to obtain at least two base classifiers; determines the classifier calculation formula corresponding to each base classifier according to a preset forgetting factor based on time changes, and determines the account recognition model according to each classifier calculation formula. The problem of identifying zombie accounts based on machine learning models is solved, but since zombie accounts have strong timeliness and machine learning models are usually relatively fixed, it is easy to cause the detection accuracy of accounts to gradually decrease over time, thereby achieving the effect of improving the accuracy of account recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 A flowchart of a method for determining an account identification model provided in Embodiment 1 of the present invention;

[0020] Figure 2 A flowchart of a method for determining an account identification model provided in Embodiment 2 of the present invention;

[0021] Figure 3 A schematic diagram of the structure of an account identification model determination device provided in Embodiment 3 of the present invention;

[0022] Figure 4 A schematic diagram of the structure of an electronic device provided in Embodiment 4 of the present invention. DETAILED DESCRIPTION

[0023] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only parts related to the present invention, rather than all structures, are shown in the accompanying drawings.

[0024] Embodiment 1

[0025] Figure 1 This is a flowchart of a method for determining an account identification model provided in the first embodiment of the present invention. This embodiment is applicable to the case of determining a zombie account identification model. The method can be executed by the account identification model determination device provided in the embodiment of the present invention. The device can be implemented in software and / or hardware. Figure 1 The account identification model determination method provided in this embodiment includes:

[0026] Step 110: Obtain account data, and extract features of the account data to obtain account features.

[0027] The account data may be obtained by crawling a web crawler or by calling a corresponding application program interface of a website, and this embodiment does not limit this.

[0028] Exemplarily, a distributed web crawler based on the Scrapy framework is used to collect account data. Since Scrapy has a highly customized, asynchronous multi-threaded crawler function, multiple hosts are deployed with crawler programs developed based on Scrapy, and crawling is coordinated. That is, the host no longer maintains its own crawling queue, but all hosts share the crawling queue and execute the same crawling task at the same time to perform distributed crawling, thereby greatly improving the performance and efficiency of data acquisition.

[0029] The account data features are extracted to obtain account features, wherein the account features may be features that are highly correlated with the characteristics of zombie accounts, and illustratively, may be the similarity of content posted in the account, etc.

[0030] Step 120: Train the account data according to the acquisition time of the account data and the account characteristics through a preset training model to obtain at least two base classifiers.

[0031] The different acquisition times of the account data can be different preset intervals, for example, the account data is acquired at intervals of days, and the account data obtained at different acquisition times are combined into a single 1 , α 2 , α 3 , ..., α t Train them separately to get the corresponding base classifier D(α 1 ), D(α 2 ), ..., D(α t ).

[0032] Among them, the base classifier can be a naive Bayes classifier. Since the naive Bayes classifier has good noise resistance, the Bayes classifier is used to convert the separation relationship between normal accounts and zombie accounts into a random variable, and conditional probability analysis can be used to analyze this relationship.

[0033] Step 130: determine the classifier calculation formula corresponding to each base classifier according to the preset forgetting factor based on time variation, and determine the account recognition model according to each classifier calculation formula.

[0034] The classifier calculation formula corresponding to each base classifier is determined according to a preset forgetting factor based on time variation, wherein the forgetting factor based on time variation is a factor whose value decreases as time increases. By substituting the forgetting factor into the classifier calculation formula originally corresponding to the base classifier, the base classifier is made time-dependent.

[0035] Optionally, set the samples of the base classifier to α according to the time at which they were acquired. 1 , α 2 , α 3 , ..., α t , the base classifier is set to D(α t ) meets the following conditions:

[0036] (1) For t<i≤T, α t =0; where T is the current time, that is, the unobtained samples will not affect the classifier.

[0037] (2) When η t When the value of is less than the preset value, the classifier is discarded.

[0038] When the forgetting factor is less than a certain value, the influence of the classifier is discarded to achieve the effect of a sliding window in time to balance the impact of too many classifiers on the efficiency of the algorithm.

[0039] In this embodiment, optionally, the classifier calculation formula corresponding to each base classifier is determined according to a preset forgetting factor based on time variation, including:

[0040] According to the following formula, the classifier calculation formula corresponding to each base classifier is determined:

[0041] D(α t )=(1-η t )×D(α t-1 )

[0042] Among them, D(α t ) is the account data α obtained at time t t The trained base classifier; η t is the forgetting factor based on time change; D(αt-1 ) is a base classifier trained based on the account data acquired at time t-1;

[0043] The forgetting factor based on time change is determined according to the following formula:

[0044] η t =C 0 ×e -ρt

[0045] Among them, C 0 and ρ are preset constants.

[0046] The base classifier calculation formula at time t is obtained, where the base classifier calculation formula at time t is determined by the forgetting factor at time t. As t increases, the value of the forgetting factor decreases, so that the proportion of the result corresponding to the base classifier in the subsequent account recognition model decreases, so that the account recognition model changes dynamically according to time to improve the accuracy of current account recognition.

[0047] In this embodiment, optionally, determining the account identification model according to the classifier calculation formulas includes:

[0048] Determine the account identification model based on the following formula:

[0049]

[0050] Among them, W t is the account identification model; C is the preset constant; T is the current time.

[0051] The results obtained by each base classifier are summarized to obtain an account recognition model, thereby improving the accuracy of the account recognition model. Moreover, since the account recognition model changes over time, the accuracy of identifying the current zombie account is improved.

[0052] The technical solution provided in this embodiment substitutes the classifier calculation formula corresponding to each base classifier determined by the forgetting factor into the formula determined by the account recognition model, that is, the results of each base classifier are summarized to obtain the overall account recognition model. Since the forgetting factor has an exponential decay characteristic, the influence of the classifier trained based on the account data acquired over a long period of time on the final result can be reduced, so that the account recognition model changes dynamically over time to improve the accuracy of the current account recognition.

[0053] Embodiment 2

[0054] Figure 2This is a flowchart of a method for determining an account identification model provided in Embodiment 2 of the present invention. This technical solution is a supplementary explanation of the process after obtaining at least two base classifiers. Compared with the above solution, this solution is specifically optimized to further include: after obtaining at least two base classifiers:

[0055] Obtaining the difference between the two base classifiers;

[0056] Determine whether the difference is greater than the difference average; if so, retain the base classifier group corresponding to the difference greater than the difference average, and determine the base classifier corresponding to the base classifier group as the target base classifier;

[0057] Correspondingly, the classifier calculation formula corresponding to each base classifier is determined according to the preset forgetting factor based on time variation, including:

[0058] The classifier calculation formula corresponding to each of the target base classifiers is determined according to the preset forgetting factor based on time variation. Specifically, the flowchart of the method for determining the account recognition model is as follows: Figure 2 As shown:

[0059] Step 210: Obtain account data, and extract features of the account data to obtain account features.

[0060] In this embodiment, optionally, the account characteristics include: at least one of the following number of users, the following number of users, the like number of users, the posting number of users, the survival time of the account, the number of repeated contents, the account associated address and the account profile.

[0061] Since zombie accounts have a single behavior pattern, less information, fewer followers and fans, and fewer posts, they were easily automatically removed by the platform where the account was located. Therefore, such zombie accounts generally have a short survival time. Later, zombie accounts gradually developed. In order to evade the platform's detection system, they often have avatars and personal profiles, and regularly send meaningless tweets. Some even have high followers and fans, and most of their fans are also zombie accounts. Some zombie accounts will send advertisements or malicious links, while others will hijack trending topics to act as opinion control robots. Therefore, the characteristics of current zombie accounts are relatively complex.

[0062] The account features selected in this embodiment include: at least one of the number of users followed, the number of users followed, the number of users liked, the number of posts by the user, the account survival time, the number of content repetitions, the account associated address and the account profile.

[0063] Among them, the number of accounts followed by a user refers to the number of accounts followed by an account. Compared with normal users who can only follow a limited number of accounts that they can handle, zombie accounts will follow a large number of other accounts in order to meet economic benefits, and even approach the upper limit of followers allowed by the platform.

[0064] The number of followers of a user refers to the number of fans of an account. From the perspective of the influence of an account, the more fans an account has, the higher the influence of the account. Operating a zombie account with a large number of fans requires more resources, so the number of fans of a zombie account is usually not too high, and most of the fans of a zombie account are also zombie accounts themselves.

[0065] The number of likes a user receives refers to the number of likes received by an account.

[0066] The number of user posts refers to the amount of content sent by an account. The number of user posts is an important indicator of whether a user is active. A normal user often sends a lot of content from registration to the present, and a carefully maintained zombie account may also have a large number of tweets, while the number of tweets of a zombie account that has stopped maintenance is often small. Advertising zombie accounts may also send tweets at a high frequency for advertising promotion.

[0067] Account survival time refers to the length of time an account exists. Compared to zombie accounts, normal accounts often survive longer.

[0068] The number of content duplicates refers to the number of similar content posted in an account. If an account posts a large number of tweets with duplicate content, it is likely to be regarded as a zombie account.

[0069] The account-associated address refers to the location information filled in by the user. Typical zombie accounts often do not fill in too much account information. The location can reflect whether the account is a zombie account to a certain extent.

[0070] An account profile refers to a personal profile in an account. Official and long-running accounts generally have a personal profile, while zombie accounts are more likely to have no profile, or the profile information is meaningless.

[0071] By extracting account features that are highly correlated with zombie accounts, the accuracy of the account recognition model obtained through subsequent training is improved, thereby improving the accuracy of subsequent zombie account recognition.

[0072] Step 220: Train the account data according to the acquisition time of the account data and the account characteristics through a preset training model to obtain at least two base classifiers.

[0073] Step 230: Obtain the difference between the two base classifiers.

[0074] The difference degree refers to the difference degree of the prediction results obtained by different base classifiers for the same prediction data.

[0075] The difference can be obtained by making different predictions on the same test set through two base classifiers. For the test set Q, the difference measure d is determined according to the following formula mn :

[0076]

[0077] Among them, Q 11 Indicates that both base classifiers predict successfully; Q 00 Indicates that both base classifiers fail to predict; Q 01 Indicates that base classifier 1 failed to predict, and base classifier 2 succeeded; Q 10 It means that base classifier 1 predicts successfully and base classifier 2 predicts unsuccessfully.

[0078] d mn ∈[0,1], the larger the value, the greater the difference between the base classifiers, and the more representative it is. A single base classifier can be used as a benchmark to obtain the difference between other base classifiers and the base classifier, which is not limited in this embodiment.

[0079] Step 240, determine whether the difference is greater than the difference average; if greater, retain the base classifier group corresponding to the difference greater than the difference average, and determine the base classifier corresponding to the base classifier group as the target base classifier.

[0080] After averaging all the obtained differences, a base classifier group whose difference is greater than the average difference is determined, two base classifiers corresponding to the base classifier group are determined as target base classifiers, and duplicate target base classifiers are deleted.

[0081] Step 250: determine the classifier calculation formula corresponding to each of the target base classifiers according to the preset forgetting factor based on time variation, and determine the account recognition model according to each of the classifier calculation formulas.

[0082] The classifier calculation formula corresponding to each target base classifier is determined according to the preset time-based forgetting factor, and the results of each target classifier are summarized to obtain the overall account recognition model.

[0083] The embodiment of the present invention screens the base classifiers to obtain representative base classifiers with stronger classification capabilities, eliminates redundant base classifiers, and determines the account recognition model based on the screened base classifiers, so that the account recognition model continuously selects a portion of time data with better classification effects, thereby improving the accuracy of the account recognition model.

[0084] Embodiment 3

[0085] Figure 3 This is a schematic diagram of the structure of an account identification model determination device provided in the third embodiment of the present invention. The device can be implemented by hardware and / or software, can execute an account identification model determination method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method. Figure 3 As shown, the device comprises:

[0086] An account feature extraction module 310 is used to obtain account data and extract features from the account data to obtain account features;

[0087] A base classifier acquisition module 320, configured to train the account data according to the acquisition time of the account data and the account features through a preset training model to obtain at least two base classifiers;

[0088] The account identification model determination module 330 is used to determine the classifier calculation formula corresponding to each of the base classifiers according to a preset forgetting factor based on time variation, and determine the account identification model according to each of the classifier calculation formulas.

[0089] The technical solution provided in this embodiment substitutes the classifier calculation formula corresponding to each base classifier determined by the forgetting factor into the formula determined by the account recognition model, that is, the results of each base classifier are summarized to obtain the overall account recognition model. Since the forgetting factor has an exponential decay characteristic, the influence of the classifier trained based on the account data acquired over a long period of time on the final result can be reduced, so that the account recognition model changes dynamically over time to improve the accuracy of the current account recognition.

[0090] Based on the above technical solutions, optionally, the account identification model determination module includes:

[0091] The first classifier calculation formula determination unit is used to determine the classifier calculation formula corresponding to each base classifier according to the following formula:

[0092] D(α t )=(1-η t )×D(α t-1 )

[0093] Among them, D(α t ) is the account data α obtained at time t t The trained base classifier; η t is the forgetting factor based on time change; D(α t-1 ) is a base classifier trained based on the account data acquired at time t-1;

[0094] The forgetting factor based on time change is determined according to the following formula:

[0095] η t =C 0 ×e -ρt

[0096] Among them, C 0 and ρ are preset constants.

[0097] Based on the above technical solutions, optionally, the account identification model determination module includes:

[0098] The account identification model determination unit is used to determine the account identification model according to the following formula:

[0099]

[0100] Among them, W t is the account identification model; C is the preset constant; T is the current time.

[0101] On the basis of the above technical solutions, optionally, the device further includes:

[0102] A difference acquisition module, used before the account recognition model determination module, to acquire the difference between the two base classifiers;

[0103] A target base classifier determination module is used to determine whether the difference is greater than the difference average; if so, retain the base classifier group corresponding to the difference greater than the difference average, and determine the base classifier corresponding to the base classifier group as the target base classifier;

[0104] Accordingly, the account identification model determination module includes:

[0105] The second classifier calculation formula determination unit is used to determine the classifier calculation formula corresponding to each of the target base classifiers according to a preset forgetting factor based on time variation.

[0106] On the basis of the above-mentioned technical solutions, optionally, the account characteristics include: at least one of the following number of users, the following number of users, the like number of users, the posting number of users, the survival time of the account, the number of repeated contents, the account associated address and the account profile.

[0107] Embodiment 4

[0108] Figure 4 A schematic diagram of the structure of an electronic device provided in Embodiment 4 of the present invention is shown in FIG. Figure 4 As shown, the electronic device includes a processor 40, a memory 41, an input device 42 and an output device 43; the number of processors 40 in the electronic device can be one or more. Figure 4 A processor 40 is taken as an example; the processor 40, the memory 41, the input device 42 and the output device 43 in the electronic device can be connected by a bus or other means. Figure 4 The example of connecting through bus is taken in the following.

[0109] The memory 41 is a computer-readable storage medium that can be used to store software programs, computer executable programs and modules, such as program instructions / modules corresponding to the account identification model determination method in the embodiment of the present invention. The processor 40 executes various functional applications and data processing of the electronic device by running the software programs, instructions and modules stored in the memory 41, that is, implements the above-mentioned account identification model determination method.

[0110] The memory 41 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system and at least one application required for a function; the data storage area may store data created according to the use of the terminal, etc. In addition, the memory 41 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some instances, the memory 41 may further include a memory remotely arranged relative to the processor 40, and these remote memories may be connected to the electronic device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0111] Embodiment 5

[0112] Embodiment 5 of the present invention further provides a storage medium including computer executable instructions, wherein the computer executable instructions are used to execute a method for determining an account identification model when executed by a computer processor, the method comprising:

[0113] Acquire account data, and extract features of the account data to obtain account features;

[0114] By using a preset training model, the account data is trained according to the acquisition time of the account data and the account features to obtain at least two base classifiers;

[0115] The classifier calculation formula corresponding to each of the base classifiers is determined according to a preset forgetting factor based on time variation, and the account recognition model is determined according to each of the classifier calculation formulas.

[0116] Of course, the storage medium containing computer executable instructions provided in an embodiment of the present invention is not limited to the method operations described above, and can also execute related operations in the account identification model determination method provided in any embodiment of the present invention.

[0117] Through the above description of the implementation methods, the technicians in the relevant field can clearly understand that the present invention can be implemented by means of software and necessary general hardware, and of course it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory (FLASH), hard disk or optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods described in each embodiment of the present invention.

[0118] It is worth noting that in the embodiment of the above-mentioned account identification model determination device, the various units and modules included are only divided according to functional logic, but are not limited to the above-mentioned division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0119] Note that the above are only preferred embodiments of the present invention and the technical principles used. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and that various obvious changes, readjustments and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in more detail through the above embodiments, the present invention is not limited to the above embodiments, and may include more other equivalent embodiments without departing from the concept of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. A method for determining an account identification model, characterized in that: include: Acquire account data, and extract features of the account data to obtain account features; By using a preset training model, the account data is trained according to the acquisition time of the account data and the account features to obtain at least two base classifiers; Determine the classifier calculation formula corresponding to each base classifier according to the preset forgetting factor based on time change, and determine the account recognition model according to each classifier calculation formula; The forgetting factor based on time variation is a factor whose value decreases with time, and when the forgetting factor based on time variation is less than a preset value, the corresponding classifier is discarded.

2. The method according to claim 1, characterized in that Determining the classifier calculation formula corresponding to each base classifier according to a preset forgetting factor based on time variation includes: According to the following formula, the classifier calculation formula corresponding to each base classifier is determined: D(a t )=(1-th t )×D(a t-1 ) Among them, D(α t ) is the account data α obtained at time t t The trained base classifier; η t is the forgetting factor based on time change; D(α t-1 ) is a base classifier trained based on the account data acquired at time t-1; The forgetting factor based on time change is determined according to the following formula: or t =C0×e -ρt Among them, C0 and ρ are preset constants.

3. The method according to claim 2, characterized in that Determining an account identification model according to the classifier calculation formulas includes: Determine the account identification model based on the following formula: Among them, W t is the account identification model; C is the preset constant; T is the current time.

4. The method according to claim 1, characterized in that After obtaining at least two base classifiers, it also includes: Obtaining the difference between the two base classifiers; Determine whether the difference is greater than the difference average; if so, retain the base classifier group corresponding to the difference greater than the difference average, and determine the base classifier corresponding to the base classifier group as the target base classifier; Correspondingly, the classifier calculation formula corresponding to each base classifier is determined according to the preset forgetting factor based on time variation, including: The classifier calculation formula corresponding to each of the target base classifiers is determined according to a preset forgetting factor based on time variation.

5. The method according to claim 1, characterized in that The account features include: at least one of the number of users followed, the number of users followed, the number of users liked, the number of posts by users, the account survival time, the number of content repetitions, the account associated address and the account profile.

6. An account identification model determination device, characterized in that: include: An account feature extraction module is used to obtain account data and extract features from the account data to obtain account features; A base classifier acquisition module, configured to train the account data according to the acquisition time of the account data and the account characteristics through a preset training model to obtain at least two base classifiers; An account identification model determination module, used to determine a classifier calculation formula corresponding to each of the base classifiers according to a preset forgetting factor based on time variation, and determine an account identification model according to each of the classifier calculation formulas; The forgetting factor based on time variation is a factor whose value decreases with time, and when the forgetting factor based on time variation is less than a preset value, the corresponding classifier is discarded.

7. The device according to claim 6, characterized in that The account identification model determination module includes: The first classifier calculation formula determination unit is used to determine the classifier calculation formula corresponding to each base classifier according to the following formula: D(a t )=(1-th t )×D(a t-1 ) Among them, D(α t ) is the account data α obtained at time t t The trained base classifier; η t is the forgetting factor based on time change; D(α t-1 ) is a base classifier trained based on the account data acquired at time t-1; The forgetting factor based on time change is determined according to the following formula: or t =C0×e -ρt Among them, C0 and ρ are preset constants.

8. The device according to claim 6, characterized in that The account identification model determination module includes: The account identification model determination unit is used to determine the account identification model according to the following formula: Among them, W t is the account identification model; C is the preset constant; T is the current time.

9. An electronic device, characterized in that: The electronic device comprises: one or more processors; a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the account identification model determination method as described in any one of claims 1-5.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the account identification model determination method as described in any one of claims 1 to 5 is implemented.

Images