Shell company identification method, device, computer equipment and storage medium

Abstract

The present invention discloses a method, apparatus, computer device, and storage medium for identifying shell companies. The method comprises obtaining company information of a target company, including the target company's address information and a portrait image of at least one target employee of the target company; obtaining a portrait image taken within a preset geographic range of the company's address information, and performing image feature comparison between the employee portrait image and the portrait image to obtain a first image comparison result between the portrait image and the employee portrait image; determining a first comparison score corresponding to the portrait image and the target company based on the first image comparison result between the portrait image and the employee portrait image; obtaining a preset recognition threshold, and identifying the target company as a suspected shell company based on the first comparison score and the preset recognition threshold to obtain a first recognition result. The present invention improves the efficiency and accuracy of shell company identification and analysis.

Description

Technical Field

[0001] The present invention relates to the field of network information technology, and in particular to a method, device, computer equipment and storage medium for identifying shell companies. Background Art

[0002] Shell companies may engage in economic crimes by engaging in various short-selling schemes, often involving large sums of money, threatening financial order and economic development. Therefore, accurately identifying shell companies can effectively reduce or even prevent the risks of illicit fund transfers and fraud.

[0003] In the existing technology, companies are usually investigated by judicial authorities, or investigated upon receiving public complaints. However, this method requires a lot of manpower and resources, is inefficient in identifying shell companies, and cannot accurately identify shell companies. Summary of the Invention

[0004] Embodiments of the present invention provide a shell company identification method, apparatus, computer device, and storage medium to address the problem of low efficiency and low accuracy in shell company identification in the prior art.

[0005] A method for identifying a shell company, comprising:

[0006] Obtaining company information of a target company, the company information including company address information of the target company and a portrait of at least one target employee of the target company;

[0007] Obtaining a portrait picture taken within a preset geographical range of the company address information, and performing image feature comparison between the employee portrait picture and the portrait picture to obtain a first image comparison result between the portrait picture and the employee portrait picture;

[0008] determining a first comparison score between the portrait image and the target company based on a first image comparison result between the portrait image and the employee portrait image;

[0009] A preset identification threshold is obtained, and the target company is identified as a suspected shell company based on the first comparison score and the preset identification threshold to obtain a first identification result.

[0010] A device for identifying a shell company, comprising:

[0011] An information acquisition module is configured to acquire company information of a target company, wherein the company information includes company address information of the target company and a portrait of at least one target employee of the target company;

[0012] an image feature comparison module, configured to obtain a portrait image taken within a preset geographical range of the company address information, and perform image feature comparison between the employee portrait image and the portrait image to obtain a first image comparison result between the portrait image and the employee portrait image;

[0013] a comparison score determination module, configured to determine a first comparison score corresponding to the portrait image and the target company based on the first image comparison result between the portrait image and the employee portrait image;

[0014] The shell company identification module is used to obtain a preset identification threshold, and identify the target company as a suspected shell company based on the first comparison score and the preset identification threshold to obtain a first identification result.

[0015] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for identifying shell companies is implemented.

[0016] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the above-mentioned method for identifying shell companies.

[0017] The above-mentioned shell company identification method, apparatus, computer device, and storage medium obtain a portrait image taken within a preset geographic range of the target company's company address information, and then compare image features between a portrait image of a target employee at the target company and the portrait image to obtain a first image comparison result. This image feature comparison can then be used to determine whether the target employee is present at the target company or within the preset geographic range of the target company. When a first comparison score determined based on the first image comparison result is greater than or equal to a preset identification threshold, it indicates that most or all of the target employees are absent from the target company or near the target company, thereby determining that the target company is a suspected shell company. This method can thus provide a preliminary shell company analysis of the target company, providing accurate data for subsequent analysis of the target company and improving the efficiency and accuracy of shell company analysis and identification of the target company. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0019] Figure 1 This is a schematic diagram of an application environment of a method for identifying a shell company in one embodiment of the present invention;

[0020] Figure 2 is a flow chart of a method for identifying a shell company in one embodiment of the present invention;

[0021] Figure 3 is another flow chart of a method for identifying a shell company in one embodiment of the present invention;

[0022] Figure 4 is another flow chart of a method for identifying a shell company in one embodiment of the present invention;

[0023] Figure 5 is another flow chart of a method for identifying a shell company in one embodiment of the present invention;

[0024] Figure 6 This is a principle block diagram of a device for identifying a shell company in one embodiment of the present invention;

[0025] Figure 7 FIG. 1 is a schematic diagram of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0026] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0027] The shell company identification method provided by the embodiment of the present invention can be applied as follows: Figure 1 Specifically, the shell company identification method is applied in the shell company identification system, and the shell company identification system includes the following: Figure 1The client and server shown can be used to implement the shell company identification method through the client or server in this shell company identification system. The client and server communicate via a network, addressing the low efficiency and accuracy of shell company identification in existing technologies. The client, also known as the user end, refers to the program that corresponds to the server and provides local services to clients. The client can be installed on, but is not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices. The server can be implemented as a standalone server or a server cluster consisting of multiple servers.

[0028] In one embodiment, if Figure 2 As shown, a method for identifying a shell company is provided, which is applied in Figure 1 The server in the example is used as an example, and the steps are as follows:

[0029] S10: Acquire company information of a target company, where the company information includes company address information of the target company and a portrait of at least one target employee of the target company.

[0030] It can be understood that the target company refers to the company that needs to be analyzed as a shell company, and the company information refers to the information associated with the target company. For example, the company information can be the address, business information, etc. of the target company. The company address information mainly refers to the address information of the target company when it is registered.

[0031] The target employee refers to an employee currently employed by the target company. Furthermore, in this embodiment, the number of target employees is not limited. That is, the target employees here can be all employees currently employed by the target company, or some employees currently employed by the target company (e.g., the legal representative, directors, supervisors, and senior managers of the target company). An employee portrait image refers to an image that includes the facial features of the target employee, and each target employee has one employee portrait image. Furthermore, the employee portrait images can be stored in a database of the target company, and an employee portrait image of each target employee can be obtained from the database.

[0032] S20: Obtain a portrait picture taken within a preset geographical range of the company address information, and compare image features of the employee portrait picture of each target employee with the portrait picture to obtain a first image comparison result corresponding to each employee portrait picture.

[0033] It can be understood that the portrait picture is a picture taken by a camera or other shooting equipment within the preset geographical range of the company's address information, such as a picture taken by a camera at the door or elevator of the building where the target company is located. The portrait picture is generally stored in the database associated with the camera of the building where the company is located, so the portrait picture can be directly obtained from the database associated with these cameras. Among them, the preset geographical range can be limited according to the specific scenario, for example, the preset geographical range can be the building to which the target company belongs, or the park to which the target company belongs. It should be noted that in this embodiment, it is necessary to perform image feature comparison of the portrait, so the portrait picture is obtained through pre-filtering, that is, after performing portrait recognition on the captured image obtained from the database associated with the camera, the captured image containing at least one portrait information (that is, facial feature) is recorded as a portrait picture.

[0034] Furthermore, the image feature comparison in this embodiment is to determine whether there are features in the portrait photo that are the same as the facial features of the target employee in the employee portrait photo, that is, to compare an employee portrait photo with the portrait photo one by one, and then to determine whether the facial features in the portrait photo are the same as the facial features in the employee portrait photo. The first image comparison result includes a result indicating a successful comparison and a result indicating a failed comparison. Among them, when the first image comparison result of an employee portrait photo indicates a successful comparison, it means that there is a facial feature of a photographed object (the photographed object is a different object in the portrait photo, which can be distinguished by different facial features) in at least one portrait photo that is the same or similar to the facial feature in the employee portrait photo, which indicates that the target employee corresponding to the employee portrait photo appears near the target company or in the target company. When the first image comparison result of an employee portrait picture indicates a comparison failure, it means that the facial features of the subjects in all portrait pictures are different from the facial features in the employee portrait image, which means that the target employee corresponding to the employee portrait picture does not appear near or inside the target company.

[0035] Specifically, after obtaining the company information of the target company, portrait pictures are obtained from the database of cameras or other camera equipment associated with the preset geographical range of the company address information of the target company, and the employee portrait pictures of each target employee are compared with the portrait pictures one by one for image features, and then it is determined whether there are portrait pictures with facial features that are the same or similar to those in the employee's facial image. In this way, the first image comparison result corresponding to each employee portrait picture can be determined.

[0036] S30: Determine a first comparison score corresponding to the portrait picture and the target company based on the first image comparison result between the portrait picture and the employee portrait picture.

[0037] It is understood that the above description indicates that the first image comparison results include results indicating successful comparisons and results indicating failed comparisons. Therefore, based on the first image comparison results corresponding to each employee portrait, a first comparison score corresponding to the target company can be determined. For example, this first comparison score can be represented by the ratio of the total number of first image comparison results indicating failed comparisons to the total number of employee portraits. The first comparison score serves as the basis for determining whether the target company is a suspected shell company.

[0038] S40: Obtain a preset identification threshold, and identify the target company as a suspected shell company based on the first comparison score and the preset identification threshold to obtain a first identification result.

[0039] Optionally, the preset recognition threshold may be selected based on, for example, the number of target employees of the target company. The first recognition result includes a recognition result indicating that the target company is a suspected shell company, and a recognition result indicating that the target company is not a suspected shell company.

[0040] Specifically, after determining a first comparison score between the portrait image and the target company based on a first image comparison result between the portrait image and the employee portrait image, a preset recognition threshold is obtained and the first comparison score is compared with the preset recognition threshold; if the first comparison score is greater than or equal to the preset recognition threshold, the first recognition result is determined to indicate that most or all of the target employees of the target company are not present near the company, and thus the target company may be a shell company, and therefore the target company needs to be marked as a suspected shell company. If the first comparison score is less than the preset recognition threshold, the first recognition result is determined to indicate that most or all of the target employees of the target company are present near the company, and thus the target company may not be a shell company, and therefore the target company is marked as a non-suspected shell company.

[0041] Furthermore, after determining that the target company is a suspected shell company, an alarm message is sent to the preset recipient, and the preset recipient can then monitor the target company in real time and provide the preset recipient with real-time data. For example, within a certain time range (such as one month, three months, six months, etc., the time range should not be too short, such as two days, three days, etc., such as when on vacation or going out for an annual meeting, or when playing, the first comparison score may be greater than or equal to the preset identification threshold), the first comparison score of the target company is continuously greater than or equal to the preset identification threshold, indicating that the target company is indeed likely to be a shell company, and the relevant department personnel can be notified to investigate the target company. If the target company is found to be a shell company, the target company will be labeled as a shell company and no shell company analysis will be performed on the target company. In this way, the efficiency and accuracy of identifying shell companies can be improved.

[0042] In this embodiment, a first image comparison result is obtained by obtaining a portrait image taken within a preset geographic range corresponding to the target company's company address information and performing an image feature comparison between a portrait image of a target employee at the target company and the portrait image. This image feature comparison can then be used to determine whether the target employee is present at the target company or within the preset geographic range of the target company. If a first comparison score determined based on the first image comparison result is greater than or equal to a preset identification threshold, this indicates that most or all of the target employees are absent from the target company or its vicinity, further confirming that the target company is a suspected shell company. This method allows for a preliminary shell company analysis of the target company, providing accurate data for subsequent analysis of the target company and improving the efficiency and accuracy of shell company analysis and identification.

[0043] In one embodiment, in step S20, that is, performing image feature comparison on the employee portrait image of each target employee and the portrait photograph to obtain a first image comparison result corresponding to each employee portrait image, includes:

[0044] The portrait pictures are subjected to portrait clustering to obtain at least one portrait category including captured portrait pictures of the same photographed subject; the captured portrait pictures are portrait pictures of each of the photographed subjects captured from the portrait pictures.

[0045] It can be understood that portrait clustering is a method for distinguishing different subjects in a portrait image, and then classifying different facial features in the portrait image into a portrait category corresponding to the corresponding subject. Specifically, one subject corresponds to one portrait category, and one portrait category includes portrait images of the subject captured from the portrait image, namely, captured portrait images. One portrait category includes at least one captured portrait image. For example, if the subject appears in different portrait images, portrait images of the subject can be captured from multiple different portrait images. In this way, after the portrait images are clustered and the portrait categories are obtained, the efficiency of image feature comparison between employee portrait images and portrait images in subsequent steps can be improved.

[0046] For each of the photographed portrait categories, determine the image similarity between the employee portrait picture and the captured portrait picture in the photographed portrait category, and determine a first image comparison result between the portrait photograph and the employee portrait picture based on the image similarity.

[0047] Specifically, after performing portrait clustering on the portrait photos containing the photographed subject and obtaining at least one portrait category containing a cropped portrait photo of the same photographed subject, the employee portrait photo can be compared with the cropped portrait photos in each portrait category to determine the image similarity between the employee portrait photo and the cropped portrait photos in each portrait category, and then determine the first image comparison result between the portrait photo and the employee portrait photo based on the image similarity.

[0048] In this embodiment, after obtaining the portrait categories by performing portrait clustering on the portrait pictures, the employee portrait pictures can be compared with the cropped portrait pictures of the same subject, rather than comparing the employee portrait pictures with any cropped portrait pictures. Since the facial features of the same subject are the same, comparing the employee portrait pictures with the cropped portrait pictures of the same subject can better notice the differences between the facial features, thereby improving the efficiency and accuracy of image feature comparison.

[0049] In one embodiment, for each of the photographed portrait categories, determining the image similarity between the employee portrait image and the captured portrait image in the photographed portrait category, and determining a first image comparison result between the portrait image and the employee portrait image based on the image similarity, includes:

[0050] For each of the portrait shooting categories, among all the intercepted portrait pictures in the portrait shooting category, a intercepted portrait picture with the highest definition is selected as the picture to be compared in the portrait shooting category.

[0051] Understandably, if the portrait image is blurry, the captured portrait image may also be relatively blurry. Comparing a relatively blurry captured portrait image with the employee portrait image is less meaningful and may even result in a failure in the image comparison between the employee portrait and the captured portrait image. Therefore, in this embodiment, it is necessary to compare the clarity of all captured portrait images in the same portrait category, and then record the highest-resolution captured portrait image in each portrait category as the image to be compared. In other words, there is one image to be compared for each portrait category.

[0052] Before selecting the highest-resolution cutout portrait image as the comparison image for the portrait category, the clarity values ​​of each cutout portrait image may be predetermined. For example, the clarity values ​​of each cutout portrait image may be determined using a Laplacian gradient function, an SMD2 (grayscale variance product) function, an NRSS gradient structural similarity method, or the like, and the cutout portrait image with the highest clarity value is recorded as the comparison image.

[0053] Determine the image similarity between the employee portrait image and the to-be-compared images of each portrait category.

[0054] Specifically, for each of the portrait categories, among all the captured portrait pictures of the portrait category, the captured portrait picture with the highest clarity is selected as the comparison picture of the portrait category, and then the picture similarity between the employee portrait picture and the comparison picture of the portrait category is determined, that is, for an employee portrait picture, there is a picture similarity between it and the comparison picture of each portrait category, and then each picture similarity is compared with a preset similarity threshold to determine whether each portrait picture contains features that are the same as the facial features in the employee portrait picture.

[0055] A preset similarity threshold is obtained, and a first image comparison result between the portrait picture and the employee portrait picture is determined according to the picture similarity and the preset similarity threshold.

[0056] Specifically, after determining the image similarity between the employee portrait image and the image to be compared in the photographed portrait category, a preset similarity threshold is obtained, and the image similarity is compared with the preset similarity threshold; if the image similarity is greater than or equal to the preset similarity threshold, it can be determined that the facial features in the employee portrait image are the same as or similar to the facial features in the captured portrait image, and the first image comparison result corresponding to the employee portrait image corresponding to the image similarity can be determined to be a successful comparison. It can be understood that as long as the image similarity between the captured portrait image in any one of the photographed portrait categories and the employee portrait image is greater than or equal to the preset similarity threshold, the first image comparison result corresponding to the employee portrait image can be determined to be a successful comparison.

[0057] Furthermore, after determining the image similarity between the employee portrait picture and the pictures to be compared in each photographed portrait category, and comparing the image similarity with the preset similarity threshold, if the similarities of all pictures corresponding to the same employee portrait picture are less than the preset similarity threshold, then the facial features in the employee portrait picture are different from the facial features of the captured portrait pictures in all photographed portrait categories, and the differences are large. In this way, it can be determined that the first image comparison result corresponding to the employee portrait picture is a comparison failure.

[0058] In this embodiment, by selecting the highest-definition captured portrait image for image feature comparison with the employee portrait image, the efficiency of the image feature comparison can be effectively improved, and the image feature comparison of the employee portrait image with the captured portrait image with poor image quality (such as low clarity) can be avoided, thereby further improving the accuracy of the image feature comparison.

[0059] In one embodiment, if Figure 3 As shown, after performing image feature comparison on each of the employee portrait pictures and the portrait photo to obtain a first image comparison result between the portrait photo and the employee portrait picture, the method further includes:

[0060] S01: When the first image comparison result indicates a comparison failure, recording the employee portrait image corresponding to the first image comparison result indicating the comparison failure as a first comparison failure image.

[0061] Specifically, the above description points out that if the similarity of all pictures corresponding to the same employee portrait picture is less than the preset similarity threshold, then the first image comparison result corresponding to the employee portrait picture is a comparison failure, and therefore the employee portrait picture corresponding to the first image comparison result representing the comparison failure can be recorded as the first comparison failure picture.

[0062] S02: determining public place address information according to the company address information, and obtaining a public place portrait picture taken within a preset geographical range where the public place address information is located.

[0063] It can be understood that the company address information pointed out in the above description refers to the company address of the target company, and then in this embodiment, the public place address information can be determined based on the company address information. Among them, the public place address information can be the address information of a place adjacent to the company address, such as a street, a place within the area where the company address belongs, etc. The public place portrait picture is a portrait picture taken by a camera or other camera equipment in the preset geographical range where the public place address information is located. The public place portrait picture can also be obtained by filtering in advance, that is, performing portrait filtering on the captured images taken by the camera in the preset geographical range where the public place address information is located, that is, filtering the images that do not contain portrait features in the captured images taken by the camera in the public place address information, and thus obtaining the public place portrait picture.

[0064] S03: performing image feature comparison between the first comparison failure picture and each of the public place portrait pictures to obtain a second image comparison result between the public place portrait picture and the first comparison failure picture.

[0065] Specifically, after determining the public place address information based on the company address information and obtaining public place portrait images taken within the preset geographical range of the public place address information, the first comparison failure image can be compared with each public place portrait image to obtain a second image comparison result corresponding to the public place portrait image and the first comparison failure image. It can be understood that the method of comparing the image features of the first comparison failure image with each public place portrait image in this embodiment is the same as the method of comparing the image features of the employee portrait image with the portrait image in the above step, and will not be repeated here. Among them, the second image comparison result includes a result indicating a successful comparison, that is, if the second image comparison result of a first comparison failure image indicates a successful comparison, it indicates that at least one public place portrait image has successfully compared with the first comparison failure image; the second image comparison result also includes a result indicating a failed comparison, that is, if the second image comparison result of a first comparison failure image indicates a failed comparison, it indicates that all public place portrait images have failed to compare with the first comparison failure image.

[0066] Furthermore, here, the image feature comparison between the portrait pictures of other employees except the picture that failed the first comparison and the portrait pictures of people in public places is no longer performed. This is because the portrait pictures of other employees except the picture that failed the first comparison have been successfully compared with the portrait pictures in the above steps. Therefore, there is no need to perform image feature comparison again on the portrait pictures of other employees except the picture that failed the first comparison, thereby improving the efficiency of identifying suspected shell companies.

[0067] S04: Determine a second comparison score corresponding to the target company based on a second image comparison result between the public place portrait picture and the first comparison failed picture.

[0068] It is understandable that the above description indicates that the second image comparison results include both results indicating successful comparisons and results indicating failed comparisons. Therefore, it is necessary to determine the total number of first comparison failure images corresponding to the second image comparison results indicating successful comparisons, as well as the total number of first comparison failure images corresponding to the second image comparison results indicating failed comparisons. Furthermore, the second comparison score can be determined based on the total number of first comparison failure images and the total number of first comparison failure images corresponding to the second image comparison results indicating failed comparisons.

[0069] S05: Identify the target company as a suspected shell company based on the second comparison score and the preset identification threshold to obtain a second identification result.

[0070] Specifically, after determining a second comparison score corresponding to the target company based on the second image comparison result between the public place portrait image and the first failed comparison image, the second comparison score is compared with a preset identification threshold. If the second comparison score is greater than or equal to the preset identification threshold, the target company is determined to be a suspected shell company. If the second comparison score is less than the preset identification threshold, it can be determined that the target company is likely not a shell company.

[0071] In this embodiment, for the first failed comparison image that fails to be compared with the portrait image, public place address information is introduced. By performing image feature comparison between the first failed comparison image and the public place portrait image in the public place address information, it is possible to further determine whether the target employee who did not appear near the target company appears in the public place to which the target company belongs. The second image comparison result obtained by performing image feature comparison with the public place portrait image is added to the calculation of the image comparison score, thereby adding a dimension to the image comparison score calculation and further improving the accuracy of the shell company analysis of the target company.

[0072] In one embodiment, determining a second comparison score corresponding to the target company based on a second image comparison result between the public place portrait image and the first comparison failure image includes:

[0073] The first comparison failure picture corresponding to the second image comparison result indicating a comparison failure is recorded as the second comparison failure picture, and the first comparison failure picture corresponding to the second image comparison result indicating a comparison success is recorded as the second comparison success picture.

[0074] It can be understood that the above description points out that the second image comparison result includes a result representing a successful comparison and a result representing a failed comparison. Therefore, the first failed comparison picture corresponding to the second image comparison result representing a failed comparison can be directly recorded as the second failed comparison picture; and the first failed comparison picture corresponding to the second image comparison result representing a successful comparison can be recorded as the second successful comparison picture.

[0075] Obtain a first comparison weight score corresponding to the first image comparison result and a second comparison weight score corresponding to the second image comparison result.

[0076] It is understandable that the first comparison weight score and the second comparison weight score can be configured according to specific needs; for example, assuming that the first comparison weight score and the second comparison weight score are both values ​​greater than 0 and less than 1, then in the current embodiment, when only the first comparison weight score and the second comparison weight score are introduced, the sum of the first comparison weight score and the second comparison weight score is 1, and the first comparison weight score is greater than the second comparison weight score, for example, the first comparison weight score is set to 0.8, the second comparison weight score is set to 0.2, etc. The above is only an example. In this embodiment, the expression form of the first comparison weight score and the second comparison weight score is not limited, and can be a percentage, a score value (such as a score greater than 1 and less than 100), etc.

[0077] The total number of first images of the employee portrait images, the total number of second images of the first images that failed comparison, and the total number of third images of the second images that failed comparison are obtained.

[0078] It can be understood that the total number of first pictures is the total number of employee portrait pictures contained in the shell company analysis instruction in step S10 (the total number of first pictures includes the total number of second pictures of first comparison failure pictures, and the total number of pictures of first successful comparison pictures, and the first successful comparison pictures are the employee portrait pictures corresponding to the first image comparison results representing successful comparison); the total number of second pictures is the total number of pictures of first comparison failure pictures (the total number of second pictures includes the total number of third pictures of second comparison failure pictures, and the total number of pictures of second successful comparison pictures, and the second successful comparison pictures are the first comparison failure pictures corresponding to the second image comparison results representing successful comparison); the total number of third pictures is the total number of pictures of second comparison failure pictures.

[0079] The ratio of the second total number of pictures to the first total number of pictures is recorded as a first picture ratio, and the ratio of the third total number of pictures to the second total number of pictures is recorded as a second picture ratio.

[0080] It can be understood that the first picture ratio is the ratio between the total number of the second pictures and the total number of the second pictures, that is, the proportion of employee portrait pictures that failed the comparison in the total number of employee portrait pictures; the second picture ratio is the ratio between the total number of the third pictures and the total number of the second pictures, that is, the proportion of the first failed comparison pictures that failed the comparison in the total number of the first failed comparison pictures.

[0081] The second comparison score is determined according to the first image ratio, the second image ratio, the first comparison weight score, and the second comparison weight score.

[0082] Specifically, after recording the ratio between the total number of the second pictures and the total number of the first pictures as the first picture ratio, and recording the ratio between the total number of the third pictures and the total number of the second pictures as the second picture ratio, the product of the first picture ratio and the first comparison weight score can be recorded as the first collision value, and the product of the second picture ratio and the second comparison weight score can be recorded as the second collision value, and then the sum of the first collision value and the second collision value can be recorded as the second comparison score.

[0083] It can be understood that the above steps are only an example of determining the second comparison score, and other methods for determining the second comparison score besides the above steps can also be implemented.

[0084] In this embodiment, different weight scores are assigned to the first image comparison result and the second image comparison result, so that the first image comparison result can account for a larger proportion and the second image comparison result accounts for a smaller proportion. This can satisfy the priority of portrait pictures over the priority of portrait pictures in public places, making the final second comparison score more convincing, further improving the accuracy of the second comparison score, and thus improving the accuracy of the shell company analysis of the target company.

[0085] In one embodiment, if Figure 4 As shown, after performing image feature comparison on the first comparison failure picture and the public place portrait picture to obtain a second image comparison result between the public place portrait picture and the first comparison failure picture, the method includes:

[0086] S11: When the second image comparison result indicates a comparison failure, a first comparison failure picture corresponding to the second image comparison result indicating the comparison failure is recorded as a second comparison failure picture.

[0087] It can be understood that the above description points out that the second image comparison result includes a result indicating a comparison failure, so the first comparison failure picture corresponding to the second image comparison result indicating a comparison failure can be directly recorded as the second comparison failure picture.

[0088] S12: Obtain the company platform data of the target company on the preset first-category third-party platform, and perform entity recognition on the company platform data to determine the platform address information of the target company on the preset first-category third-party platform.

[0089] It can be understood that company platform data refers to the data stored by the target company on the first type of third-party platform. The first type of third-party platform may include a recruitment platform, a courier platform, or the like. This first type of third-party platform may contain data related to the target company, such as recruitment information posted by the target company on the recruitment platform, courier delivery and receipt information on the courier platform, and so on. Entity recognition in this embodiment can employ a BiLSTM-CRF model, for example, to perform entity analysis on the company platform data to identify address entities within the company platform data, and then determine the data associated with the address entities as platform address information. Platform address information refers to detailed address data related to the target company recorded on the first type of third-party platform, such as the shipping and receiving addresses in the courier platform data, or the company address in the recruitment information on the recruitment platform. The BiLSTM-CRF model can be pre-trained using a large amount of sample data. For example, this sample data may include data containing different address information. This sample data can then be used to identify the sample data, and the parameters of the BiLSTM-CRF model can be continuously adjusted based on the recognition results, so that the trained BiLSTM-CRF model can accurately identify address entities within the first type of third-party platform data.

[0090] S13: Determine a first data comparison result corresponding to the target company according to the company address information and the platform address information.

[0091] It is understandable that the above description indicates that the company address information is the company address of the target company, and the platform address information is the detailed address in the company platform data of the first type of third-party platform. The company address information and the platform address information can then be matched to determine the first data comparison result corresponding to the target company. Among them, the first data comparison result includes a successful match result, where a successful match means that the target company's address is the same as the address in the first type of third-party platform, and a failed match result, where a failed match means that the target company's address is different from the address in the first type of third-party platform.

[0092] S14: Determine a third comparison score corresponding to the target company according to the first comparison failure image, the second comparison failure image, and the first data comparison result.

[0093] It is understandable that the above description indicates that the first data comparison results include both results indicating a successful match and results indicating a failed match. Therefore, it is necessary to determine the total number of second comparison failure images corresponding to the first data comparison results indicating a successful match, as well as the total number of second comparison failure images corresponding to the first data comparison results indicating a failed match. Furthermore, the third comparison score can be determined based on the total number of first comparison failure images, the total number of second comparison failure images, and the total number of second comparison failure images corresponding to the first data comparison results indicating a failed match.

[0094] Furthermore, in the above steps, a first comparison weight score is assigned to the first image comparison result, and a second comparison weight score is assigned to the second image comparison result. Therefore, in this embodiment, a third collision weight score is assigned to the first data comparison result, and the first comparison weight score and the second comparison weight score in this embodiment are different from when the second comparison score is determined in the above steps. In this embodiment, assuming that the first comparison weight score, the second comparison weight score, and the third collision weight score are all values ​​greater than 0 and less than 1, the sum of the first comparison weight score, the second comparison weight score, and the third collision weight score is 1, and the first comparison weight score is greater than the second comparison weight score and the third collision weight score, while the second comparison weight score and the third collision weight score can be arbitrarily assigned. For example, the first comparison weight score is set to 0.6, the second comparison weight score is set to 0.2, the third collision weight score is set to 0.2, and so on.

[0095] Furthermore, the above description points out that the ratio between the total number of second pictures and the total number of first pictures is recorded as the first picture ratio, and the ratio between the total number of third pictures and the total number of second pictures is recorded as the second picture ratio. In this embodiment, a third dimension is added, that is, the second failed comparison picture corresponding to the first data comparison result representing the matching failure is recorded as the first failed matching picture, and then the total number of first failed matching pictures is obtained, and the ratio between the total number of first failed matching pictures and the total number of third pictures (the total number of third pictures is the total number of second collision pictures) is recorded as the third picture ratio; and then the product of the first picture ratio and the first comparison weight score is recorded as the new first collision value, the product of the second picture ratio and the second comparison weight score is recorded as the new second collision value, and the product of the third picture ratio and the third collision weight score is recorded as the third collision value. In this way, the sum of the new first collision value, the new second collision value and the third collision value is recorded as the third comparison score.

[0096] S15: Identify the target company as a suspected shell company based on the third comparison score and the preset identification threshold to obtain a third identification result.

[0097] Specifically, after determining the third comparison score corresponding to the target company based on the first failed comparison image, the second failed comparison image and the first data comparison result, the third image comparison score is compared with the preset recognition threshold. When the third image comparison score is greater than or equal to the preset recognition threshold, the target company is determined to be a suspected shell company; when the third image comparison score is less than the preset recognition threshold, it is determined that the target company may not be a shell company.

[0098] In this embodiment, for the second failed comparison picture that failed to be compared with the portrait picture in the public place, the company platform data of the first type of third-party platform is introduced, and then the platform address information in the company platform data is matched with the company address information of the target company. It can be further determined whether there is other data in the target company that can detect that the target company is still operating normally (such as the address of the recruitment platform is the same, and the delivery address of the express platform is the same). This can reduce the errors caused by the failure of the employee portrait picture comparison of the target employee due to missed shots due to camera damage, and further improve the accuracy of the shell company analysis of the target company.

[0099] In one embodiment, if Figure 5 As shown, after performing image feature comparison on the first comparison failure picture and the public place portrait picture to obtain a second image comparison result between the public place portrait picture and the first comparison failure picture, the method further includes:

[0100] S21: When the second image comparison result indicates a comparison failure, a first comparison failure picture corresponding to the second image comparison result indicating the comparison failure is recorded as a second comparison failure picture.

[0101] It can be understood that the above description points out that the second image comparison result includes a result indicating a comparison failure, so the first comparison failure picture corresponding to the second image comparison result indicating a comparison failure can be directly recorded as the second comparison failure picture.

[0102] S22: Obtain employee platform data of the target employee corresponding to the second failed comparison image on a preset second-category third-party platform, and perform entity recognition on the employee platform data to determine the employee address information of the target employee corresponding to the second failed comparison image on the preset second-category third-party platform.

[0103] It can be understood that employee platform data refers to the data stored by the target employee in the second type of third-party platform. Among them, the second type of third-party platform data can be a food delivery platform. The second type of third-party platform may contain data related to the target employee, such as the delivery address information contained in the food delivery information of the target employee on the food delivery platform. The entity recognition in this embodiment can use a BiLSTM-CRF model to perform entity analysis on the employee platform data to identify the address entity in the employee platform data, and then determine the data associated with the address entity as employee address information. Among them, the employee address information is the detailed address data related to the target employee recorded in the second type of third-party platform, such as the delivery address in the food delivery data.

[0104] S23: Determine a second data comparison result corresponding to the second comparison failure image according to the company address information and the employee address information.

[0105] It is understandable that the above description indicates that the company address information is the company address of the target company, and the employee address information is the detailed address in the employee platform data of the second type of third-party platform. The company address information and the employee address information can then be matched to determine the second data comparison result corresponding to the second comparison failure image. Among them, the second data comparison result includes a successful match result, where a successful match means that the target company's address is the same as the address in the second type of third-party platform, and a failed match result, where a failed match means that the target company's address is different from the address in the second type of third-party platform.

[0106] S24: Determine a fourth comparison score corresponding to the target company according to the first comparison failure image, the second comparison failure image, and the second data comparison result.

[0107] It is understandable that the above description indicates that the second data comparison results include both results indicating a successful match and results indicating a failed match. Therefore, it is necessary to determine the total number of second comparison failure images corresponding to the second data comparison results indicating a successful match, as well as the total number of second comparison failure images corresponding to the second data comparison results indicating a failed match. Furthermore, the fourth comparison score can be determined based on the total number of second comparison failure images, the total number of second comparison failure images, and the total number of second comparison failure images corresponding to the second data comparison results indicating a failed match.

[0108] Furthermore, in the above steps, a first comparison weight score is assigned to the first image comparison result, and a second comparison weight score is assigned to the second image comparison result. Therefore, in this embodiment, a fourth collision weight score is assigned to the second data comparison result, and the first comparison weight score and the second comparison weight score in this embodiment are different from those when the second comparison score is determined in the above steps. In this embodiment, assuming that the first comparison weight score, the second comparison weight score, and the fourth collision weight score are all values ​​greater than 0 and less than 1, the sum of the first comparison weight score, the second comparison weight score, and the fourth collision weight score is 1, and the first comparison weight score is greater than the second comparison weight score and the fourth collision weight score, while the second comparison weight score and the fourth collision weight score can be arbitrarily assigned. For example, the first comparison weight score is set to 0.7, the second comparison weight score is set to 0.2, the fourth collision weight score is set to 0.1, and so on.

[0109] Furthermore, the above description points out that the ratio between the total number of second pictures and the total number of first pictures is recorded as the first picture ratio, and the ratio between the total number of third pictures and the total number of second pictures is recorded as the second picture ratio. In this embodiment, a third dimension is added, that is, the second failed matching picture corresponding to the second data comparison result representing the matching failure is recorded as the second failed matching picture, and then the total number of second failed matching pictures is obtained, and the ratio between the total number of second failed matching pictures and the total number of third pictures (the total number of third pictures is the total number of second collision pictures) is recorded as the fourth picture ratio; and then the product of the first picture ratio and the first comparison weight score is recorded as the new first collision value, the product of the second picture ratio and the second comparison weight score is recorded as the new second collision value, and the product of the fourth picture ratio and the third collision weight score is recorded as the fourth collision value. In this way, the sum of the new first collision value, the new second collision value and the fourth collision value is recorded as the fourth comparison score.

[0110] S25: Identify the target company as a suspected shell company based on the fourth comparison score and the preset identification threshold to obtain a fourth identification result.

[0111] Specifically, after determining the fourth comparison score corresponding to the target company based on the first failed comparison image, the second failed comparison image and the comparison result with the second data, the fourth image comparison score is compared with the preset recognition threshold. When the fourth image comparison score is greater than or equal to the preset recognition threshold, the target company is determined to be a suspected shell company; when the fourth image comparison score is less than the preset recognition threshold, it is determined that the target company may not be a shell company.

[0112] Furthermore, the employee platform data of the second type of third-party platform introduced in this embodiment may be implemented in combination with the company platform data of the first type of third-party platform introduced in the above embodiment, or may be implemented separately.

[0113] In this embodiment, for the second failed comparison picture that failed to be compared with the portrait picture of a person in a public place, the employee platform data of the second type of third-party platform is introduced, and then the employee address information in the employee platform data is matched with the company address information of the target company. It can be further determined whether there is other data in the target company that can detect that the target company is still operating normally (such as the same delivery address of the food delivery platform). This can reduce the errors caused by the failure of the employee portrait picture comparison of the target employee due to missed shots due to camera damage, and further improve the accuracy of the shell company analysis of the target company.

[0114] It should be understood that the order of execution of the steps in the above embodiments does not necessarily mean the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0115] In one embodiment, a shell company identification device is provided, which corresponds to the shell company identification method in the above embodiment. Figure 6 As shown, the shell company identification device includes an information acquisition module 10, an image feature comparison module 20, a comparison score determination module 30 and a shell company identification module 40. The functional modules are described in detail as follows:

[0116] An information acquisition module 10 is configured to acquire company information of a target company, wherein the company information includes company address information of the target company and a portrait of at least one target employee of the target company;

[0117] An image feature comparison module 20 is configured to obtain a portrait image taken within a preset geographical range of the company address information, and perform image feature comparison between the employee portrait image and the portrait image to obtain a first image comparison result between the portrait image and the employee portrait image;

[0118] a comparison score determination module 30 for determining a first comparison score between the portrait image and the target company based on the first image comparison result between the portrait image and the employee portrait image;

[0119] The shell company identification module 40 is configured to obtain a preset identification threshold, and identify the target company as a suspected shell company based on the first comparison score and the preset identification threshold to obtain a first identification result.

[0120] The specific definitions of the shell company identification device can be found in the definitions of the shell company identification method above and will not be repeated here. Each module in the shell company identification device described above may be implemented in whole or in part through software, hardware, or a combination thereof. Each of the modules described above may be embedded in or independent of a processor in a computer device in hardware form, or may be stored in a computer device memory in software form, so that the processor can call and execute the corresponding operations of each module.

[0121] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 7 As shown. The computer device includes a processor, memory, network interface, and database connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used to store data used in the method for identifying shell companies in the above embodiment. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a method for identifying shell companies is implemented.

[0122] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for identifying shell companies in the above embodiment is implemented.

[0123] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the method for identifying a shell company in the above embodiment is implemented.

[0124] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0125] Those skilled in the art will clearly understand that for the sake of convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0126] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. A method for identifying a shell company, characterized in that: include: Obtaining company information of a target company, the company information including company address information of the target company and a portrait of at least one target employee of the target company; Obtaining a portrait picture taken within a preset geographical range of the company address information, and performing image feature comparison between the employee portrait picture and the portrait picture to obtain a first image comparison result between the portrait picture and the employee portrait picture; determining a first comparison score between the portrait image and the target company based on a first image comparison result between the portrait image and the employee portrait image; Obtaining a preset identification threshold, and identifying the target company as a suspected shell company based on the first comparison score and the preset identification threshold, to obtain a first identification result; After performing image feature comparison between the employee portrait picture and the portrait photo to obtain a first image comparison result between the portrait photo and the employee portrait picture, the method further includes: When the first image comparison result indicates a comparison failure, recording the employee portrait image corresponding to the first image comparison result indicating the comparison failure as a first comparison failure image; Determining public place address information based on the company address information, and obtaining public place portrait images taken within a preset geographical range of the public place address information; performing an image feature comparison between the first comparison-failed picture and the public place portrait picture to obtain a second image comparison result between the public place portrait picture and the first comparison-failed picture; determining a second comparison score corresponding to the target company based on a second image comparison result between the public place portrait image and the first comparison-failed image; Identify the target company as a suspected shell company based on the second comparison score and the preset identification threshold to obtain a second identification result; After performing image feature comparison on the first comparison failure picture and the public place portrait picture to obtain a second image comparison result between the public place portrait picture and the first comparison failure picture, the method further includes: When the second image comparison result indicates a comparison failure, recording the first comparison failure picture corresponding to the second image comparison result indicating the comparison failure as a second comparison failure picture; Obtaining employee platform data of a target employee corresponding to the second image that failed comparison on a preset second-type third-party platform, and performing entity recognition on the employee platform data to determine employee address information of the target employee corresponding to the second image that failed comparison on the preset second-type third-party platform; Determining a second data comparison result corresponding to the second comparison-failed image based on the company address information and the employee address information; determining a fourth comparison score corresponding to the target company based on the first comparison failure image, the second comparison failure image, and the second data comparison result; The target company is identified as a suspected shell company based on the fourth comparison score and the preset identification threshold to obtain a fourth identification result.

2. The method for identifying a shell company according to claim 1, wherein: The performing image feature comparison between the employee portrait picture and the portrait photo to obtain a first image comparison result between the portrait photo and the employee portrait picture includes: Performing portrait clustering on the portrait pictures to obtain at least one portrait category containing a captured portrait picture of the same photographed subject; the captured portrait picture refers to a portrait picture captured from the portrait pictures and containing the photographed subject; For each of the photographed portrait categories, the image similarity between the employee portrait picture and the captured portrait picture in the photographed portrait category is determined, and a first image comparison result between the portrait photograph and the employee portrait picture is determined based on the image similarity.

3. The method for identifying a shell company according to claim 2, wherein: The step of determining, for each of the photographed portrait categories, a picture similarity between the employee portrait picture and a captured portrait picture in the photographed portrait category, and determining a first image comparison result between the portrait picture and the employee portrait picture based on the picture similarity includes: For each of the portrait categories, selecting a portrait picture with the highest definition from all the captured portrait pictures of the portrait category as a picture to be compared for the portrait category; Determining the image similarity between the employee portrait image and the image to be compared of the portrait category; A preset similarity threshold is obtained, and a first image comparison result between the portrait picture and the employee portrait picture is determined according to the picture similarity and the preset similarity threshold.

4. The method for identifying a shell company according to claim 1, wherein: Determining a second comparison score corresponding to the target company based on a second image comparison result between the public place portrait picture and the first comparison failure picture includes: Recording the first comparison failure image corresponding to the second image comparison result indicating a comparison failure as the second comparison failure image, and recording the first comparison failure image corresponding to the second image comparison result indicating a comparison success as the second comparison success image; Obtaining a first comparison weight score corresponding to the first image comparison result and a second comparison weight score corresponding to the second image comparison result; Obtain the total number of first images of the employee portrait images, the total number of second images of the first images that failed comparison, and the total number of third images of the second images that failed comparison; Recording the ratio of the second total number of pictures to the first total number of pictures as a first picture ratio, and recording the ratio of the third total number of pictures to the second total number of pictures as a second picture ratio; The second comparison score is determined according to the first image ratio, the second image ratio, the first comparison weight score, and the second comparison weight score.

5. The method for identifying a shell company according to claim 1, wherein: After performing image feature comparison on the first comparison failure picture and the public place portrait picture to obtain a second image comparison result between the public place portrait picture and the first comparison failure picture, the method further includes: When the second image comparison result indicates a comparison failure, recording the first comparison failure picture corresponding to the second image comparison result indicating the comparison failure as a second comparison failure picture; Obtaining company platform data of the target company on a preset first-category third-party platform, and performing entity recognition on the company platform data to determine platform address information of the target company on the preset first-category third-party platform; Determining a first data comparison result corresponding to the target company according to the company address information and the platform address information; Determining a third comparison score corresponding to the target company based on the first comparison failure image, the second comparison failure image, and the first data comparison result; The target company is identified as a suspected shell company based on the third comparison score and the preset identification threshold to obtain a third identification result.

6. A device for identifying a shell company, characterized in that: include: An information acquisition module is configured to acquire company information of a target company, wherein the company information includes company address information of the target company and a portrait of at least one target employee of the target company; an image feature comparison module, configured to obtain a portrait image taken within a preset geographical range of the company address information, and perform image feature comparison between the employee portrait image and the portrait image to obtain a first image comparison result between the portrait image and the employee portrait image; a comparison score determination module, configured to determine a first comparison score corresponding to the portrait image and the target company based on the first image comparison result between the portrait image and the employee portrait image; a shell company identification module, configured to obtain a preset identification threshold, and identify the target company as a suspected shell company based on the first comparison score and the preset identification threshold, to obtain a first identification result; After performing image feature comparison between the employee portrait picture and the portrait photo to obtain a first image comparison result between the portrait photo and the employee portrait picture, the method further includes: When the first image comparison result indicates a comparison failure, recording the employee portrait image corresponding to the first image comparison result indicating the comparison failure as a first comparison failure image; Determining public place address information based on the company address information, and obtaining public place portrait images taken within a preset geographical range of the public place address information; performing an image feature comparison between the first comparison-failed picture and the public place portrait picture to obtain a second image comparison result between the public place portrait picture and the first comparison-failed picture; determining a second comparison score corresponding to the target company based on a second image comparison result between the public place portrait image and the first comparison-failed image; Identify the target company as a suspected shell company based on the second comparison score and the preset identification threshold to obtain a second identification result; After performing image feature comparison on the first comparison failure picture and the public place portrait picture to obtain a second image comparison result between the public place portrait picture and the first comparison failure picture, the method further includes: When the second image comparison result indicates a comparison failure, recording the first comparison failure picture corresponding to the second image comparison result indicating the comparison failure as a second comparison failure picture; Obtaining employee platform data of a target employee corresponding to the second image that failed comparison on a preset second-type third-party platform, and performing entity recognition on the employee platform data to determine employee address information of the target employee corresponding to the second image that failed comparison on the preset second-type third-party platform; Determining a second data comparison result corresponding to the second comparison-failed image based on the company address information and the employee address information; determining a fourth comparison score corresponding to the target company based on the first comparison failure image, the second comparison failure image, and the second data comparison result; The target company is identified as a suspected shell company based on the fourth comparison score and the preset identification threshold to obtain a fourth identification result.

7. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the method for identifying a shell company according to any one of claims 1 to 5 is implemented.

8. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method for identifying a shell company according to any one of claims 1 to 5 is implemented.

Images