Methods and apparatus for ad labeling, ad filtering and model training

Abstract

This disclosure provides a method and apparatus for ad labeling, ad filtering, and model training, relating to the field of internet technology, specifically deep learning and data labeling, and applicable to scenarios involving the labeling and filtering of web page ads. The specific implementation includes: acquiring a page image corresponding to a target page; inputting the page image into an ad detection model to determine the ad information corresponding to a first ad contained in the page image; determining the code of the first ad based on the ad information; and labeling the first ad based on its code. This disclosure can reduce resource consumption, lower costs, and improve the efficiency of ad labeling and filtering.

Description

Technical Field

[0001] This disclosure relates to the field of Internet technology, specifically to the fields of deep learning and data annotation, and can be applied to scenarios of labeling and filtering web page advertisements. In particular, it relates to a method and apparatus for advertising labeling, advertising filtering and model training. Background Technology

[0002] With the rapid development of information technology, more and more users are obtaining various information through web browsing. However, web pages contain various advertisements, and filtering these advertisements can improve the user's browsing experience.

[0003] The current ad filtering solution relies on experienced software engineers to browse the page and read the webpage source code, or to use web development tools to annotate the ad code on the webpage. When the webpage is displayed, the ads are filtered based on the results of the annotated ad code.

[0004] However, among current ad filtering methods, manual labeling is costly and resource-intensive. Summary of the Invention

[0005] This disclosure provides a method and apparatus for ad labeling, ad filtering, and model training, which can reduce resource consumption, lower costs, and improve the efficiency of ad labeling and ad filtering.

[0006] According to a first aspect of this disclosure, an advertising labeling method is provided, the method comprising:

[0007] Obtain the page image corresponding to the target page; input the page image into the ad detection model, and use the ad detection model to determine the ad information corresponding to the first ad contained in the page image; determine the code of the first ad based on the ad information; and annotate the first ad based on the code of the first ad.

[0008] According to a second aspect of this disclosure, an ad filtering method is provided, the method comprising:

[0009] Obtain the ad labeling information corresponding to the target page. The ad labeling information is generated according to the ad labeling method described in the first aspect. Filter the corresponding ads on the target page based on the ad labeling information.

[0010] According to a third aspect of this disclosure, a model training method is provided, the method comprising:

[0011] Obtain sample page images and sample labels for the sample page images. The sample labels for the sample page images are used to indicate the advertising information corresponding to the advertisements contained in the sample page images. Using the sample page images as input and the sample labels for the sample page images as output, a preset neural network is trained to obtain an advertisement detection model. The advertisement detection model has the function of determining the advertising information corresponding to the advertisements contained in the page images based on the input page images.

[0012] According to a fourth aspect of this disclosure, a model training method is provided, the method comprising:

[0013] The process involves obtaining sample advertisement information and sample tags for the sample advertisement information. The sample tags are used to indicate the code of the sample advertisement corresponding to the sample advertisement information and the code of the advertisement contained in the code nodes related to the code of the sample advertisement information. The related code nodes include at least one of parent nodes, child nodes, and sibling nodes. Using the sample advertisement information as input and the sample tags of the sample advertisement information as output, a preset neural network is trained to obtain an advertisement search model. The advertisement search model has the function of determining the code of the advertisement corresponding to the advertisement information and the code of the advertisement contained in the code nodes related to the code of the advertisement corresponding to the advertisement information based on the input advertisement information.

[0014] According to a fifth aspect of this disclosure, a model training method is provided, the method comprising:

[0015] Obtain sample ad images and their sample labels. The sample labels are used to indicate whether a sample ad image is a real ad image. Use the sample ad images as input and their sample labels as output to train a preset neural network to obtain an ad verification model. The ad verification model has the function of verifying whether an ad image is a real ad image based on the input ad image.

[0016] According to a sixth aspect of this disclosure, an advertising labeling device is provided, the device comprising: an acquisition unit, a determination unit, and a labeling unit.

[0017] The acquisition unit is used to acquire the page image corresponding to the target page; the determination unit is used to input the page image into the ad detection model and determine the ad information corresponding to the first ad contained in the page image through the ad detection model; the determination unit is also used to determine the code of the first ad based on the ad information; the annotation unit is used to annotate the first ad based on the code of the first ad.

[0018] According to a seventh aspect of this disclosure, an advertising filtering device is provided, the device comprising: an acquisition unit and a filtering unit.

[0019] The acquisition unit is used to acquire the advertising labeling information corresponding to the target page, and the advertising labeling information is generated according to the advertising labeling method described in the first aspect; the filtering unit is used to filter the corresponding advertisements in the target page according to the advertising labeling information.

[0020] According to the eighth aspect of this disclosure, a model training apparatus is provided, the apparatus comprising: an acquisition unit and a training unit.

[0021] The acquisition unit is used to acquire sample page images and sample labels of sample page images. The sample labels of sample page images are used to indicate the advertising information corresponding to the advertisements contained in the sample page images. The training unit is used to train a preset neural network with sample page images as input and sample labels of sample page images as output to obtain an advertisement detection model. The advertisement detection model has the function of determining the advertising information corresponding to the advertisements contained in the page images based on the input page images.

[0022] According to a ninth aspect of this disclosure, a model training apparatus is provided, the apparatus comprising: an acquisition unit and a training unit.

[0023] The acquisition unit is used to acquire sample advertisement information and sample tags of the sample advertisement information. The sample tags of the sample advertisement information are used to indicate the code of the sample advertisement corresponding to the sample advertisement information and the code of the advertisement contained in the code node related to the code of the sample advertisement information. The related code node includes at least one of parent node, child node, and sibling node. The training unit is used to train a preset neural network with the sample advertisement information as input and the sample tags of the sample advertisement information as output to obtain an advertisement search model. The advertisement search model has the function of determining the code of the advertisement corresponding to the advertisement information and the code of the advertisement contained in the code node related to the code of the advertisement corresponding to the advertisement information based on the input advertisement information.

[0024] According to a tenth aspect of this disclosure, a model training apparatus is provided, the apparatus comprising: an acquisition unit and a training unit.

[0025] The acquisition unit is used to acquire sample advertising images and sample labels of sample advertising images. The sample labels of sample advertising images are used to indicate whether the sample advertising image is a real advertising image. The training unit is used to train a preset neural network with sample advertising images as input and sample labels of sample advertising images as output to obtain an advertising verification model. The advertising verification model has the function of verifying whether the advertising image is a real advertising image based on the input advertising image.

[0026] According to the eleventh aspect of this disclosure, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the method as described in any one of the first to fifth aspects.

[0027] According to a twelfth aspect of this disclosure, a non-transitory computer-readable storage medium is provided storing computer instructions for causing a computer to perform the method described according to any one of the first to fifth aspects.

[0028] According to a thirteenth aspect of this disclosure, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the method described in any one of the first to fifth aspects.

[0029] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0030] The accompanying drawings are provided to better understand this solution and do not constitute a limitation of this disclosure. Wherein:

[0031] Figure 1 A flowchart illustrating the advertising labeling method provided in this embodiment of the disclosure;

[0032] Figure 2 Another schematic diagram of the advertising labeling method provided in this disclosure embodiment;

[0033] Figure 3 Provided for the embodiments of this disclosure Figure 1 A schematic diagram of an implementation process for S104 in the middle;

[0034] Figure 4 Provided for the embodiments of this disclosure Figure 2 A schematic diagram of one implementation process of S202 in China;

[0035] Figure 5 A schematic diagram illustrating the implementation process of the ad filtering method provided in this embodiment of the disclosure;

[0036] Figure 6 A schematic flowchart of a model training method provided in an embodiment of this disclosure;

[0037] Figure 7 A flowchart illustrating another model training method provided in this embodiment of the disclosure;

[0038] Figure 8A flowchart illustrating yet another model training method provided in this embodiment of the disclosure;

[0039] Figure 9 A schematic diagram of the composition of the advertising labeling device provided in the embodiments of this disclosure;

[0040] Figure 10 A schematic diagram of the composition of the advertising filtering device provided in the embodiments of this disclosure;

[0041] Figure 11 A schematic diagram of the composition of the model training apparatus provided in the embodiments of this disclosure;

[0042] Figure 12 Another schematic diagram of the model training apparatus provided in the embodiments of this disclosure;

[0043] Figure 13 Another schematic diagram of the model training apparatus provided in the embodiments of this disclosure;

[0044] Figure 14 A schematic block diagram of an example electronic device 1400 provided for implementation of embodiments of the present disclosure. Detailed Implementation

[0045] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0046] It should be understood that in the embodiments of this disclosure, the character " / " generally indicates that the preceding and following objects are in an "or" relationship. The terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated.

[0047] With the rapid development of information technology, more and more users are obtaining various information through web browsing. However, web pages contain various advertisements, and filtering these advertisements can improve the user's browsing experience.

[0048] For example, when users browse the web, their browsers are often filled with various non-compliant advertising images. To ensure display and click-through rates, these ads use pop-ups, floating displays, or even obscure key content, misleading users and significantly impacting the browsing experience. To address the problem of non-compliant ads on web pages, various browsers and open-source communities have been working tirelessly to filter low-quality advertisements.

[0049] The current ad filtering solution relies on experienced software engineers to browse the page and read the webpage source code, or to use web development tools to annotate the ad code on the webpage. When the webpage is displayed, the ads are filtered based on the results of the annotated ad code.

[0050] However, among current ad filtering methods, manual labeling is costly and resource-intensive.

[0051] For example, current ad filtering solutions are rule-based, with rules mainly falling into two categories: ad resource-based and ad node-based. The ad node-based approach primarily determines the corresponding HTML tags for low-quality ad images by reading the Hypertext Markup Language (HTML) code in the webpage, then hides or deletes these images. Existing solutions rely heavily on the open-source community and experienced software engineers who manually analyze and determine the corresponding HTML tags for ads using web development tools. This process is time-consuming and labor-intensive, and the annotation results require secondary verification, taking anywhere from several hours to several days. Furthermore, the accuracy of the annotation and verification results is dependent on the engineer's experience, leading to inconsistent quality.

[0052] Against this background, this disclosure provides an ad labeling method that can reduce resource consumption, lower costs, and improve the efficiency of ad labeling, thereby improving the efficiency of ad filtering.

[0053] For example, the entity executing the advertising labeling method can be a computer or server, or other device with data processing capabilities. No limitation is placed on the entity executing this method.

[0054] In some embodiments, the server can be a single server, or it can be a server cluster consisting of multiple servers. In some embodiments, the server cluster can also be a distributed cluster. This disclosure does not limit the specific implementation of the server.

[0055] Figure 1 This is a flowchart illustrating the advertising labeling method provided in an embodiment of this disclosure. Figure 1 As shown, the method may include S101-S104.

[0056] S101. Obtain the page image corresponding to the target page.

[0057] For example, taking a mobile phone accessing a target webpage using an Android operating system, an automated script can be written using a computer programming language (such as Python). This script can then control the phone's access to the target webpage via the Android Debug Bridge (ADB) command in Android Developer Tools. A cloud phone platform can be used to access different pages simultaneously using multiple cloud phone instances. Screenshots of the target page can be captured. This process can generate webview screenshots for ad target detection, identification, and logs needed for subsequent analysis.

[0058] S102. Input the page image into the ad detection model, and use the ad detection model to determine the ad information corresponding to the first ad contained in the page image.

[0059] For example, screenshots of pages containing advertisements can be collected as sample page images, and the advertisements in the screenshots can be labeled to obtain sample labels for the sample page images. The format of the labeled sample images can adopt the format of a visual object classes (VOC) dataset. Using the collected sample page images as input and the labeled sample labels as output, a pre-defined neural network is trained using a YOLO (youonly live once) framework-based object detection algorithm to obtain an advertisement detection model. This model has the function of determining the advertisement information corresponding to the advertisements contained in the input page image. The sample labels of the sample page images indicate the advertisement information corresponding to the advertisements contained in the sample page image, which may include the size, position, and number of advertisements.

[0060] For example, if a screenshot of a page in a browser is taken and input into a trained ad detection model, and if there are ads in the screenshot, the ad detection model can mark the ads contained in the screenshot (for example, by highlighting them with a red box), thereby obtaining ad information such as the location, size, and number of ads.

[0061] For example, this advertising detection model can also be called an object detection model.

[0062] S103. Determine the code of the first advertisement based on the advertisement information.

[0063] For example, after determining the ad's location, size, and other ad information, clicking on any ad area will yield the corresponding ad code.

[0064] For example, once the location of ad 1 is determined, clicking on ad 1 will display the corresponding code on the page.

[0065] S104. Mark the first advertisement according to its code.

[0066] For example, the first advertisement can be annotated based on its code, which contains the advertisement image. The first advertisement can be annotated by marking the code of the first advertisement with a hypertext markup language (HTML) tag.

[0067] For example, once the code for Ad 1 is determined, Ad 1 can be annotated by marking the code for Ad 1 with an HTML tag.

[0068] This disclosure obtains a page image corresponding to a target page, inputs the page image into an ad detection model, and uses the ad detection model to determine the ad information corresponding to a first ad contained in the page image; based on the ad information, it determines the code of the first ad; and based on the code of the first ad, it annotates the first ad. This reduces the resources and costs consumed by ad annotation, improves the efficiency of ad annotation, and can determine the ad information and ad code corresponding to the ad, providing data support for ad filtering. By improving the efficiency of ad annotation, the efficiency of ad filtering is improved.

[0069] Figure 2 This is another schematic diagram of the advertising labeling method provided in an embodiment of this disclosure. For example... Figure 2 As shown, the method may include S201-S202.

[0070] S201. Determine the code of the second advertisement contained in the code node related to the code of the first advertisement.

[0071] For example, the code of a second advertisement can be determined from the code node associated with the code of the first advertisement, wherein the associated code node includes at least one of a parent node, a child node, and a sibling node.

[0072] For example, after determining the code of the first advertisement, you can continue to traverse the parent node, sibling node, and child node of the code node of the first advertisement to determine that the code of the second advertisement is at least one of the parent node, child node, and sibling node.

[0073] S202. Mark the second advertisement according to the code of the second advertisement.

[0074] For example, the code of the second advertisement includes the advertisement image of the second advertisement, and after the code of the second advertisement is determined, the code of the second advertisement can be annotated. The second advertisement may contain multiple advertisements.

[0075] For example, once the code for ad 2 is determined, ad 2 can be annotated by marking the code with an HTML tag.

[0076] This embodiment determines the code of the second advertisement contained in the code node related to the code of the first advertisement, and annotates the second advertisement according to the code of the second advertisement, thereby increasing the number of annotated advertisements and improving the comprehensiveness of the advertisement annotation coverage.

[0077] In some embodiments, S201 may include: inputting advertising information into an advertising search model, and determining the code of a second advertisement through the advertising search model.

[0078] For example, the advertising search model can be a reinforcement learning agent model trained based on a deep Q-network (DQN) algorithm. The acquired advertising information labels are used as output to train a pre-defined neural network, resulting in the advertising search model. The advertising search model can search for at least one of the parent, child, and sibling nodes of the currently acquired advertising information's corresponding advertising code, based on the code's context information. This results in the output of the advertising information's label, which indicates the advertising code corresponding to the advertising information and the advertising code contained in the code nodes related to that advertising code. The related code nodes include at least one of parent, child, and sibling nodes. The trained advertising search model has the function of determining the advertising code corresponding to the input advertising information and the advertising code contained in the code nodes related to that advertising information. Inputting advertising information into the advertising search model allows the model to determine the code of a second advertisement. During the training of the advertising search model, a reward value can be calculated. When the advertising search model finds the correct advertising code, the reward value is positive; when the advertising search model finds the wrong advertising code or fails to find the advertising code, the reward value is negative, thereby improving the accuracy of the advertising search model.

[0079] For example, suppose we obtain the advertising information and tags for Ad 1, Ad 2, and Ad 3, respectively. Using the advertising information of Ad 1, Ad 2, and Ad 3 as input and their tags as output, we train a pre-defined neural network to obtain an ad search model. This model can determine the code of the corresponding ad and the ad code contained in the code nodes related to that ad code based on the input advertising information. When the trained ad search model finds Ad 4, Ad 5, or Ad 6, a positive reward is awarded. When the trained ad search model finds Ad 5, which is not a real ad, or no ad is found, a negative reward is awarded.

[0080] For example, this advertising search model can also be called a reinforcement learning agent model.

[0081] This embodiment trains an advertising search model, inputs advertising information into the advertising search model, and uses the advertising search model to determine the code of the second advertisement, thereby improving the efficiency of determining the code of the second advertisement and thus improving the efficiency of advertising labeling.

[0082] Alternatively, the code for determining the second advertisement can also be implemented in other ways, such as by using other models and algorithms.

[0083] Figure 3 Provided for the embodiments of this disclosure Figure 1 A schematic diagram of one implementation process of S104. For example... Figure 3 As shown, in some embodiments, Figure 1 The S104 shown may include S301-S304.

[0084] S301, Obtain the first advertisement image contained in the code of the first advertisement.

[0085] For example, the code of the first advertisement contains a first advertisement image, and the first advertisement image can be obtained through the code of the first advertisement.

[0086] For example, suppose the code for ad 1 is X. By clicking on the code X of ad 1, you can get the ad image of ad 1 with code X on the webpage.

[0087] S302. Input the first advertising image into the advertising verification model, and verify whether the first advertising image is a real advertising image through the advertising verification model to obtain the verification result of the first advertising image.

[0088] For example, an advertising image and its label are obtained, the label of which indicates whether the advertising image is a real advertising image; the obtained advertising image is used as input and the label of the advertising image is used as output to train a preset neural network to obtain an advertising verification model; wherein, the trained advertising verification model has the function of verifying whether the advertising image is a real advertising image based on the input advertising image.

[0089] For example, assuming that the obtained advertising image 1 is a real advertising image and advertising image 2 is not a real advertising image, then the label of advertising image 1 can be "yes" and the label of advertising image 2 can be "no". By taking advertising image 1 and advertising image 2 as inputs and the labels of advertising image 1 "yes" and advertising image 2 "no" as outputs, a preset neural network can be trained to obtain an advertising verification model. This advertising verification model can verify whether an advertising image is a real advertising image based on the input advertising image.

[0090] For example, this ad verification model can also be called an ad image recognition model.

[0091] S303. Based on the verification result of the first advertisement image, the code of the first advertisement is filtered to obtain the code of the target first advertisement where the real advertisement image is located.

[0092] For example, the code of the first advertisement can be filtered based on the verification result of the first advertisement image to obtain the code of the target first advertisement where the real advertisement image is located. When the first advertisement image is a real advertisement image, the code of the first advertisement image is the code of the target first advertisement where the real advertisement image is located. When the first advertisement image is not a real advertisement image, no mark is made on the code of the first advertisement.

[0093] For example, assuming that the ad image of Ad 1 is a real ad image, then the code of Ad 1 is the code of the target first ad where the real ad image is located. The code of Ad 1 can be recorded to provide a code basis for subsequent annotation of Ad 1.

[0094] S304. Mark the first target ad according to its code.

[0095] For example, the code of the first target advertisement includes the advertisement image of the first target advertisement. After determining that the advertisement image of the first target advertisement is a real advertisement image, the code of the first target advertisement can be annotated.

[0096] For example, if it is determined that the ad image of the target first ad is a real ad image, the target first ad can be annotated by marking the code of the target first ad with an HTML tag.

[0097] This embodiment obtains the first advertisement image contained in the code of the first advertisement, inputs the first advertisement image into the advertisement verification model, verifies whether the first advertisement image is a real advertisement image through the advertisement verification model, obtains the verification result of the first advertisement image, filters the code of the first advertisement based on the verification result of the first advertisement image, obtains the code of the target first advertisement where the real advertisement image is located, and annotates the target first advertisement based on the code of the target first advertisement. This can reduce the annotation of code that is not a real advertisement image and improve the accuracy of advertisement annotation.

[0098] Figure 4 Provided for the embodiments of this disclosure Figure 2 A schematic diagram of one implementation process of S202. For example... Figure 4 As shown, in some embodiments, Figure 2 The S202 shown may include S401-S404.

[0099] S401, Obtain the second advertisement image contained in the code of the second advertisement.

[0100] For example, the code of the second advertisement contains a second advertisement image, which can be obtained through the code of the second advertisement.

[0101] For example, suppose the code for ad 2 is X. By clicking on the code X of ad 2, you can get the ad image of ad 2 with code X on the webpage.

[0102] S402. Input the second advertising image into the advertising verification model, and verify whether the second advertising image is a real advertising image through the advertising verification model to obtain the verification result of the second advertising image.

[0103] For example, the obtained second advertising image can be input into the advertising verification model trained in S302 above, and the advertising verification model can be used to verify whether the second advertising image is a real advertising image to obtain the verification result of the second advertising image.

[0104] For example, assuming that the obtained ad image 1 is a real ad image and ad image 2 is not a real ad image, then if ad image 1 is input into the ad verification model, the ad verification model can output the label "yes" for ad image 1, and if ad image 2 is input into the ad verification model, the ad verification model can output the label "no" for ad image 2.

[0105] S403. Based on the verification result of the second advertisement image, filter the code of the second advertisement to obtain the code of the target second advertisement where the real advertisement image is located.

[0106] For example, the code of the second advertisement can be filtered based on the verification result of the second advertisement image to obtain the code of the target second advertisement where the real advertisement image is located. When the second advertisement image is a real advertisement image, the code of the second advertisement image is the code of the target second advertisement where the real advertisement image is located. When the second advertisement image is not a real advertisement image, no mark is made on the code of the second advertisement.

[0107] For example, assuming that the ad image of Ad 2 is a real ad image, then the code of Ad 2 is the code of the target second ad where the real ad image is located. The code of Ad 2 can be recorded to provide a code basis for subsequent annotation of Ad 2.

[0108] S404. Mark the target second advertisement according to its code.

[0109] For example, the code of the target second advertisement contains the advertisement image of the target second advertisement. After determining that the advertisement image of the target second advertisement is a real advertisement image, the code of the target second advertisement can be annotated.

[0110] For example, if it is determined that the image of the target second advertisement is a real advertisement image, the target second advertisement can be annotated by marking the code of the target second advertisement with an HTML tag.

[0111] This embodiment obtains the second advertisement image contained in the code of the second advertisement, inputs the second advertisement image into the advertisement verification model, verifies whether the second advertisement image is a real advertisement image through the advertisement verification model, obtains the verification result of the second advertisement image, filters the code of the second advertisement based on the verification result of the second advertisement image, obtains the code of the target second advertisement where the real advertisement image is located, and annotates the target second advertisement based on the code of the target second advertisement. This can quickly annotate advertisements on web pages, improve the accuracy of advertisement annotation, and further improve the efficiency of advertisement annotation.

[0112] Based on the advertising labeling method provided in the foregoing embodiments, this disclosure also provides an advertising filtering method. The executing entity of this advertising filtering method can refer to the executing entity of the advertising labeling method described in the foregoing embodiments, and will not be repeated here. Figure 5 This is a schematic diagram illustrating the implementation flow of the ad filtering method provided in this embodiment of the disclosure. Figure 5 As shown, the method may include S501-S502.

[0113] S501. Obtain the ad labeling information corresponding to the target page.

[0114] For example, the advertising labeling information corresponding to the target page can be obtained by collecting the HTML tags of the target page, wherein the advertising labeling information is generated by the advertising labeling method described in the foregoing embodiments.

[0115] For example, if a page has three HTML tags, then the page will have three ads marked. The tagging information for these three ads can be determined based on these three HTML tags.

[0116] S502. Filter the corresponding advertisements on the target page based on the advertisement labeling information.

[0117] For example, during ad labeling, an ad can be hidden as soon as it is labeled, or all labeled ads on the target page can be hidden at once after all corresponding ads on the target page have been labeled, thus filtering the corresponding ads on the target page. After filtering all ads, the ad verification model trained in S302 can be used to verify whether the ad images of the filtered ads are real ad images. If the verification shows that the ad image of a certain ad is not a real ad image, the filtered ad can be restored.

[0118] For example, suppose a page is labeled with Ad 1, Ad 2, Ad 3, Ad 4, and Ad 5. Based on the ad labeling information, all five ads are hidden. Then, the ad verification model is used to verify the ad images of these five ads. The result is that the ad image of Ad 2 is not a real ad image, so Ad 2 is restored.

[0119] This embodiment obtains the advertising labeling information corresponding to the target page, and filters the corresponding advertisements on the target page based on the advertising labeling information. This can reduce the resources and costs consumed by advertising filtering, improve the quality and efficiency of advertising filtering, and enhance the user's browsing experience.

[0120] This disclosure also provides a model training method, which can be used to train the advertisement detection model in the foregoing embodiments. The entity executing this model training method can be a computer or server, or other devices with data processing capabilities. No limitation is placed on the entity executing this method.

[0121] Figure 6 This is a flowchart illustrating a model training method provided in an embodiment of this disclosure. Figure 6 As shown, the method may include S601-S602.

[0122] S601. Obtain the sample page image and the sample label of the sample page image.

[0123] For example, the sample label of the sample page image is used to indicate the advertising information corresponding to the advertisement contained in the sample page image.

[0124] S602. Using sample page images as input and sample labels of sample page images as output, train the preset neural network to obtain an advertisement detection model.

[0125] For example, the specific method for training the ad detection model can be referred to in step S102, which will not be repeated here.

[0126] This embodiment obtains sample page images and sample labels for sample page images, uses the sample page images as input and the sample labels as output, trains a preset neural network to obtain an ad detection model, and the ad detection model can determine the ad information corresponding to the ad contained in the page image based on the input page image.

[0127] This disclosure also provides a model training method, which can be used to train the advertising search model in the foregoing embodiments. The entity executing this model training method can be a computer or server, or other devices with data processing capabilities. No limitation is placed on the entity executing this method.

[0128] Figure 7 This is a flowchart illustrating another model training method provided in an embodiment of this disclosure. Figure 7 As shown, the method may include S701-S702.

[0129] S701. Obtain sample advertising information and sample tags for sample advertising information.

[0130] For example, the sample tag of the sample advertisement information is used to indicate the code of the sample advertisement corresponding to the sample advertisement information, the code of the advertisement contained in the code node related to the code of the sample advertisement, and the related code node includes at least one of parent node, child node, and sibling node.

[0131] S702. Using sample advertising information as input and sample labels of the sample advertising information as output, train the preset neural network to obtain the advertising search model.

[0132] For example, the advertising search model has the function of determining the code of the advertisement corresponding to the input advertising information, and the code of the advertisement contained in the code node related to the code of the advertisement corresponding to the input advertising information, based on the input advertising information. The specific method for training the advertising search model can be referred to the aforementioned method for training advertising search models, and will not be repeated here.

[0133] This embodiment obtains sample advertising information and sample tags of sample advertising information, uses the sample advertising information as input and the sample tags of sample advertising information as output, trains a preset neural network to obtain an advertising search model, and the advertising search model can determine the code of the advertisement corresponding to the input advertising information.

[0134] This disclosure also provides a model training method, which can be used to train the advertising verification model in the foregoing embodiments. The entity executing this model training method can be a computer or server, or other devices with data processing capabilities. No limitation is placed on the entity executing this method.

[0135] Figure 8 This is a flowchart illustrating yet another model training method provided in an embodiment of this disclosure. Figure 8 As shown, the method may include S801-S802.

[0136] S801. Obtain the sample advertisement image and the sample label of the sample advertisement image.

[0137] For example, the sample label of the sample advertisement image is used to indicate whether the sample advertisement image is a real advertisement image.

[0138] S802. Using sample advertisement images as input and sample labels of sample advertisement images as output, train the preset neural network to obtain the advertisement verification model.

[0139] For example, the specific method for training the advertising verification model can be found in step S302, and will not be repeated here.

[0140] This embodiment obtains sample advertisement images and sample labels of sample advertisement images, uses the sample advertisement images as input and the sample labels of sample advertisement images as output, trains a preset neural network to obtain an advertisement verification model, and the advertisement verification model can verify whether the advertisement image is a real advertisement image based on the input advertisement image.

[0141] In an exemplary embodiment, this disclosure also provides an advertising labeling device, which can be used to achieve the aforementioned... Figures 1 to 4 The advertising labeling method in the illustrated embodiment.

[0142] Figure 9 This is a schematic diagram illustrating the composition of the advertising labeling device provided in an embodiment of this disclosure. Figure 9 As shown, the device may include: an acquisition unit 901, a determination unit 902, and a labeling unit 903.

[0143] The acquisition unit 901 is used to acquire the page image corresponding to the target page.

[0144] The determining unit 902 is used to input the page image into the advertising detection model and determine the advertising information corresponding to the first advertisement contained in the page image through the advertising detection model; the determining unit 902 is also used to determine the code of the first advertisement based on the advertising information.

[0145] Labeling unit 903 is used to label the first advertisement according to the code of the first advertisement.

[0146] Optionally, the determining unit 902 is specifically used to determine the code of the second advertisement contained in the code node related to the code of the first advertisement, and the related code node includes at least one of parent node, child node, and sibling node; the labeling unit 903 is also used to label the second advertisement according to the code of the second advertisement.

[0147] Optionally, the determining unit 902 is specifically used to input advertising information into the advertising search model and determine the code of the second advertisement through the advertising search model.

[0148] Optionally, the annotation unit 903 is specifically used to obtain the first advertisement image contained in the code of the first advertisement; input the first advertisement image into the advertisement verification model, verify whether the first advertisement image is a real advertisement image through the advertisement verification model, and obtain the verification result of the first advertisement image; filter the code of the first advertisement based on the verification result of the first advertisement image to obtain the code of the target first advertisement where the real advertisement image is located; and annotate the target first advertisement based on the code of the target first advertisement.

[0149] Optionally, the annotation unit 903 is specifically used for: obtaining the second advertisement image contained in the code of the second advertisement; inputting the second advertisement image into the advertisement verification model, verifying whether the second advertisement image is a real advertisement image through the advertisement verification model, and obtaining the verification result of the second advertisement image; filtering the code of the second advertisement based on the verification result of the second advertisement image to obtain the code of the target second advertisement where the real advertisement image is located; and annotating the target second advertisement based on the code of the target second advertisement.

[0150] In an exemplary embodiment, this disclosure also provides an ad filtering device that can be used to implement the aforementioned... Figure 5 The advertising filtering method in the illustrated embodiment.

[0151] Figure 10 This is a schematic diagram illustrating the composition of an advertising filtering device provided in an embodiment of this disclosure. Figure 10 As shown, the device may include: an acquisition unit 1001 and a filtering unit 1002.

[0152] The acquisition unit 1001 is used to acquire the advertising labeling information corresponding to the target page. The advertising labeling information is generated according to the advertising labeling method described in the foregoing embodiments.

[0153] The filtering unit 1002 is used to filter the corresponding advertisements on the target page based on the advertisement labeling information.

[0154] In an exemplary embodiment, this disclosure also provides a model training apparatus, which can be used to implement the aforementioned... Figure 6 The model training method in the illustrated embodiment.

[0155] Figure 11 This is a schematic diagram illustrating the composition of the model training apparatus provided in an embodiment of this disclosure. Figure 11 As shown, the device may include: an acquisition unit 1101 and a training unit 1102.

[0156] The acquisition unit 1101 is used to acquire a sample page image and a sample tag for the sample page image. The sample tag for the sample page image is used to indicate the advertising information corresponding to the advertisement contained in the sample page image.

[0157] The training unit 1102 is used to train a preset neural network with sample page images as input and sample labels of the sample page images as output to obtain an ad detection model; wherein, the ad detection model has the function of determining the ad information corresponding to the ad contained in the page image based on the input page image.

[0158] In an exemplary embodiment, this disclosure also provides a model training apparatus, which can be used to implement the aforementioned... Figure 7 The model training method in the illustrated embodiment.

[0159] Figure 12 This is another schematic diagram of a model training apparatus provided in an embodiment of this disclosure. (See diagram below.) Figure 11 As shown, the device may include: an acquisition unit 1201 and a training unit 1202.

[0160] The acquisition unit 1201 is used to acquire sample advertisement information and sample tags of sample advertisement information. The sample tags of sample advertisement information are used to indicate the code of the sample advertisement corresponding to the sample advertisement information and the code of the advertisement contained in the code node related to the code of the sample advertisement. The related code node includes at least one of parent node, child node, and sibling node.

[0161] The training unit 1202 is used to train a preset neural network with sample advertising information as input and sample labels of the sample advertising information as output to obtain an advertising search model. The advertising search model has the function of determining the code of the advertisement corresponding to the input advertising information and the code of the advertisement contained in the code node related to the code of the advertisement corresponding to the advertising information, based on the input advertising information.

[0162] In an exemplary embodiment, this disclosure also provides a model training apparatus, which can be used to implement the aforementioned... Figure 8 The model training method in the illustrated embodiment.

[0163] Figure 13 This is another schematic diagram of a model training apparatus provided in an embodiment of this disclosure. (See diagram below.) Figure 11 As shown, the device may include: an acquisition unit 1301 and a training unit 1302.

[0164] The acquisition unit 1301 is used to acquire a sample advertisement image and a sample label of the sample advertisement image. The sample label of the sample advertisement image is used to indicate whether the sample advertisement image is a real advertisement image.

[0165] The training unit 1302 is used to train a preset neural network with sample advertising images as input and sample labels of the sample advertising images as output to obtain an advertising verification model; wherein, the advertising verification model has the function of verifying whether the advertising image is a real advertising image based on the input advertising image.

[0166] The acquisition, storage, and application of user personal information involved in the technical solution disclosed herein comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0167] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0168] In an exemplary embodiment, an electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the method as described in the above embodiments.

[0169] In an exemplary embodiment, the readable storage medium may be a non-transitory computer-readable storage medium storing computer instructions for causing a computer to perform the method described in the above embodiments.

[0170] In an exemplary embodiment, the computer program product includes a computer program that, when executed by a processor, implements the method described in the above embodiments.

[0171] Figure 14A schematic block diagram of an example electronic device 1400 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0172] like Figure 14 As shown, the electronic device 1400 includes a computing unit 1401, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 1402 or a computer program loaded from a storage unit 1408 into a random access memory (RAM) 1403. The RAM 1403 may also store various programs and data required for the operation of the electronic device 1400. The computing unit 1401, ROM 1402, and RAM 1403 are interconnected via a bus 1404. An input / output (I / O) interface 1405 is also connected to the bus 1404.

[0173] Multiple components in electronic device 1400 are connected to I / O interface 1405, including: input unit 1406, such as keyboard, mouse, etc.; output unit 1407, such as various types of monitors, speakers, etc.; storage unit 1408, such as disk, optical disk, etc.; and communication unit 1409, such as network card, modem, wireless transceiver, etc. Communication unit 1409 allows electronic device 1400 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0174] The computing unit 1401 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 1401 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 1401 performs the various methods and processes described above, such as ad labeling, ad filtering, and model training methods. For example, in some embodiments, the ad labeling, ad filtering, and model training methods can be implemented as computer software programs tangibly contained in a machine-readable medium, such as storage unit 1408. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 1400 via ROM 1402 and / or communication unit 1409. When the computer program is loaded into RAM 1403 and executed by the computing unit 1401, one or more steps of the ad labeling, ad filtering, and model training methods described above can be performed. Alternatively, in other embodiments, the computing unit 1401 may be configured in any other suitable manner (e.g., by means of firmware) to perform ad labeling, ad filtering, and model training methods.

[0175] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0176] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0177] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0178] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0179] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with embodiments of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0180] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.

[0181] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0182] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. An advertising labeling method, the method comprising: Get the page image corresponding to the target page; The page image is input into the ad detection model, and the ad detection model is used to determine the ad information corresponding to the first ad contained in the page image. Based on the advertising information, determine the code of the first advertisement; The first advertisement is labeled according to its code; The method involves determining the code of a second advertisement contained in a code node related to the code of the first advertisement, wherein the related code node includes at least one of a parent node, a child node, and a sibling node; the method includes: inputting the advertisement information into an advertisement search model, determining the code of the second advertisement through the advertisement search model; the advertisement search model searches for at least one of the parent node, child node, and sibling node of the code of the advertisement corresponding to the current advertisement information based on the currently obtained advertisement code and the environmental information in which the advertisement code corresponding to the current advertisement information is located, thereby outputting a tag for the advertisement information, wherein the tag for the advertisement information is used to indicate the code of the advertisement corresponding to the advertisement information and the code of the advertisement contained in the code node related to the advertisement code; The second advertisement is labeled according to its code.

2. The method according to claim 1, wherein labeling the first advertisement according to the code of the first advertisement includes: Obtain the first advertisement image contained in the code of the first advertisement; The first advertisement image is input into the advertisement verification model, and the advertisement verification model verifies whether the first advertisement image is a real advertisement image, and the verification result of the first advertisement image is obtained. Based on the verification result of the first advertisement image, the code of the first advertisement is filtered to obtain the code of the target first advertisement where the real advertisement image is located. The target first advertisement is labeled according to its code.

3. The method according to claim 1, wherein labeling the second advertisement according to the code of the second advertisement includes: The code that retrieves the second advertisement image contained in the second advertisement; The second advertisement image is input into the advertisement verification model, and the advertisement verification model verifies whether the second advertisement image is a real advertisement image, and the verification result of the second advertisement image is obtained. Based on the verification result of the second advertisement image, the code of the second advertisement is filtered to obtain the code of the target second advertisement where the real advertisement image is located; The target second advertisement is labeled according to its code.

4. An ad filtering method, the method comprising: Obtain the advertising labeling information corresponding to the target page, wherein the advertising labeling information is generated by the method according to any one of claims 1-3; Based on the advertising labeling information, the corresponding advertisements on the target page are filtered.

5. A model training method, the method comprising: Obtain a sample page image and its sample tags, wherein the sample tags are used to indicate the advertising information corresponding to the advertisement contained in the sample page image; Using the sample page image as input and the sample label of the sample page image as output, a preset neural network is trained to obtain an advertisement detection model; The ad detection model has the function of determining the ad information corresponding to the ad contained in the page image based on the input page image, and the ad detection model is used for ad detection of the page image corresponding to the target page in the method of claim 1.

6. A model training method, the method comprising: Obtain sample advertisement information and sample tags of the sample advertisement information. The sample tags of the sample advertisement information are used to indicate the code of the sample advertisement corresponding to the sample advertisement information and the code of the advertisement contained in the code node related to the code of the sample advertisement. The related code node includes at least one of parent node, child node, and sibling node. Using the sample advertisement information as input and the sample labels of the sample advertisement information as output, a preset neural network is trained to obtain an advertisement search model; The advertising search model has the function of determining the code of the advertisement corresponding to the input advertising information and the code of the advertisement contained in the code node related to the code of the advertisement corresponding to the advertising information, and the advertising search model is used to determine the second advertising code in the method of claim 1.

7. A model training method, the method comprising: Obtain a sample advertisement image and a sample tag for the sample advertisement image, wherein the sample tag is used to indicate whether the sample advertisement image is a real advertisement image; Using the sample advertisement image as input and the sample label of the sample advertisement image as output, a preset neural network is trained to obtain an advertisement verification model; The advertising verification model has the function of verifying whether the input advertising image is a real advertising image; the advertising verification model is used in the method of claim 2.

8. An advertising labeling device, the device comprising: The acquisition unit is used to acquire the page image corresponding to the target page. The determining unit is used to input the page image into the ad detection model and determine the ad information corresponding to the first ad contained in the page image through the ad detection model. The determining unit is further configured to determine the code of the first advertisement based on the advertisement information; The annotation unit is used to annotate the first advertisement according to the code of the first advertisement; The determining unit is specifically used for: The method involves determining the code of a second advertisement contained in a code node related to the code of the first advertisement, wherein the related code node includes at least one of a parent node, a child node, and a sibling node; specifically, it involves: inputting the advertisement information into an advertisement search model, determining the code of the second advertisement through the advertisement search model; the advertisement search model searches for at least one of the parent node, child node, and sibling node of the code of the advertisement corresponding to the current advertisement information based on the currently obtained advertisement code and the environmental information in which the advertisement code corresponding to the current advertisement information is located, thereby outputting a tag for the advertisement information, wherein the tag for the advertisement information is used to indicate the code of the advertisement corresponding to the advertisement information and the code of the advertisement contained in the code node related to the advertisement code; The annotation unit is also used to annotate the second advertisement according to the code of the second advertisement.

9. The apparatus according to claim 8, wherein the labeling unit is specifically used for: Obtain the first advertisement image contained in the code of the first advertisement; The first advertisement image is input into the advertisement verification model, and the advertisement verification model verifies whether the first advertisement image is a real advertisement image, and the verification result of the first advertisement image is obtained. Based on the verification result of the first advertisement image, the code of the first advertisement is filtered to obtain the code of the target first advertisement where the real advertisement image is located. The target first advertisement is labeled according to its code.

10. The apparatus according to claim 8, wherein the marking unit is specifically used for: The code that retrieves the second advertisement image contained in the second advertisement; The second advertisement image is input into the advertisement verification model, and the advertisement verification model verifies whether the second advertisement image is a real advertisement image, and the verification result of the second advertisement image is obtained. Based on the verification result of the second advertisement image, the code of the second advertisement is filtered to obtain the code of the target second advertisement where the real advertisement image is located; The target second advertisement is labeled according to its code.

11. An ad filtering device, the device comprising: An acquisition unit is used to acquire advertising labeling information corresponding to a target page, wherein the advertising labeling information is generated by the advertising labeling device according to any one of claims 8-10; The filtering unit is used to filter the corresponding advertisements on the target page based on the advertisement labeling information.

12. A model training apparatus, the apparatus comprising: The acquisition unit is used to acquire a sample page image and a sample tag of the sample page image, wherein the sample tag of the sample page image is used to indicate the advertising information corresponding to the advertisement contained in the sample page image; The training unit is used to train a preset neural network with the sample page image as input and the sample label of the sample page image as output to obtain an advertisement detection model. The ad detection model has the function of determining the ad information corresponding to the ad contained in the page image based on the input page image, and the ad detection model is used for ad detection of the page image corresponding to the target page in the method of claim 1.

13. A model training apparatus, the apparatus comprising: The acquisition unit is used to acquire sample advertising information and sample tags of the sample advertising information. The sample tags of the sample advertising information are used to indicate the code of the sample advertisement corresponding to the sample advertising information and the code of the advertisement contained in the code node related to the code of the sample advertisement. The related code node includes at least one of parent node, child node, and sibling node. The training unit is used to train a preset neural network with the sample advertising information as input and the sample labels of the sample advertising information as output to obtain an advertising search model. The advertising search model has the function of determining the code of the advertisement corresponding to the input advertising information and the code of the advertisement contained in the code node related to the code of the advertisement corresponding to the advertising information, and the advertising search model is used to determine the second advertising code in the method of claim 1.

14. A model training apparatus, the apparatus comprising: The acquisition unit is used to acquire a sample advertisement image and a sample tag of the sample advertisement image, wherein the sample tag of the sample advertisement image is used to indicate whether the sample advertisement image is a real advertisement image; The training unit is used to train a preset neural network with the sample advertisement image as input and the sample label of the sample advertisement image as output to obtain an advertisement verification model. The advertising verification model has the function of verifying whether the input advertising image is a real advertising image; the advertising verification model is used in the method of claim 2.

15. An electronic device comprising: At least one processor; and a memory communicatively connected to the at least one processor; The memory stores instructions executable by the at least one processor, which are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1-3, or the method according to claim 4, or the method according to any one of claims 5-7.

16. A non-transitory computer-readable storage medium storing computer instructions for causing a computer to perform the method according to any one of claims 1-3, or the method according to claim 4, or the method according to any one of claims 5-7.

17. A computer program product comprising a computer program that, when executed by a processor, implements the method according to any one of claims 1-3, or the method according to claim 4, or the method according to any one of claims 5-7.

Images