An illegal advertisement processing method, device and electronic equipment

Abstract

The application provides a method, device and electronic equipment for processing illegal advertisements, wherein the method comprises: obtaining a plurality of newly launched advertisements; grouping similar advertisements in the plurality of advertisements to obtain at least two advertisement groups, wherein each of the at least two advertisement groups comprises at least two similar advertisements; and marking the advertisement group with an illegal identifier when an advertisement in the advertisement group is determined as an illegal advertisement. Through the method, device and electronic equipment for processing illegal advertisements provided by the application, similar advertisements in newly launched advertisements are grouped, and the advertisement group is marked with an illegal identifier when an advertisement in the advertisement group is determined as an illegal advertisement, so that the review of illegal advertisements can be performed in groups, and the processing efficiency of illegal advertisements is greatly improved.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a method, device and electronic device for processing illegal advertisements. Background Art

[0002] At present, among the advertisements released every day, there are often a large number of illegal advertisements released, which cause adverse effects. In order to discover the illegal advertisements among the advertisements released every day, it is necessary to review the content of the advertisements released. Currently, illegal advertisements are usually reviewed manually, and the review efficiency is low. Summary of the Invention

[0003] To solve the above problems, the embodiments of the present invention aim to provide a method, device and electronic device for processing illegal advertisements.

[0004] In a first aspect, an embodiment of the present invention provides a method for processing illegal advertisements, comprising:

[0005] Get multiple newly placed ads;

[0006] Grouping advertisements with similar contents among the plurality of advertisements into one group to obtain at least two advertisement groups, wherein each advertisement group in the at least two advertisement groups includes: at least two similar advertisements;

[0007] When one advertisement in an advertisement group is determined to be an illegal advertisement, the advertisement group is marked with an illegal identifier.

[0008] In a second aspect, an embodiment of the present invention further provides a device for processing illegal advertisements, including:

[0009] The acquisition module is used to obtain multiple newly placed advertisements;

[0010] a grouping module, configured to group advertisements with similar contents among the plurality of advertisements into a group to obtain at least two advertisement groups, wherein each of the at least two advertisement groups includes: at least two similar advertisements;

[0011] The processing module is configured to mark the advertisement group with a violation identifier when one advertisement in the advertisement group is determined to be a violation advertisement.

[0012] In a third aspect, an embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method described in the first aspect are executed.

[0013] In a fourth aspect, an embodiment of the present invention further provides an electronic device, comprising a memory, a processor, and one or more programs, wherein the one or more programs are stored in the memory and configured so that the processor executes the steps of the method described in the first aspect above.

[0014] In the solution provided by the first aspect of the embodiment of the present invention, similar advertisements in newly released advertisements are grouped to obtain advertisement groups composed of similar advertisements. When one advertisement in an advertisement group is determined to be an illegal advertisement, the advertisement group is marked with a violation identifier. Compared with the method of manually reviewing illegal advertisements in related technologies, similar advertisements in newly released advertisements can be grouped, and when one advertisement in an advertisement group is determined to be an illegal advertisement, the advertisement group is marked with a violation identifier, so that illegal advertisements can be reviewed in groups, which greatly improves the processing efficiency of illegal advertisements.

[0015] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. 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 work.

[0017] Figure 1 A flowchart of a method for processing illegal advertisements provided by Example 1 of the present invention is shown;

[0018] Figure 2 A schematic diagram showing the structure of an illegal advertisement processing device provided by embodiment 2 of the present invention is shown;

[0019] Figure 3 A schematic structural diagram of an electronic device provided by Embodiment 3 of the present invention is shown. DETAILED DESCRIPTION

[0020] In the description of the present application, it needs to be understood that the terms "center", "longitudinal", "transverse", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise" and the like indicate the orientation or positional relationship shown in the drawings, which are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the present application.

[0021] In addition, the terms "first", "second" are only for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the technical features referred to. Therefore, the features defined with "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "multiple" is two or more, unless otherwise explicitly specified and limited.

[0022] In the present application, unless otherwise explicitly specified and limited, the terms "mounting", "connection", "connection", "fixing" and the like should be understood broadly, for example, it can be fixed connection, or detachable connection, or integrally connected; it can be mechanical connection, or electrical connection; it can be directly connected, or indirectly connected through intermediate medium, or the communication inside two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0023] At present, among the advertisements put out every day, a large number of illegal advertisements are often put out, causing adverse effects. In order to find illegal advertisements among the advertisements put out every day, the content of the advertisements put out needs to be examined. At present, the illegal advertisements are usually examined by manual, and the examination efficiency is low.

[0024] Based on this, the embodiment provides an illegal advertisement processing method and device and electronic equipment, similar advertisements in newly put out advertisements are grouped to obtain an advertisement group composed of similar advertisements, when an advertisement in the advertisement group is determined as an illegal advertisement, the advertisement group is marked with an illegal identifier, so that the examination of illegal advertisements can be performed in groups, and the processing efficiency of illegal advertisements is greatly improved.

[0025] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0026] Embodiment 1

[0027] The embodiment provides an illegal advertisement processing method, and the execution subject of the illegal advertisement processing method is a server.

[0028] See also Figure 1 The flowchart of a method for handling illegal advertisements is shown. This embodiment provides a method for handling illegal advertisements, including the following specific steps:

[0029] Step 100: Acquire multiple newly placed advertisements.

[0030] In the above step 100, when the time length since the last illegal advertisement processing reaches a preset time length, the server obtains multiple advertisements newly placed on the advertising system between the time point when the last illegal advertisement processing was performed and the current time point.

[0031] The advertising system may be run in the server or in a computing device that interacts with the server.

[0032] Each of the multiple advertisements is an advertisement with video as main content.

[0033] Step 102: Group the advertisements with similar contents among the plurality of advertisements into one group to obtain at least two advertisement groups, wherein each of the at least two advertisement groups includes at least two similar advertisements.

[0034] In the above step 102, in order to group the advertisements with similar contents among the plurality of advertisements, the following specific steps (1) to (5) are included:

[0035] (1) extracting text information of each advertisement from the plurality of advertisements, and processing the text information of each advertisement using a word2vec model to obtain a text feature vector of each advertisement;

[0036] (2) Extract the image information of each advertisement respectively, and use the InceptionV3 model to process the image information of each advertisement to obtain the image feature vector of each advertisement;

[0037] (3) extracting video frames of each advertisement respectively, processing the video frames of each advertisement, and obtaining a video feature vector of each advertisement;

[0038] (4) respectively calculating a first similarity of text feature vectors between each advertisement, a second similarity of image feature vectors between each advertisement, and a third similarity of video feature vectors between each advertisement;

[0039] (5) Determine the advertisements whose first similarity, second similarity, and third similarity are all greater than the similarity threshold as advertisements with similar contents, and group the advertisements with similar contents into the same advertisement group.

[0040] In the above step (1), the specific process of extracting the text information of each of the multiple advertisements is prior art and will not be described in detail here.

[0041] After obtaining the text information of each advertisement, first, the text information of each advertisement is translated into English text information using the translation software running on the server itself; then, the English text information of each advertisement is input into the word2vec model, and the English text information of each advertisement is processed using the word2vec model to obtain the text feature vector of each advertisement.

[0042] The specific process of using the word2vec model to process the English text information of each advertisement to obtain the text feature vector of each advertisement is prior art and will not be described in detail here.

[0043] In the above step (2), the specific process of extracting the image information of each advertisement is prior art and will not be described in detail here.

[0044] The InceptionV3 model is used to process the image information of each advertisement. Specifically, the image feature vector is extracted by returning it through convolution operations (convolution layer) and pooling layers; eigenvalues ​​are optimized through fully connected layers; 3. Image similarity is calculated using the eigenvalues ​​to classify the images (classification); thereby obtaining the image feature vector of each advertisement. The specific process of obtaining the image feature vector of each advertisement is existing technology and will not be repeated here.

[0045] In the above step (3), in order to process the video frames of each advertisement and obtain the video feature vector of each advertisement, the following steps (31) to (34) may be performed:

[0046] (31) extracting the video frames of each advertisement according to the playback order of each advertisement and obtaining the playback duration of each advertisement;

[0047] (32) extracting the video frames in each advertisement that are located at a preset ratio of the playback duration, and extracting the first video frame and the second to last video frame of each advertisement;

[0048] (33) Using the InceptionV3 model, the extracted video frames located at the preset proportion position of the playback time, the first video frame of each advertisement, and the penultimate video frame of each advertisement are processed respectively to obtain the feature vectors of the video frames located at the preset proportion position of the playback time, the feature vectors of the first video frame of each advertisement, and the feature vectors of the penultimate video frame of each advertisement;

[0049] (34) Calculate the average value of the feature vectors of the video frames located at a preset ratio of the playback duration in each advertisement, the feature vectors of the first video frame of each advertisement, and the feature vectors of the second to last video frame of each advertisement, and use the calculated results as the video feature vectors of each advertisement.

[0050] In the above step (31), the specific process of extracting the video frames of each advertisement according to the playback order of each advertisement is a prior art and will not be repeated here.

[0051] The playing time of each advertisement is extracted by the server from each advertisement. The specific extraction process is prior art and will not be described in detail here.

[0052] In the above step (32), a plurality of preset playback duration ratio positions can be set in the server, and then the video frames at the plurality of preset playback duration ratio positions need to be extracted.

[0053] In one embodiment, the preset proportion positions of the playback time may include but are not limited to: the position of 10% of the playback time, the position of 20% of the playback time, the position of 30% of the playback time, the position of 40% of the playback time, the position of 50% of the playback time, the position of 60% of the playback time, the position of 70% of the playback time, the position of 80% of the playback time and the position of 90% of the playback time.

[0054] Of course, the preset ratio position of the playback time can also be set to positions of other ratios of the playback time, which will not be described here one by one.

[0055] For example, the video frames at a preset proportion of the playback time in each advertisement are extracted, that is, the video frames at 10% of the playback time, the video frames at 20% of the playback time, the video frames at 30% of the playback time, the video frames at 40% of the playback time, the video frames at 50% of the playback time, the video frames at 60% of the playback time, the video frames at 70% of the playback time, the video frames at 80% of the playback time and the video frames at 90% of the playback time are extracted.

[0056] The specific process of the server extracting the video frame is prior art and will not be described in detail here.

[0057] In step (33), the InceptionV3 model is used to process the video frame located at the preset proportion of the playing time, the first video frame and the second last video frame of each advertisement, respectively, to obtain the feature vectors of the video frame located at the preset proportion of the playing time, the first video frame and the second last video frame of each advertisement. The specific implementation process is known in the art and will not be described here.

[0058] In step (34), the average values of the feature vectors of the video frame located at the preset proportion of the playing time, the first video frame and the second last video frame of each advertisement are calculated. The specific process is known in the art and will not be described here.

[0059] In step (4), when calculating the first similarity of the text feature vectors between the advertisements, the cosine similarity between the text feature vectors of the advertisements can be calculated to obtain the first similarity of the text feature vectors between the advertisements.

[0060] The specific process of calculating the cosine similarity between the text feature vectors of the advertisements is known in the art and will not be described here.

[0061] The specific calculation processes of the second similarity of the picture feature vectors between the advertisements and the third similarity of the video feature vectors between the advertisements are similar to the specific process of the first similarity of the text feature vectors between the advertisements, and will not be described here.

[0062] In step (5), the similarity threshold is cached in the server. The similarity threshold can be set to any value between 0.6 and 0.9, and will not be described here.

[0063] In an embodiment, the similarity threshold can be set to 0.8.

[0064] That is, if the first similarity, the second similarity and the third similarity of two advertisements are all greater than 0.8, the two advertisements are determined to be similar in content.

[0065] After the advertisement grouping of similar advertisements is completed through step 102, the following step 104 can be performed to determine whether the advertisements in the advertisement grouping are illegal advertisements.

[0066] In step 104, when an advertisement in the advertisement grouping is determined to be an illegal advertisement, the advertisement grouping is marked with an illegal identifier.

[0067] Specifically, the above step 104 may perform the following steps (1) to (5):

[0068] (1) Obtaining text feature vectors, image feature vectors, and video feature vectors of illegal advertisements;

[0069] (2) selecting a target advertisement from at least two similar advertisements in each of the advertisement groups;

[0070] (3) calculating a fourth similarity between the text feature vector of the target advertisement in each of the advertisement groups and the text feature vector of the illegal advertisement, calculating a fifth similarity between the image feature vector of the target advertisement in each of the advertisement groups and the image feature vector of the illegal advertisement, and calculating a sixth similarity between the video feature vector of the target advertisement in each of the advertisement groups and the video feature vector of the illegal advertisement;

[0071] (4) when the fourth similarity, the fifth similarity, and the sixth similarity are all greater than the illegal image similarity threshold, determining that the target advertisement in the advertisement group is determined to be an illegal advertisement;

[0072] (5) Using a violation flag, mark the advertisement group to which the target advertisement belongs.

[0073] In the above step (1), the text feature vector, the image feature vector and the video feature vector of the illegal advertisement are stored in an illegal advertisement feature library, and the illegal advertisement feature library is set in the server.

[0074] In the above step (2), a target advertisement is selected from at least two similar advertisements in each advertisement group by random selection.

[0075] In the above step (3), the specific process of calculating the fourth similarity between the text feature vector of the target advertisement in each of the advertisement groups and the text feature vector of the illegal advertisement, calculating the fifth similarity between the image feature vector of the target advertisement in each of the advertisement groups and the image feature vector of the illegal advertisement, and calculating the sixth similarity between the video feature vector of the target advertisement in each of the advertisement groups and the video feature vector of the illegal advertisement is similar to the specific process of calculating the first similarity of the text feature vectors between the advertisements in the above step 102, and will not be repeated here.

[0076] In the above step (4), the similarity threshold of the illegal images can be set to any value between 0.8 and 0.95.

[0077] In one embodiment, the violation image similarity threshold may be set to 0.9.

[0078] In the above step (5), the violation identifier is cached in the server.

[0079] After an advertisement group is marked with the violation identifier, the advertisements in the advertisement group marked with the violation identifier are sent to staff for secondary confirmation of the violation advertisements to ensure the accuracy of identifying the violation advertisements. Moreover, only the advertisements in the advertisement group marked with the violation identifier are sent to staff for confirmation, which greatly improves the efficiency of handling violation advertisements.

[0080] To sum up, this embodiment proposes a method for processing illegal advertisements, which groups similar advertisements in newly released advertisements to obtain advertisement groups composed of similar advertisements. When one advertisement in an advertisement group is determined to be an illegal advertisement, the advertisement group is marked with a violation identifier. Compared with the method of manually reviewing illegal advertisements in related technologies, similar advertisements in newly released advertisements can be grouped, and when one advertisement in an advertisement group is determined to be an illegal advertisement, the advertisement group is marked with a violation identifier, so that illegal advertisements can be reviewed in groups, which greatly improves the processing efficiency of illegal advertisements.

[0081] Example 2

[0082] This embodiment provides an illegal advertisement processing device, which is used to execute the illegal advertisement processing method proposed in the above-mentioned embodiment 1.

[0083] See also Figure 2 The structure diagram of a device for processing illegal advertisements is shown in FIG. This embodiment provides a device for processing illegal advertisements, including:

[0084] An acquisition module 200 is used to acquire multiple newly placed advertisements;

[0085] A grouping module 202 is configured to group advertisements with similar content among the plurality of advertisements to obtain at least two advertisement groups, wherein each advertisement group in the at least two advertisement groups includes at least two similar advertisements;

[0086] The processing module 204 is configured to mark the advertisement group with a violation identifier when one advertisement in the advertisement group is determined to be a violation advertisement.

[0087] Specifically, the grouping module is configured to group advertisements with similar contents among the plurality of advertisements into one group, including:

[0088] extracting text information of each advertisement from the plurality of advertisements, and processing the text information of each advertisement using a word2vec model to obtain a text feature vector of each advertisement;

[0089] Extract the image information of each advertisement separately, and use the InceptionV3 model to process the image information of each advertisement to obtain the image feature vector of each advertisement;

[0090] Extracting video frames of each advertisement respectively, processing the video frames of each advertisement, and obtaining a video feature vector of each advertisement;

[0091] respectively calculating a first similarity of text feature vectors between each advertisement, a second similarity of image feature vectors between each advertisement, and a third similarity of video feature vectors between each advertisement;

[0092] The advertisements whose first similarity, second similarity and third similarity are all greater than a similarity threshold are determined as advertisements with similar contents, and the advertisements with similar contents are grouped into the same advertisement group.

[0093] Specifically, the grouping module is used to extract the video frames of each advertisement respectively, process the video frames of each advertisement, and obtain the video feature vector of each advertisement, including:

[0094] Extract the video frames of each advertisement according to the playback order of each advertisement, and obtain the playback duration of each advertisement;

[0095] Extracting the video frames at a preset ratio of the playback duration of each advertisement, and extracting the first video frame and the second to last video frame of each advertisement;

[0096] The InceptionV3 model is used to process the extracted video frames at the preset proportion of the playback time of each advertisement, the first video frame of each advertisement, and the second-to-last video frame of each advertisement, respectively, to obtain the feature vectors of the video frames at the preset proportion of the playback time of each advertisement, the feature vectors of the first video frame of each advertisement, and the feature vectors of the second-to-last video frame of each advertisement;

[0097] Calculate the average of the feature vectors of the video frames at the preset ratio of the playback duration in each advertisement, the feature vector of the first video frame of each advertisement, and the feature vector of the second to last video frame of each advertisement, and use the calculated results as the video feature vector of each advertisement.

[0098] Specifically, the processing module is specifically used to:

[0099] Obtaining text feature vectors, image feature vectors, and video feature vectors of illegal advertisements;

[0100] selecting a target advertisement from at least two similar advertisements in each of the advertisement groups;

[0101] Calculating a fourth similarity between the text feature vector of the target advertisement in each of the advertisement groups and the text feature vector of the illegal advertisement, calculating a fifth similarity between the image feature vector of the target advertisement in each of the advertisement groups and the image feature vector of the illegal advertisement, and calculating a sixth similarity between the video feature vector of the target advertisement in each of the advertisement groups and the video feature vector of the illegal advertisement;

[0102] When the fourth similarity, the fifth similarity, and the sixth similarity are all greater than the illegal image similarity threshold, determining that the target advertisement in the advertisement group is determined to be an illegal advertisement;

[0103] The advertisement group to which the target advertisement belongs is marked with a violation identifier.

[0104] In summary, this embodiment proposes a device for processing illegal advertisements, which groups similar advertisements among newly released advertisements to obtain advertisement groups composed of similar advertisements. When one advertisement in an advertisement group is determined to be an illegal advertisement, the advertisement group is marked with a violation identifier. Compared with the method of manually reviewing illegal advertisements in related technologies, similar advertisements among newly released advertisements can be grouped, and when one advertisement in an advertisement group is determined to be an illegal advertisement, the advertisement group is marked with a violation identifier, so that illegal advertisements can be reviewed in groups, thereby greatly improving the processing efficiency of illegal advertisements.

[0105] Example 3

[0106] This embodiment provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the steps of the method for handling illegal advertisements described in the above embodiment 1 are executed. The specific implementation can be found in the method embodiment 1 and will not be described in detail here.

[0107] In addition, see Figure 3 The electronic device shown in FIG. 1 is a structural diagram of an electronic device. This embodiment further provides an electronic device, which includes a bus 51 , a processor 52 , a transceiver 53 , a bus interface 54 , a memory 55 , and a user interface 56 . The electronic device includes the memory 55 .

[0108] In this embodiment, the electronic device further includes: one or more programs stored in the memory 55 and executable on the processor 52, and is configured to execute the one or more programs by the processor to perform the following steps (1) to (3):

[0109] (1) Obtain multiple newly placed advertisements;

[0110] (2) grouping advertisements with similar contents among the plurality of advertisements to obtain at least two advertisement groups, wherein each advertisement group in the at least two advertisement groups includes at least two similar advertisements;

[0111] (3) When one advertisement in an advertisement group is determined to be an illegal advertisement, the advertisement group is marked with an illegal identification.

[0112] The transceiver 53 is configured to receive and send data under the control of the processor 52 .

[0113] The bus architecture (represented by bus 51) may include any number of interconnected buses and bridges. Bus 51 links various circuits, including one or more processors represented by processor 52 and memory represented by memory 55. Bus 51 may also link various other circuits, such as peripheral devices, voltage regulators, and power management circuits. These are well known in the art and are therefore not further described in this embodiment. Bus interface 54 provides an interface between bus 51 and transceiver 53. Transceiver 53 may be a single component or multiple components, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. For example, transceiver 53 receives external data from other devices. Transceiver 53 is used to transmit data processed by processor 52 to other devices. Depending on the nature of the computing system, a user interface 56, such as a keypad, display, speaker, microphone, or joystick, may also be provided.

[0114] The processor 52 is responsible for managing the bus 51 and general processing, such as running the general operating system mentioned above, while the memory 55 can be used to store data used by the processor 52 when performing operations.

[0115] Optionally, the processor 52 may be, but is not limited to: a central processing unit, a single chip microcomputer, a microprocessor or a programmable logic device.

[0116] It is to be appreciated that the memory 55 in the embodiments of the present application can be a volatile memory or a nonvolatile memory, or can include both volatile and nonvolatile memory. Among them, the nonvolatile memory can be a Read-Only Memory (ROM), a Programmable ROM (PROM), an Erasable PROM (EPROM), an Electrically EPROM (EEPROM), or a flash memory. The volatile memory can be a Random Access Memory (RAM) used as an external cache. By way of example, and not limitation, many forms of RAM can be used, such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDR SDRAM), Enhanced SDRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct Rambus RAM (DRRAM). The memory 55 of the system and method described in the embodiments is intended to include, without being limited to, these and any other suitable types of memory.

[0117] In some embodiments, the memory 55 stores the following elements, executable modules or data structures, or a subset thereof, or an extended set thereof: an operating system 551 and an application program 552.

[0118] Among them, the operating system 551 contains various system programs, such as a framework layer, a core library layer, a driver layer, etc., for implementing various basic services and processing hardware-based tasks. The application program 552 contains various application programs, such as a media player (Media Player), a browser (Browser), etc., for implementing various application services. The program for implementing the method of the embodiments of the present application can be contained in the application program 552.

[0119] In summary, this embodiment proposes a computer-readable storage medium and an electronic device, which group similar advertisements in newly released advertisements to obtain an advertisement group composed of similar advertisements. When one advertisement in an advertisement group is determined to be an illegal advertisement, the advertisement group is marked with a violation identifier. Compared with the method of manually reviewing illegal advertisements in related technologies, similar advertisements in newly released advertisements can be grouped, and when one advertisement in an advertisement group is determined to be an illegal advertisement, the advertisement group is marked with a violation identifier, so that illegal advertisements can be reviewed in groups, thereby greatly improving the processing efficiency of illegal advertisements.

[0120] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A method for handling illegal advertisements, characterized in that: include: Get multiple newly placed ads; Grouping advertisements with similar contents among the plurality of advertisements into one group to obtain at least two advertisement groups, wherein each advertisement group in the at least two advertisement groups includes: at least two similar advertisements; The grouping of advertisements with similar contents from the plurality of advertisements includes: extracting text information from each of the plurality of advertisements, and processing the text information of each advertisement using a word2vec model to obtain a text feature vector for each advertisement; extracting image information from each advertisement, and processing the image information of each advertisement using an InceptionV3 model to obtain an image feature vector for each advertisement; extracting video frames from each advertisement, processing the video frames of each advertisement to obtain a video feature vector for each advertisement; calculating a first similarity between the text feature vectors of each advertisement, a second similarity between the image feature vectors of each advertisement, and a third similarity between the video feature vectors of each advertisement; determining advertisements whose first similarity, second similarity, and third similarity are all greater than a similarity threshold as advertisements with similar contents, and grouping the advertisements with similar contents into the same advertisement group; When one advertisement in the advertisement group is determined to be an illegal advertisement, marking the advertisement group with an illegal identification; The extracting of video frames of each advertisement and processing the video frames of each advertisement to obtain a video feature vector of each advertisement include: Extract the video frames of each advertisement according to the playback order of each advertisement, and obtain the playback duration of each advertisement; Extracting video frames at preset proportion positions of the playback time of each advertisement, and extracting the first video frame of each advertisement and the penultimate video frame of each advertisement; the preset proportion positions of the playback time include any one or more of the following: a position at 10% of the playback time, a position at 20% of the playback time, a position at 30% of the playback time, a position at 40% of the playback time, a position at 50% of the playback time, a position at 60% of the playback time, a position at 70% of the playback time, a position at 80% of the playback time, and a position at 90% of the playback time; The InceptionV3 model is used to process the extracted video frames at the preset proportion of the playback time of each advertisement, the first video frame of each advertisement, and the second-to-last video frame of each advertisement, respectively, to obtain the feature vectors of the video frames at the preset proportion of the playback time of each advertisement, the feature vectors of the first video frame of each advertisement, and the feature vectors of the second-to-last video frame of each advertisement; The feature vectors of the video frames at the preset proportion of the playback duration of each advertisement, the feature vectors of the first video frame of each advertisement, and the average feature vectors of the second to last video frame of each advertisement are calculated, and the calculated results are used as the video feature vectors of each advertisement.

2. The method according to claim 1, characterized in that When one advertisement in the advertisement group is determined to be an illegal advertisement, marking the advertisement group with an illegal identifier includes: Obtaining text feature vectors, image feature vectors, and video feature vectors of illegal advertisements; selecting a target advertisement from at least two similar advertisements in each of the advertisement groups; Calculating a fourth similarity between the text feature vector of the target advertisement in each of the advertisement groups and the text feature vector of the illegal advertisement, calculating a fifth similarity between the image feature vector of the target advertisement in each of the advertisement groups and the image feature vector of the illegal advertisement, and calculating a sixth similarity between the video feature vector of the target advertisement in each of the advertisement groups and the video feature vector of the illegal advertisement; When the fourth similarity, the fifth similarity, and the sixth similarity are all greater than the illegal image similarity threshold, determining that the target advertisement in the advertisement group is determined to be an illegal advertisement; The advertisement group to which the target advertisement belongs is marked with a violation identifier.

3. A device for processing illegal advertisements, characterized in that: include: The acquisition module is used to obtain multiple newly placed advertisements; a grouping module, configured to group advertisements with similar contents among the plurality of advertisements into a group to obtain at least two advertisement groups, wherein each of the at least two advertisement groups includes: at least two similar advertisements; The grouping module is configured to group advertisements with similar contents among the plurality of advertisements into one group, and includes: extracting text information of each advertisement from the plurality of advertisements, and processing the text information of each advertisement using a word2vec model to obtain a text feature vector of each advertisement; Extract the image information of each advertisement separately, and use the InceptionV3 model to process the image information of each advertisement to obtain the image feature vector of each advertisement; Extracting video frames of each advertisement respectively, processing the video frames of each advertisement, and obtaining a video feature vector of each advertisement; respectively calculating a first similarity of text feature vectors between each advertisement, a second similarity of image feature vectors between each advertisement, and a third similarity of video feature vectors between each advertisement; determining, among the advertisements, advertisements for which the first similarity, the second similarity, and the third similarity are all greater than a similarity threshold as advertisements with similar contents, and grouping the advertisements with similar contents into the same advertisement group; a processing module, configured to mark the advertisement group with a violation identifier when one advertisement in the advertisement group is determined to be an illegal advertisement; The grouping module is configured to extract video frames of each advertisement respectively, process the video frames of each advertisement, and obtain a video feature vector of each advertisement, including: Extract the video frames of each advertisement according to the playback order of each advertisement, and obtain the playback duration of each advertisement; Extracting the video frames at a preset ratio of the playback duration of each advertisement, and extracting the first video frame and the second to last video frame of each advertisement; The InceptionV3 model is used to process the extracted video frames at the preset proportion of the playback time of each advertisement, the first video frame of each advertisement, and the second-to-last video frame of each advertisement, respectively, to obtain the feature vectors of the video frames at the preset proportion of the playback time of each advertisement, the feature vectors of the first video frame of each advertisement, and the feature vectors of the second-to-last video frame of each advertisement; The feature vectors of the video frames at the preset proportion of the playback duration of each advertisement, the feature vectors of the first video frame of each advertisement, and the average feature vectors of the second to last video frame of each advertisement are calculated, and the calculated results are used as the video feature vectors of each advertisement.

4. The device according to claim 3, characterized in that The processing module is specifically used to: Obtaining text feature vectors, image feature vectors, and video feature vectors of illegal advertisements; selecting a target advertisement from at least two similar advertisements in each of the advertisement groups; Calculating a fourth similarity between the text feature vector of the target advertisement in each of the advertisement groups and the text feature vector of the illegal advertisement, calculating a fifth similarity between the image feature vector of the target advertisement in each of the advertisement groups and the image feature vector of the illegal advertisement, and calculating a sixth similarity between the video feature vector of the target advertisement in each of the advertisement groups and the video feature vector of the illegal advertisement; When the fourth similarity, the fifth similarity, and the sixth similarity are all greater than the illegal image similarity threshold, determining that the target advertisement in the advertisement group is determined to be an illegal advertisement; The advertisement group to which the target advertisement belongs is marked with a violation identifier.

5. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 2 are executed.

6. An electronic device, characterized in that: The electronic device includes a memory, a processor, and one or more programs, wherein the one or more programs are stored in the memory and are configured so that the processor executes the steps of the method according to any one of claims 1 to 2.

Images