An Internet advertising risk detection system and method based on big data architecture

Abstract

The present invention discloses an Internet advertising risk detection system and method based on a big data architecture, including data collection; text and image scanning; text keyword extraction and image capture analysis; keyword screening and detection; image screening and detection; sensitive word interception and sensitive image interception; sensitive word storage and sensitive image storage; advertiser matching and storage; extracting advertiser names; data aggregation; advertising delivery. The present invention adopts intelligent extraction and analysis to intercept and filter sensitive words and images, prevent illegal advertisements from being delivered to the Internet, aggregate the filtered text and images, ensure the integrity of compliant and legal advertisements, and mark illegal advertisers for storage, conduct key detection when the advertiser subsequently delivers advertisements, facilitate platform supervision, effectively reduce the probability of illegal advertisements entering the public's field of vision, and greatly improve the compliance of platform advertising delivery.

Description

Technical Field

[0001] The present invention relates to the field of network big data application technology, and specifically to an Internet advertising risk detection system and method based on big data architecture. Background Art

[0002] Big data applications need to process large amounts of information, and the size of the information becomes larger and larger when the data is replicated to multiple locations for elasticity. However, the most important attribute of big data is not its size, but its ability to split large jobs into many small jobs, which can distribute the resources for processing a task to multiple locations for parallel processing.

[0003] Especially for Internet advertising, after the platform receives the advertiser's request to place an advertisement, in order to ensure the legality and compliance of the advertising content, the platform needs to detect the information involved in the advertisement to prevent illegal and sensitive words from appearing in the advertisement, so as to reduce the chance of illegal and illegal advertisements appearing in the public eye.

[0004] Existing Internet advertising risk detection systems usually only detect the entire advertising file, which may easily lead to some sensitive words or sensitive areas in the image escaping the platform's interception and filtering, and fail to highlight advertisers that violate laws and regulations, causing such advertisers to not pay attention to the legality and compliance of the advertising content. If such advertisers continue to place advertisements in the future, it will also increase the chances of illegal and irregular advertising content being displayed on the Internet, greatly reducing the compliance of the platform's advertising. Summary of the invention

[0005] The purpose of the present invention is to provide an Internet advertising risk detection system and method based on big data architecture to solve the problems raised in the above background technology.

[0006] To achieve the above-mentioned purpose, the present invention provides the following technical solutions: an Internet advertising risk detection system based on a big data architecture, including data collection, the data collection is for text data and picture data, and the collected data is stored and summarized;

[0007] Scan the text and pictures in the database, scan the entire advertisement file, and make a preliminary screening of the text and pictures designed in the advertisement;

[0008] Extract keywords from texts, extract texts from advertisements in sections; capture and analyze pictures, and capture pictures from advertisements in sections;

[0009] Keyword screening and detection: compare the segmented extracted keywords with the preset word library in the library to detect compliant and illegal words; image screening and detection: compare the captured images with the preset images in the library to detect compliant and illegal images;

[0010] Sensitive word blocking, blocking illegal words; sensitive picture blocking, blocking illegal pictures;

[0011] Sensitive words and images are saved. Violating sensitive words and images are saved in the library, and advertisers of illegal advertisements are matched and stored in the library.

[0012] Extract the names of advertisers, extract the names of advertisers that violate the regulations, and add them to the scanning library for marking as the key detection targets for subsequent advertisements;

[0013] Data aggregation: aggregate compliant texts and pictures to generate complete advertising files;

[0014] Advertisement delivery: by delivering the aggregated advertising files, compliant advertisements can be displayed on the Internet.

[0015] In a further embodiment, the data collection is data collection of advertisements placed by advertisers on the Internet, and the data consists of text data and image data.

[0016] In a further embodiment, the text and image scanning is a holistic analysis of all advertisements in the platform server.

[0017] In a further embodiment, the text keyword extraction is to segment the text portion in the advertisement, and the text segmentation is to divide the entire text into several phrases. The image capture analysis is to partition and capture the image portion in the advertisement, and the partition capture is to divide the entire image into several blocks of the same size.

[0018] In a further embodiment, the keyword screening and detection are performed by comparing the extracted keywords with the vocabulary preset by the platform, and deeply distinguishing between compliant and illegal phrases. The image screening and detection are performed by comparing the captured image blocks with the image library preset by the platform, and deeply distinguishing between compliant and illegal images.

[0019] In a further embodiment, the sensitive words and pictures are stored, and after detection and filtering, the sensitive words and pictures are stored in a database.

[0020] In a further embodiment, the advertiser matching and storage is to match the advertiser's name with the stored data to achieve a one-to-one correspondence between the advertisement file and the advertiser.

[0021] In a further embodiment, the extracting of advertiser names extracts and marks the names of the offending advertisers through the platform media library, and then transmits them back to the platform server.

[0022] In a further embodiment, the data aggregation and advertisement delivery, after interception and filtering, aggregates the compliant text information and image information to ensure the integrity of the compliant advertisement files, and then delivers them to the Internet for display.

[0023] Preferably, the detection method based on the above-mentioned Internet advertising risk detection system based on big data architecture includes the following steps:

[0024] A1. The platform collects text and image information from the advertisements and puts the collected data into the platform database for scanning and analysis. A preliminary analysis is conducted to determine whether the overall advertisement file violates any regulations or laws.

[0025] A2. Segment the text information to extract short phrases, and then perform in-depth analysis and screening on the extracted phrases; segment the image information to extract the segmented image blocks, and then perform in-depth analysis and screening on the extracted blocks;

[0026] A3. Compare the extracted phrases with the words in the word library to quickly obtain sensitive words and block them. Compare the screenshot image blocks with the images in the library to quickly obtain sensitive images and block them.

[0027] A4. Save the intercepted sensitive words and images to the database, extract the names of illegal advertisers from the database, send the extracted advertiser names back to the platform media library and mark them. In subsequent advertising detection, focus on the products of illegal advertisers to reduce the pressure on the platform to identify illegal advertisements;

[0028] A5: Aggregate the data of the intercepted and filtered compliant text and information to ensure the integrity of the compliant advertising files, and then put the compliant advertisements as a whole on the Internet platform for display, so as to ensure that the advertisements displayed to the public are legal and compliant.

[0029] Compared with the prior art, the present invention has the following beneficial effects:

[0030] The present invention adopts intelligent extraction and analysis, gives priority to overall analysis of advertising files, and then intercepts and filters sensitive words and images to prevent illegal advertisements from being placed on the Internet. The filtered text and images are aggregated to ensure the integrity of compliant and legal advertisements, and illegal advertisers are marked for entry into the database. When the advertisers subsequently place advertisements, key inspections are carried out to facilitate platform supervision, effectively reduce the probability of illegal advertisements entering the public eye, and greatly improve the compliance of platform advertising. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 A flowchart of an Internet advertising risk detection system and method based on a big data architecture is provided in the present invention. DETAILED DESCRIPTION

[0032] The following will be described clearly and completely in conjunction with the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0033] Example

[0034] See also Figure 1 ,This embodiment provides an Internet advertising risk detection system and method based on big data architecture, including data collection, characterized in that: data collection is performed for text data and image data, and the collected data is stored and summarized to ensure that the platform database can perform preliminary analysis on the advertising information;

[0035] Scanning text and images in the database, scanning the entire advertisement file, can initially screen the text and images designed in the advertisement, and then feed back all the information involved in the advertisement to the backend terminal, distinguishing its legality and compliance in the overall structure of the advertisement, and allowing the platform to grasp the overall content of the advertisement file;

[0036] Text keyword extraction: extract text from advertisements in segments. Segmented text extraction can prevent text information from being too long and difficult to detect accurately, thus reducing the chance of missed detection. Image capture and analysis: segment images in advertisements and divide image information into a large number of blocks, which can quickly eliminate compliant parts. Only sensitive blocks need to be detected, thus reducing the difficulty of platform detection.

[0037] Keyword screening and detection: compare the segmented extracted keywords with the preset word library in the library to detect compliant and illegal words; image screening and detection: compare the captured images with the preset images in the library to detect compliant and illegal images. The comparison method can reduce the computing pressure of the server and reduce the platform maintenance cost;

[0038] Sensitive word blocking, blocking illegal words, sensitive picture blocking, blocking illegal pictures, filtering and blocking sensitive information, effectively preventing illegal and illegal advertising information from being placed on the Internet;

[0039] Sensitive word storage and sensitive image storage, save the sensitive words and images that violate the law into the library, match the advertisers of the illegal advertisements into the library, and after the illegal information is saved in the library, the information of the illegal advertisers can be permanently stored in the server terminal;

[0040] Extract the names of advertisers, extract the names of advertisers that violate the regulations, and add them to the scanning library for marking. They will be used as the key detection targets for subsequent advertisements. Targeted detection will be carried out on the advertising products placed by the marked advertisers. If such advertisers place advertisements again, their information can be quickly obtained, and they will be subject to key detection during the initial screening process, so as to encourage such advertisers to improve the content of their advertisements and move closer to the basic direction of legality and compliance.

[0041] Data aggregation: aggregate compliant text and images to generate complete advertising files, ensuring that legal and compliant advertisements can be fully displayed on the Internet;

[0042] Advertising delivery, by delivering the aggregated advertising files, can display compliant advertisements on the Internet, ensuring that advertisements that enter the public eye are legal and compliant, and effectively reducing the chances of risky advertisements entering the public eye.

[0043] Among them, the data collection is the data collection of advertisements placed by advertisers on the Internet. The data consists of text data and image data. The advertisement-related information is continuously collected and transmitted to the platform server to ensure that the platform can effectively track and detect the placed advertisements, avoid the phenomenon of fault detection, and reduce the probability of missed detection of advertising files.

[0044] The text and image scanning is to conduct an overall analysis of all advertisements in the platform server, read the advertisement information delivered to the platform and feed the read information back to the platform server, so as to obtain a detailed understanding of the advertisement content delivered by the advertiser.

[0045] Among them, the text keyword extraction is to segment the text part in the advertisement. The text segmentation is to divide the entire text into several phrases, and divide the paragraphs composed of text into several phrases, so as to facilitate the storage and comparison of text information. The image capture and analysis is to partition and capture the image part in the advertisement. The partition capture is to divide the entire image into several blocks of the same size. After the image is divided into blocks, the size of the image can be reduced, which is convenient for comparison and analysis of the blocks.

[0046] Among them, the keyword screening and detection compares the extracted keywords with the preset vocabulary of the platform, deeply distinguishes between compliant and illegal phrases, and facilitates the capture of illegal texts. The image screening and detection compares the preset image library of the platform with the intercepted image blocks, deeply distinguishes between compliant and illegal images, marks the illegal blocks, and quickly analyzes the compliance of the entire image.

[0047] Among them, the sensitive word storage and sensitive picture storage, after detection and filtering, the sensitive text and picture parts are stored in the database. After the sensitive words and sensitive pictures are saved, they can be fed back to advertisers, which can not only urge advertisers to make timely improvements, but also retain sufficient evidence to facilitate cooperation between the platform and advertisers. If the advertiser raises an objection, the platform can provide sufficient evidence and guarantee its own interests.

[0048] Among them, the advertiser matching and warehousing is to match the name of the advertiser through the data stored in the database, to achieve a one-to-one correspondence between the advertising file and the advertiser, and to save the information of advertisers with violations in the platform database. If such advertisers still violate the regulations when placing advertisements in the subsequent period, the platform can implement warnings and blacklisting measures, which can not only reduce the detection pressure of the platform, but also purify the purity of Internet advertising.

[0049] The process of extracting advertiser names involves extracting and marking the names of illegal advertisers through the platform media library, and then transmitting them back to the platform server. If illegal advertisers subsequently place advertisements, the platform can promptly extract the products of such advertisers separately and conduct key inspections to prevent advertisers from optimizing to evade platform inspections, thereby further reducing the chance of risky advertisements entering the Internet.

[0050] Among them, the data aggregation and advertising delivery, after interception and filtering, aggregates the compliant text information and image information to ensure the integrity of the compliant advertising files, and then delivers them to the Internet for display. This can not only reduce the chance of illegal and irregular advertisements entering the Internet, but also ensure that legal and compliant advertising products can be displayed on the Internet in a timely manner.

[0051] Working principle: The platform collects text and image information in the advertisement, and puts the collected data into the platform database for scanning and analysis. It initially analyzes whether the overall advertisement file violates regulations and laws, and segments the text information to extract short phrases, which are then deeply analyzed and screened. The image information is partitioned and intercepted to extract segmented image blocks, which are then deeply analyzed and screened. The extracted phrases are compared with the text in the thesaurus, which can quickly obtain sensitive words and intercept them. The screenshot image blocks are compared with the images in the gallery, which can quickly obtain sensitive images and conduct The intercepted sensitive words and images are saved in the database, the names of the illegal advertisers are extracted from the database, the extracted advertiser names are sent back to the platform media library and marked, and in the subsequent advertising detection, the products of the illegal advertisers are tested intensively to reduce the pressure on the platform to identify illegal advertisements, and the data of the compliant text and information after interception and filtration is summarized to ensure the integrity of the compliant advertising files, and then the compliant advertisements are delivered to the Internet platform for display as a whole, so as to ensure that the advertisements displayed in the public eye are legal and compliant, effectively reduce the chances of illegal and illegal advertisements being displayed in the public eye, and greatly improve the compliance of the platform's advertising.

[0052] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An Internet advertising risk detection system based on big data architecture, including data collection, characterized in that: Data collection collects text data and image data, and summarizes the collected data into the database; Scan the text and pictures in the database, scan the entire advertisement file, and make a preliminary screening of the text and pictures designed in the advertisement; Extract keywords from texts, extract texts from advertisements in sections; capture and analyze pictures, and capture pictures from advertisements in sections; Keyword screening and detection: compare the segmented extracted keywords with the preset word library in the library to detect compliant and illegal words; image screening and detection: compare the captured images with the preset images in the library to detect compliant and illegal images; Sensitive word blocking, blocking illegal words; sensitive picture blocking, blocking illegal pictures; Sensitive words and images are saved. Violating sensitive words and images are saved in the library, and advertisers of illegal advertisements are matched and stored in the library. Extract the names of advertisers, extract the names of advertisers that violate the regulations, and add them to the scanning library for marking as the key detection targets for subsequent advertisements; Data aggregation: aggregate compliant texts and pictures to generate complete advertising files; Advertisement delivery: by delivering the aggregated advertising files, compliant advertisements can be displayed on the Internet.

2. The Internet advertising risk detection system based on big data architecture according to claim 1 is characterized by: The data collection is to collect data from advertisements placed by advertisers on the Internet, and the data consists of text data and image data.

3. The Internet advertising risk detection system based on big data architecture according to claim 1 is characterized by: The text and image scanning is an overall analysis of all advertisements in the platform server.

4. The Internet advertising risk detection system based on big data architecture according to claim 1 is characterized by: The text keyword extraction is to segment the text part in the advertisement, and the text segmentation is to divide the entire text into several phrases. The image capture analysis is to partition the image part in the advertisement, and the partition capture is to divide the entire image into several blocks of the same size.

5. The Internet advertising risk detection system based on big data architecture according to claim 1 is characterized by: The keyword screening and detection is performed by comparing the platform preset vocabulary with the extracted keywords, and deeply distinguishing between compliant and illegal phrases. The image screening and detection is performed by comparing the platform preset gallery with the intercepted image blocks, and deeply distinguishing between compliant and illegal images.

6. The Internet advertising risk detection system based on big data architecture according to claim 1 is characterized by: The sensitive words and pictures are stored, and after detection and filtering, the sensitive words and pictures are stored in the database.

7. The Internet advertising risk detection system based on big data architecture according to claim 1 is characterized by: The advertiser matching storage is to match the advertiser's name with the stored data to achieve a one-to-one correspondence between the advertisement file and the advertiser.

8. The Internet advertising risk detection system based on big data architecture according to claim 1 is characterized by: The extracting of advertiser names extracts and marks the names of illegal advertisers through the platform media library, and then transmits them back to the platform server.

9. The Internet advertising risk detection system based on big data architecture according to claim 1 is characterized by: The data aggregation and advertisement delivery, after interception and filtering, aggregates the compliant text information and image information to ensure the integrity of the compliant advertisement files, and then delivers them to the Internet for display.

10. A detection method for an Internet advertising risk detection system based on a big data architecture, according to the Internet advertising risk detection system based on a big data architecture according to claims 1-9, characterized in that: The steps include: A1. The platform collects text and image information from the advertisements and puts the collected data into the platform database for scanning and analysis. A preliminary analysis is conducted to determine whether the overall advertisement file violates any regulations or laws. A2. Segment the text information to extract short phrases, and then perform in-depth analysis and screening on the extracted phrases; segment the image information to extract the segmented image blocks, and then perform in-depth analysis and screening on the extracted blocks; A3. Compare the extracted phrases with the words in the word library to quickly obtain sensitive words and block them. Compare the screenshot image blocks with the images in the library to quickly obtain sensitive images and block them. A4. Save the intercepted sensitive words and images to the database, extract the names of illegal advertisers from the database, send the extracted advertiser names back to the platform media library and mark them. In subsequent advertising detection, focus on the products of illegal advertisers to reduce the pressure on the platform to identify illegal advertisements; A5: Aggregate the data of the intercepted and filtered compliant text and information to ensure the integrity of the compliant advertising files, and then put the compliant advertisements as a whole on the Internet platform for display, so as to ensure that the advertisements displayed to the public are legal and compliant.

Images