JP7145443B1 - ADVERTISING REVIEW SYSTEM, ADVERTISING REVIEW METHOD, AND ADVERTISING REVIEW PROGRAM - Google Patents

Abstract

The present invention provides an advertisement screening system and the like that streamlines the screening of advertisement videos by making it possible for examiners to easily use screening results based on a learning model of advertisement videos. A second pre-examination processing unit (61) in an advertisement examination server (3) of an advertisement examination system extracts a plurality of image components included in an image. The second video screening unit 63 generates a video-related determination result by determining whether each of the plurality of extracted video components is appropriate as an advertisement based on the learning model. The second examination result output unit 65 displays an image display area for displaying a still image including an inappropriate element that is a video component judged to be inappropriate by the second video examination unit 63. outputting a review result screen including a first review result display area for displaying the determination result for each of the elements and a second review result display area for inputting comments by the reviewer for each of the inappropriate elements regarding the still image; . [Selection drawing] Fig. 5

Description

The present disclosure relates to an advertisement examination system, an advertisement examination method, and an advertisement examination program for examining whether a video produced for advertisement is suitable as an advertisement.

In transportation facilities, buildings, outdoor spaces, etc. in public facilities and commercial facilities, images produced for advertisement are broadcast on digital signage such as liquid crystal displays. A media company (media owner) that owns the advertising media examines in advance whether or not it is suitable as an advertisement.

Judging of videos for advertisement is generally performed by a person (jury). For example, the examiner must actually view the entire video and determine whether or not it contains content that violates preset advertising rules. It is known that the examination of such videos takes a long time and thus imposes a heavy workload on the examiners.

Therefore, conventionally, a technique has been developed for judging advertisement videos using a learning model generated by machine learning. For example, in order to shorten the screening time for advertising videos, pass/fail information on whether or not at least one of the images, characters, or sounds contained in the advertising video passes predetermined screening conditions is acquired, and the videos are and a technique of learning a learning model using pass/fail information as teacher data (see Patent Document 1).

In addition, in order to make effective use of the content (advertisement) screening results, after the content screening using the model, information on the results of the visual screening of the content is acquired as learning data, and the acquired learning data is used. There is known a technique of learning a model based on this (see Patent Literature 2).

Japanese Patent Application Laid-Open No. 2020-4248 Japanese Patent Application Laid-Open No. 2020-101940

By the way, it may be difficult to accurately examine the suitability of a video produced for advertisement as an advertisement only by examining the video based on the learning model disclosed in Patent Documents 1 and 2. Therefore, in addition to the examination based on the learning model, it is conceivable that the judges examine the videos. In that case, if the reviewers can easily use (i.e. refer to) the review results of the video based on the learning model when performing the review, it may be possible to streamline the review of the video for advertisement. There is

However, in the conventional technology described above, no special consideration is given to the examiner's use of the examination result of the video based on the learning model.

Therefore, the present disclosure provides an advertisement examination system and an advertisement examination that can improve the efficiency of examination of advertisement images by making examination results based on learning models of advertisement images easily available to examiners. The main purpose is to provide a method and an advertisement review program.

The advertisement examination system of the present disclosure is an advertisement examination system that examines a video produced for advertisement, and includes a video acquisition unit that acquires a video to be reviewed, and a learning model for reviewing whether the video is appropriate as an advertisement. , a pre-examination processing unit that extracts a plurality of video components included in the video, and a plurality of video components extracted by the pre-examination processing unit, based on the learning model A video examination unit that generates a judgment result regarding the video by judging whether it is suitable as an advertisement, and a examination result output unit that outputs a examination result screen including the judgment result to an examiner involved in advertising examination of the video. and an image display area for displaying a still image showing, in an identifiable form, the inappropriate elements, which are the video constituent elements determined to be inappropriate by the video review section, on the examination result screen. , a first examination result display area for displaying the judgment results for each of the inappropriate elements with respect to the still image; 2 examination result display area , wherein a first symbol associated with each of the inappropriate elements in the still image is superimposed and displayed in the image display area, and the second examination result is displayed. Third symbols corresponding to each of the first symbols are displayed in the display area in association with comments by the examiner corresponding to each of the inappropriate elements .

The advertisement examination method of the present disclosure is an advertisement examination method by an advertisement examination system that examines a video produced for advertisement. extracting and judging whether each of the plurality of extracted video components is appropriate as an advertisement based on a learning model for examination, thereby generating a judgment result regarding the video, and conducting examination related to advertising examination of the video and outputting a screening result screen including the judgment result to the person, and the screening result screen shows a still image showing the inappropriate element, which is the video component judged to be inappropriate, in an identifiable form. an image display area to be displayed; a first examination result display area for displaying the judgment results for each of the inappropriate elements with respect to the still image; a second examination result display area in which comments are input , wherein a first symbol associated with each of the inappropriate elements in the still image is superimposed and displayed in the image display area; Third symbols corresponding to each of the first symbols are displayed in the second examination result display area in association with comments by the examiner corresponding to each of the inappropriate elements. .

An advertisement screening program of the present disclosure is an advertisement screening program for screening a video produced for an advertisement, and includes a video acquisition process for acquiring a video to be reviewed and a plurality of video components included in the video. Pre-examination processing, and video examination for generating a judgment result regarding the video by judging whether each of the plurality of video components extracted by the pre-examination processing is appropriate as an advertisement based on a learning model for examination. and a review result output process of outputting a review result screen including the determination result to the reviewer involved in reviewing the video advertisement, and determining that the review result screen is inappropriate. a first image display area for displaying a still image showing inappropriate elements, which are said video constituent elements, in an identifiable form; and a second examination result display area in which comments by the examiner on each of the inappropriate elements are input with respect to the still image , wherein the image display area contains the still image. are superimposed and displayed, and in the second examination result display area, third symbols corresponding to each of the first symbols are displayed in the It is configured to be displayed in association with the comments by the examiner corresponding to each of the inappropriate elements .

According to the present disclosure, it is possible to improve the efficiency of the examination of advertising videos by making it possible for examiners to easily use the examination results based on the learning model of advertising videos.

Overall configuration diagram of an advertisement screening system according to the present embodiment Explanatory diagram showing an overview of preliminary screening of video in the advertisement screening system shown in FIG. Explanatory diagram showing a modification of the preliminary examination shown in FIG. Explanatory diagram showing the outline of the main examination of the video in the advertisement examination system shown in Fig. 1 Functional block diagram of the advertisement screening server shown in FIG. Flow chart showing the flow of preliminary examination processing by the advertisement examination server shown in FIG. Flow chart showing a modified example of preliminary examination processing by the advertisement examination server shown in FIG. Flow chart showing the flow of main examination processing by the advertisement examination server shown in FIG. Explanatory diagram showing an example of a manuscript submission screen used by a user Explanatory diagram showing an example of an examination result display screen used by a user Explanatory diagram showing the final examination screen used by the examiner Explanatory diagram showing an example of an examination result display screen used by an examiner Explanatory diagram showing an example of the review history display screen used by the reviewer to refer to the review history Overall configuration diagram showing a modification of the advertisement screening system shown in FIG.

A first invention made to solve the above-mentioned problems is an advertisement screening system for screening a video produced for an advertisement, comprising a video acquisition unit for acquiring a video to be reviewed, and an advertisement for the video. A storage unit that stores a learning model for examination regarding propriety, a pre-examination processing unit that extracts a plurality of image constituent elements included in the image, and each of the plurality of image constituent elements extracted by the pre-examination processing unit. , a video examination unit that generates a judgment result regarding the video by judging whether it is suitable as an advertisement based on the learning model, and an examination result screen that includes the judgment result for an examiner involved in advertising examination of the video wherein the screening result screen is a still image showing, in an identifiable form, the inappropriate elements, which are the video constituent elements determined to be inappropriate by the video screening section. a first examination result display area for displaying the judgment results for each of the inappropriate elements with respect to the still image; and the examiner for each of the inappropriate elements with respect to the still image and a second examination result display area in which a comment is input , wherein a first symbol associated with each of the inappropriate elements in the still image is superimposed and displayed in the image display area. , in the second examination result display area, a third symbol corresponding to each of the first symbols is displayed in association with the comment by the examiner corresponding to each of the inappropriate elements; do.

According to this, it is possible to improve the efficiency of the examination of advertisement images by making it possible for the examiner to easily use the examination results based on the learning model of the advertisement images.

In a second aspect of the invention, a still image superimposed with a frame for identifying the inappropriate element is displayed in the image display area as a still image showing the inappropriate element in an identifiable form. Configuration .

In the third invention , in the first examination result display area, second symbols corresponding to each of the first symbols are associated with the judgment results corresponding to each of the inappropriate elements. It is configured to be displayed as follows. According to this, the examiner can efficiently examine the advertisement video by referring to the first symbol and the corresponding second symbol.

In a fourth aspect of the invention, the pre-examination processing unit converts voice as the video component into characters and extracts them, and the characters converted from the voice correspond to the image display area. It is configured to be superimposed on the still image and displayed. According to this, the examiner can appropriately execute the examination regarding the sound as the video component.

In a fifth aspect of the invention, the learning model includes a plurality of industry learning models generated for each industry based on industry rules respectively defined for a plurality of preset industries, and a media company distributing the video. including a media company learning model generated based on defined media company rules, and a common learning model generated based on common rules defined regardless of the plurality of industries and the media companies, wherein the video screening unit, The determination result is generated based on the industry learning model, the media company learning model, and the common learning model corresponding to at least one industry to which the advertisement target of the video belongs among the plurality of industries.

According to this, it is possible to properly examine the advertisement video based on a plurality of learning models suitable for the examination of the advertisement video.

A sixth aspect of the invention is an advertisement screening method by an advertisement screening system for screening a video produced for advertisement, wherein the video to be reviewed is acquired, and each of a plurality of video components contained in the video is evaluated. extracting and judging whether each of the plurality of extracted video components is appropriate as an advertisement based on a learning model for examination, thereby generating a judgment result regarding the video, and conducting examination related to advertising examination of the video and outputting a screening result screen including the judgment result to the person, and the screening result screen shows a still image showing the inappropriate element, which is the video component judged to be inappropriate, in an identifiable form. an image display area to be displayed; a first examination result display area for displaying the judgment results for each of the inappropriate elements with respect to the still image; a second examination result display area in which comments are input , wherein a first symbol associated with each of the inappropriate elements in the still image is superimposed and displayed in the image display area; Third symbols corresponding to each of the first symbols are displayed in the second examination result display area in association with comments by the examiner corresponding to each of the inappropriate elements. .

According to this, it is possible to improve the efficiency of the examination of advertisement images by making it possible for the examiner to easily use the examination results based on the learning model of the advertisement images.

A seventh aspect of the invention is an advertisement screening program for screening a video produced for an advertisement, comprising: a video acquisition process for acquiring a video to be reviewed; and extracting a plurality of video components contained in the video. Pre-examination processing, and video examination for generating a judgment result regarding the video by judging whether each of the plurality of video components extracted by the pre-examination processing is appropriate as an advertisement based on a learning model for examination. and a review result output process of outputting a review result screen including the determination result to the reviewer involved in reviewing the video advertisement, and determining that the review result screen is inappropriate. a first image display area for displaying a still image showing inappropriate elements, which are said video constituent elements, in an identifiable form; and a second examination result display area in which comments by the examiner on each of the inappropriate elements are input with respect to the still image , wherein the image display area contains the still image. are superimposed and displayed, and in the second examination result display area, third symbols corresponding to each of the first symbols are displayed in the It is configured to be displayed in association with the comments by the examiner corresponding to each of the inappropriate elements .

According to this, it is possible to improve the efficiency of the examination of advertisement images by making it possible for the examiner to easily use the examination results based on the learning model of the advertisement images.

Embodiments of the present disclosure will be described below with reference to the drawings.

FIG. 1 is an overall configuration diagram of an advertisement screening system 1 according to this embodiment.

The advertisement screening system 1 includes an advertisement screening server 3 , a user terminal 5 and an examiner terminal 7 . The advertisement screening server 3, user terminals 5, and examiner terminals 7 are each connected to a known communication network 9 such as the Internet.

The advertisement screening server 3 is an information processing device that executes processing for determining whether a video produced for advertisement is appropriate as an advertisement. The advertisement screening server 3 appropriately includes known hardware such as a processor (CPU, MPU, etc.), memory, display, input device, network interface, and storage. Note that the advertisement images handled by the advertisement examination system 1 include not only moving images but also a plurality of still images (for example, animated still images). Advertising images are produced by advertising companies according to requests from advertisers. The produced video is subject to examination by the server 3 for advertisement examination.

The user terminal 5 is used by an advertising video creator or the like (hereinafter referred to as a user) in an advertising company (advertising agency). As the user terminal 5, a known computer having a communication function such as a desktop PC (Personal Computer), a notebook PC, a tablet terminal, and a smart phone is used. Note that the user terminal 5 may be used by an advertiser.

The examiner terminal 7 is used by an examiner (person in charge of advertisement examination) at a media company (media owner). A media company owns advertising media for providing advertising services (that is, broadcasting videos for advertising). In addition, the media company examines the adequacy of the video to be aired as an advertisement. As the examiner terminal 7, a known computer such as a desktop PC, a notebook PC, a tablet terminal, and a smart phone is used. Note that the examiner terminal 7 may be used by an administrator of the advertisement examination system 1 . The term "broadcasting" is used here because it refers to the screening of advertisement videos, but the term "broadcasting" also includes the distribution of content to an unspecified number of viewers. It is not intended to limit the interpretation of rights in this application.

In the advertisement examination system 1, the number and arrangement of the advertisement examination server 3, the user terminals 5, and the examiner terminals 7 can be changed as appropriate. Further, the advertisement screening server 3 does not necessarily have to be configured by one device. In other words, the advertisement screening server 3 may be composed of a plurality of devices (computers, etc.) each capable of implementing a plurality of functions thereof (for example, processes related to preliminary screening and main screening, which will be described later).

As will be described below, the advertisement screening server 3 can perform a preliminary screening for preliminarily screening the suitability of a video as an advertisement in order to notify the user of at least a part of the screening results in advance. In addition, the advertisement screening server 3 can carry out a final examination for judging whether or not a video that has been screened by a preliminary screening (hereinafter referred to as a pre-screened video) is judged as an advertisement by an examiner. . Typically, in a pre-screened video, at least some of the video components (elements that make up the video to be screened) that were determined to be inappropriate in the above-described preliminary screening have been modified by the user. The video component includes at least one of still images, text, and audio.

The preliminary screening process in the advertisement screening server 3 is executed based on a plurality of prepared learning models (or at least one learning model selected from them). A learning model can be generated by machine learning. Machine learning may use known algorithms such as neural networks, support vector machines, logistic regression, random forests, and linear regression. Further, the main examination process (excluding the examination by the examiner) in the advertisement examination server 3 is executed based on a plurality of learning models, like the preliminary examination process.

FIG. 2 is an explanatory diagram showing an overview of preliminary screening of video in the advertisement screening system 1 shown in FIG.

In the preliminary screening of the advertisement screening system 1, the common rule screening (ST1001) is performed on the video to be screened submitted by the user (for example, transmitted from the user terminal 5 to the advertisement screening server 3). . Common rule screening is performed using a common learning model.

A common learning model is generated by learning using a data set related to common rules as teacher data. Common rules include rules that should be generally observed regardless of the industry to which the advertiser (or advertised product or service) belongs or the media company (or advertising medium). For example, common rules include things that violate laws and regulations, things that go against public order or good morals, things that are related to specific politics, ideas, or religions, things that mislead people with false or exaggerated expressions, and things that are insulting or discriminatory. It is set in order to avoid inappropriate expressions and to avoid infringement of privacy (that is, it is determined to be inappropriate as an advertisement).

Next, the result of common rule examination is judged (ST1002). Therefore, when it is determined that the image to be examined is appropriate as an advertisement as a result of the common rule examination, the industry rule examination (ST1003) is subsequently carried out. Industry rule screening is performed using an industry learning model corresponding to the industry to which the video's advertising target belongs. That is, the industry learning model is an industry learning model generated based on the industry rule corresponding to the industry to which the video advertisement target belongs, among the plurality of learning models prepared for each industry. In industry rule screening, for example, advertisements related to industry A are screened for industry A rules based on the learning model for industry A, and advertisements for industry B are screened for industry B rules based on the learning model for industry B. A review is carried out.

The industry learning model is generated by learning using a data set related to industry rules defined for the industry as training data. Such industries include, for example, the pharmaceutical industry, the medical device industry, the cosmetics industry, the real estate industry, the education industry, the financial industry, the mail order industry, and the travel industry.

For example, advertisement videos about cosmetics are judged based on a learning model generated using a data set about industry rules of the cosmetics industry as training data. To avoid the use of brand names that have not been approved or permitted, the use of names other than generic names, and unapproved expressions regarding efficacy and effects, etc. Contains the configured rules. The learning models for other industries are similar to those for the cosmetics industry, except that there are differences in the rules specific to each industry.

Next, the result of industry rule examination is judged (ST1004). As a result of industry rule examination, therefore, when it is determined that the image to be examined is appropriate as an advertisement, media company rule examination (ST1005) is carried out. The publisher rule review is performed using the publisher learning model.

The media company learning model is generated by learning using, as training data, a data set related to media company rules established by the media company that broadcasts the advertisement. For example, media company rules include rules for prohibiting the airing of advertisements for specific industries and rules for prohibiting the airing of advertisements for specific products and services depending on the location where each advertising medium is installed. and so on. Note that the media company learning model is prepared for each media company that broadcasts advertisements. In the advertisement screening server 3, media company learning models corresponding to a plurality of media companies may be stored in advance.

Next, the result of media company rule examination is determined (ST1006). Thereafter, examination result processing (ST1007) is performed. In examination result processing, the results of common rule examination, industry rule examination, and media company rule examination are archived (ST1008). In addition, the results of these examinations are notified to the user who submitted the video (ST1009). Furthermore, if at least a part of the video component is judged to be appropriate as an advertisement in any of the common rule review, industry rule review, and media company rule review, the video will be sent to the user along with notification of the review results. A re-submitting instruction (ST1010) is issued to prompt correction of the .

It should be noted that if at least a part of the video component is determined to be appropriate as an advertisement in the common rule examination or the industry rule examination, even if examination result processing (ST1007) is performed without proceeding to the next examination. good (see dashed arrow in FIG. 2). This allows the user to start modifying the advertising video without waiting for all review results.

FIG. 2 shows an example in which common rule examination, industry rule examination, and media company rule examination are performed in this order, but the order of these examinations can be changed as appropriate.

In this disclosure, a common learning model for common rules, an industry learning model for industry rules, and a media company learning model for media company rules are cited as examples of learning models suitable for judging videos for advertising. The learning model is not limited to these as long as it is a learning model based on a predetermined rule that allows video screening.

FIG. 3 is an explanatory diagram showing a modification of the preliminary examination shown in FIG. In FIG. 3, the same symbols are attached to the same components as those shown in FIG. In addition, regarding the modified examples, details that are not specifically mentioned below are the same as those described above, and detailed description thereof will be omitted.

As shown in FIG. 3, when submitting a video to be reviewed, the user can select one of common rule review, industry rule review, and media company rule review to perform a preliminary review. can.

For example, if the user fully understands the common rules and the industry rules of the industry to which the advertisement belongs, the produced video will include video components that are determined to be inappropriate in the common rules review and the industry rules review. unlikely to contain. In such a case, the user can select publisher rule review and obtain only the result of that review. Even if the produced video contains video components inappropriate for the common rules review and the industry rules review, those inappropriate video components may be extracted later in the main review.

Thus, in the modified example, the preliminary examination is made more efficient by performing the examination based on one learning model selected by the user.

FIG. 4 is an explanatory diagram showing the outline of the main examination of the video in the advertisement examination system shown in FIG.

In the main examination by the advertisement examination system 1, the examination through judgment (ST2001) is performed on the pre-examined video submitted by the user. In screening through determination, videos that have been aired as advertisements in the past are excluded from screening (that is, they are determined to be appropriate as advertisements). Examination through determination may be performed by the examiner as necessary.

For videos that have not been aired as advertisements in the past, display restriction determination (ST2002) is subsequently carried out. In the publication regulation determination, videos inappropriate as advertisements are excluded from broadcasting targets based on the dignity and ethics of advertisements. The posting restriction determination may be performed by the examiner as necessary.

For videos determined to be appropriate in the posting regulation determination, the common rule review (ST2003) is executed in the same manner as the common rule review (ST1001) and the common rule result determination (ST1002) in the preliminary review, and the common rule is determined. The examination result is determined (ST2004). Furthermore, industry rule examination (ST2005) and industry rule examination (ST2005) and industry rule examination (ST1005) and industry rule examination (ST1005) and industry rule examination (ST1004) and media company rule examination (ST1005) and media company rule examination (ST1006), respectively, in the preliminary examination. Rule result determination (ST2006), media company rule examination (ST2007), and media company rule result determination (ST2008) are executed.

After that, comprehensive judgment (ST2009) is executed to summarize the results of common rule result judgment, industry rule result judgment, and media company rule result judgment. After that, the examiner reviews (views the video) while referring to the results of the overall judgment (that is, the presence or absence of video components determined to be inappropriate by the review based on each learning model). ST2010).

If the video is approved by the reviewer (that is, determined to be appropriate as an advertisement), total volume restriction determination (ST2011) is executed. In judging the total volume regulation, whether or not the video subject to examination, including other advertising videos aired at the same time, exceeds the upper limit of the total volume judging criteria regarding the advertisement airing time and airing volume (number). is determined. Here, if the video to be screened does not satisfy the total amount criteria, even if the video is suitable as an advertisement, it is excluded from being broadcast. The total volume control determination may be performed by the examiner as necessary.

Thereafter, examination result processing (ST2012) is performed in the same manner as the examination result processing (ST1007) in the preliminary examination. In the examination result processing, archiving of examination results (ST2013), notification of examination results (ST2014), and resubmission instruction (ST2015) are executed as in the preliminary examination. In addition, learning data generation (ST2016) is executed in the examination result processing in the main examination. In the examination by the examiner, if video components inappropriate for advertisement (hereinafter referred to as "inappropriate elements") are extracted, examination result processing may be performed without execution of total volume control judgment ( See dashed arrow in FIG. 4).

In the learning data generation, teacher data for each learning model is generated based on data of archived examination results (for example, video components determined to be inappropriate as advertisements in the final examination). In learning data generation, the results of the common rule review, industry rule review, and media company rule review are used to generate teacher models for the common learning model, industry learning model, and media company learning model, respectively. By appropriately relearning each learning model using such teaching data, it becomes possible to perform the preliminary examination and the main examination with higher accuracy.

FIG. 4 shows an example in which the main examination is performed using all the learning models used in the preliminary examination. However, this examination is not limited to such a mode.

In the final screening, all learning models that have not been used in the preliminary screening may be used among the multiple learning models prepared in advance. For example, if the preliminary screening was conducted using only the publisher learning model (i.e., if only the publisher rule screening was conducted in the preliminary screening), the learning model that was not used in the preliminary screening will be used in the main screening. Certain common learning models and industry learning models can all be used (i.e., perform common rules review and industry rules review). Alternatively, part of the learning model that was not used in the preliminary examination may be used in the main examination (for example, one of the common rule examination and the industry rule examination that were not conducted in the preliminary examination may be conducted). . Also, at least one learning model not used in the preliminary screening may be used in combination with the learning model used in the preliminary screening. (For example, if the preliminary screening was conducted using only the media company learning model, in the main screening, the learning model used in the preliminary screening and the common learning model, which is the learning model not used in the preliminary screening, or (can be used in conjunction with industry learning models).

FIG. 5 is a functional block diagram of the advertisement screening server 3 shown in FIG.

The advertisement screening server 3 has a communication section 11 , a storage section 13 and a control section 15 .

The communication unit 11 includes a network interface for communicating with other devices (the user terminal 5 and the examiner terminal 7 here) via the communication network 9 . By communicating with the user terminal 5 and the examiner terminal 7, the advertisement examination server 3 can transmit and receive image data to be examined, transmit and receive examination results, and the like.

The storage unit 13 includes storage devices such as HDDs (Hard Disk Drives) and SSDs (Solid State Drives). The storage unit 13 stores information and data such as a learning model 21, a video to be screened 23, a preliminary screening result 25, a final screening result 27, and the like. The learning model 21 includes a plurality of learning models such as a common learning model 31 , an industry learning model 33 , and a media company learning model 35 .

The control unit 15 includes an image acquisition unit 41 , a preliminary examination unit 43 , a main examination unit 45 and a learning data generation unit 47 .

The image acquisition unit 41 acquires advertisement images (including pre-screened images) transmitted from the user terminal 5 and stores them in the storage unit 13 as images 23 to be screened.

The preliminary examination section 43 includes a first examination preprocessing section 51 , a first video examination section 53 , a first examination result output section 55 , and a first examination result notification section 57 .

The first pre-examination processing unit 51 executes a process of extracting images and characters included in each frame (still image) constituting a video to be examined, and audio in the video. A known method can be used for the process of extracting images, characters, and sounds from video. Those frames, the images and text they contain, and the audio are the visual components of the video being judged. Furthermore, the first pre-examination processing unit 51 can convert the extracted voice into text information, and acquire the converted text information as a video component. Such video components may be obtained as information of feature amounts related to images, characters, sounds, and the like.

The first video screening unit 53 performs video screening (hereinafter referred to as AI screening) based on the learning model 21 . Reviews of such videos include common rule reviews, industry rule reviews, and publisher rule reviews as described above. The first video examination unit 53 determines whether each video component acquired by the first pre-examination processing unit 51 is appropriate as an advertisement, and obtains the judgment result (AI examination result or AI examination result). The same shall apply hereinafter) to generate an examination result (preliminary examination result 25). The learning model used here is appropriately changed according to the type of examination to be executed (here, either common rule examination, industry rule examination, or media company rule examination). For example, in the common rule examination (see FIG. 2), the common learning model 31 is used to determine the adequacy of each video component as an advertisement.

The first screening result output unit 55 outputs, as screening results, information on video constituent elements that have been determined by the first video screening unit 53 to be inappropriate as advertisements in a user-confirmable manner. For example, the first examination result output unit 55 generates an examination result screen including the judgment result by the first examination preprocessing unit 51 and stores it in the storage unit 13 as part of the preliminary examination result 25 . The user can display the screening result screen on the user terminal 5 by accessing the advertisement screening server 3 .

The first examination result notification unit 57 generates an examination result notification including information on the examination result by the first pre-examination processing unit 51 . The first examination result notification unit 57 can transmit the examination result notification to the user terminal 5 (that is, notify the user) using a known communication tool such as e-mail. When using e-mail, the first examination result notification unit 57 can transmit an examination result notification to a preset address. Further, for example, when electronic means such as SNS (Social Networking Service) messages or chat applications are used, it is possible to send examination result notifications to preset destinations (accounts).

Note that the first video examination unit 53 executes examination based on a plurality of learning models (here, the common learning model 31, the industry learning model 33, and the media company learning model 35) in a predetermined order set in advance. By doing so, a preliminary examination result for each of these learning models may be generated. At this time, the first screening result output unit 55 can sequentially output preliminary screening results for each learning model to the user. For example, if one of the preliminary examination results for each learning model includes information (judgment result) indicating that the video component is inappropriate as an advertisement, the first examination result output unit 55 Judging results (including information about inappropriate video components) can be output to the user immediately (i.e., whether or not the generation of preliminary judging results based on other learning models has been completed). can. Also, at this time, the first examination result notification unit 57 can generate an examination result notification including information on the preliminary examination result for each learning model and sequentially transmit it to the user. For example, if one of the preliminary screening results for each learning model includes information indicating that the video component is inappropriate as an advertisement, the first screening result notifying unit 57 provides information about the preliminary screening result. immediately (i.e., regardless of whether or not the generation of preliminary screening results based on other learning models has been completed) and immediately send it to the user.

In addition, if the first video screening unit 53 determines that at least one video component in the video to be reviewed is inappropriate as an advertisement, the first video screening unit 53 may generate a preliminary examination result. At this time, the first screening result output unit 55 can sequentially output preliminary screening results for each video component to the user. For example, if the preliminary screening result includes information (determination result) indicating that the video component is inappropriate as an advertisement, the first screening result output unit 55 outputs the preliminary screening result (inappropriate video (including information about the components) can be immediately output to the user. Also, at this time, the first screening result notification unit 57 can generate screening result notification including information on the preliminary screening result for each video component and sequentially transmit it to the user. For example, if the preliminary examination result includes information indicating that the video component is inappropriate as an advertisement, the first examination result notification unit 57 immediately notifies the examination result including the information regarding the preliminary examination result. can be generated and sent to the user immediately.

The main examination section 45 includes a second examination preprocessing section 61 , a second image examination section 63 , a second examination result output section 65 and a second examination result notification section 67 . The second examination preprocessing section 61, the second image examination section 63, the second examination result output section 65, and the second examination result notification section 67 are the preliminary examination section except for the matters mentioned below. It has the same functions as the first examination preprocessing section 51, the first image examination section 53, the first examination result output section 55, and the first examination result notification section 57 in 43, respectively.

In the main examination section 45, the pre-examined video is subject to examination. The pre-screened video is transmitted from the user terminal 5 and acquired by the video acquisition unit 41 . However, if there are no items pointed out in the preliminary screening (that is, if it is determined that the video to be reviewed does not contain inappropriate elements), the transmission of the pre-screened video from the user will be omitted, and after the preliminary screening The video may be used as it is for judging. In addition, if at least part of a video that has been preliminarily screened in the past is included in the manuscript submission, the manuscript submission user may decide to skip the preliminary screening and start from the main screening.

The second video examination section 63 performs AI examination on the pre-screened video in the same manner as the first video examination section 53 . In addition to the AI examination, the second video examination section 63 can also allow the examiner to perform the examination of the video using the main examination screen (see FIG. 11) including information on the results of the preliminary examination. In addition, the second image examination section 63 can appropriately execute examination through judgment, notice regulation judgment, and total volume regulation judgment (see FIG. 4).

The second screening result output unit 65 outputs information on the video components determined by the second video screening unit 63 to be inappropriate as an advertisement as a screening result in a manner that can be confirmed by the reviewers. For example, the second examination result output unit 65 generates a main examination screen (see FIG. 11) including the result of the AI examination by the second video examination unit 63, and outputs it (see FIG. 11) in response to a request from the examiner. For example, it can be displayed on the examiner terminal 7). The second examination result output unit 65 generates an examination result screen including the results of the AI examination and the examination by the examiner, and stores it in the storage unit 13 as the final examination result 27 .

The second examination result notification unit 67 generates an examination result notification including information on the results of AI examination by the second video examination unit 63 . The second examination result notification unit 67 can use a known communication tool such as e-mail to transmit the examination result notification to the examiner terminal 7 (that is, notify the examiner). The second examination result notification unit 67 can use electronic means such as SNS (Social Networking Service) messages and chat applications as communication tools.

The learning data generation unit 47 generates teacher data for each learning model based on the final examination result 27 . The generated teacher data is stored in the storage unit 13 together with the learning model 21 . The learning data generation unit 47 uses the results of the common rule examination, the industry rule examination, and the media company rule examination in the main examination results 27 for the corresponding common learning model 31, industry learning model 33, and media company learning model 35. used to generate a teacher model for The learning data generation unit 47 is a re-learning unit that performs processing for updating (that is, re-learning) the common learning model 31, the industry learning model 33, and the media company learning model 35 using the generated teacher model data. may function.

The control unit 15 includes one or more processors (CPU, MPU, GPU, etc.) as hardware. The function of each part in the control part 15 described above is realized by the processor executing a predetermined control program. Also, at least part of the functions of the control unit 15 can be realized in cooperation with hardware other than the processor.

FIG. 6 is a flowchart showing the flow of preliminary examination processing by the advertisement examination server 3 shown in FIG.

When the advertisement examination server 3 acquires a video to be examined (ST101), it executes examination (common rule examination) based on the common learning model 31 (ST102).

If there are no indications in the examination in step ST102 (that is, there is no video component that has been determined to be inappropriate as an advertisement) (No in ST103), then the advertisement examination server 3 instructs the industry learning model 33 to (ST104).

If there are no indications in the examination of step ST104 (No in ST105), then the advertisement examination server 3 executes examination (media company rule examination) based on the media company learning model 35 (ST106).

If there are no indications in the examination in step ST106 (No in ST107), then the advertisement examination server 3 archives those examination results (ST108). At this time, the results obtained in the examinations in steps ST102, ST104, and ST106 are stored in the storage section 13 as preliminary examination results 25. FIG. After that, the advertisement examination server 3 notifies the examination result to the user (for example, the user terminal 5) who submitted the image to be examined (ST109). In this case, the examination result notification includes information (for example, a message) indicating that the image to be examined is appropriate as an advertisement.

Note that even if there are any indications in steps ST103, ST105, and ST107 (Yes), the examination results (including the indications) may be archived in the same manner as in step ST108.

On the other hand, if there are items pointed out in the examination in step ST102 (Yes in ST103), the advertisement examination server 3 submits the video to be examined without executing the remaining examinations (here, steps ST104 and ST105). The examination result is notified to the user who has done so (ST109). In this case, the video under review includes video components that have been determined to be inappropriate for advertising. Therefore, the examination result notification includes information (for example, a message) instructing the user to resubmit the video (that is, notifying that the video needs to be corrected and re-examined).

Note that the order of examination in steps ST102, ST104, and ST106 may be changed as appropriate.

FIG. 7 is an explanatory diagram showing a modification of the preliminary examination process shown in FIG.

When the advertisement examination server 3 acquires the video to be examined (ST201), it selects one learning model to be used in examination based on the user's selection (ST202). In this embodiment, one of the common learning model 31, the industry learning model 33, and the media company learning model 35 is selected as the learning model used in the examination.

Next, advertisement examination server 3 executes examination based on the selected learning model (ST203). Here, the results of the examination are archived (ST204), regardless of whether or not there are items pointed out in the examination of step ST203. At this time, the examination result is stored in the storage unit 13 as the preliminary examination result 25 . After that, the advertisement examination server 3 notifies the examination result to the user who submitted the video to be examined (ST205), as in step ST109 of FIG.

In the modified example shown in FIG. 7, the user can quickly obtain the examination result based on the selected learning model, so that it is possible to correct the video (video component) that is determined to be inappropriate as an advertisement, The transition to examination can be performed quickly.

FIG. 8 is a flow chart showing the flow of final examination processing by the final examination section 45 shown in FIG.

When the advertisement screening server 3 acquires a video to be screened (ST301), it sequentially executes screening based on the common learning model 31, industry learning model 33, and media company learning model 35 (ST302- ST304).

If there are no indications (that is, video components that are determined to be inappropriate as an advertisement) in these examinations (No in ST305), the advertisement examination server 3 confirms the viewing of the video by the examiner ( ST306). At this time, the advertisement examination server 3 can display a main examination screen (see FIG. 11) for examination based on a request from the examiner on the examiner terminal 7 . The judges can easily check whether inappropriate elements are included by viewing at least a part of the video (especially the part that is considered difficult to judge by the learning model) using the main examination screen. do.

After that, when the video to be reviewed is approved by the reviewer (Yes in ST307), the advertisement review server 3 executes the total amount restriction determination (ST308).

After that, the advertisement screening server 3 archives the screening results in steps ST302 to ST304 (ST309), and notifies the screening result to the user who submitted the video to be screened (ST310).

On the other hand, if any of the examinations in steps ST302 to ST304 indicates an item to be pointed out (Yes in ST305), the advertisement examination server 3 executes the judgment of video viewing by the examiner (ST311). At this time, the advertisement examination server 3 can display the main examination screen on the examiner terminal 7 based on a request from the examiner, in the same manner as the viewing confirmation described above. Judges will use the final screening screen to view at least a portion of the video (especially the parts that are considered difficult to judge by the learning model, as well as the video components judged to be inappropriate by the learning model). By doing so, we will confirm in detail whether the results of the AI examination are appropriate.

After that, when the video to be reviewed is approved by the reviewer (Yes in ST312), the advertisement review server 3 determines the total volume restriction (ST308), archives the review result (ST309), and Examination result notification (ST310) is sequentially executed.

In step ST310, the examination result may be notified to the examiner (for example, examination terminal, etc.) or its manager. In that case, the advertisement screening server 3 can use a known communication tool to notify the screening results to a preset media company account (e-mail address, etc.).

The notification timing of such examination results is immediately after or at the same time as the completion of the main examination process (for example, step ST309 or step ST310), after a predetermined time or a predetermined number of days have passed since the completion of the main examination process, and on the broadcast date. (or the due date for completion of the advertising video) may be at least one timing of a predetermined time or a predetermined number of days before. Further, the examination results may be notified multiple times at different timings after a predetermined period of time or a predetermined number of days have elapsed since the completion of the main examination process. For example, the examination results may be notified one day after the completion of the main examination process, and further two days after the completion of the main examination process. Similarly, the notification of the examination results a predetermined time or days before the broadcast date (or the deadline for completion of the advertising video) may be made multiple times at different timings.

FIG. 9 is an explanatory diagram showing an example of a manuscript submission screen used by a user to submit a video to be reviewed.

In the advertisement screening system 1 , the user can access the advertisement screening server 3 from the user terminal 5 to display an operation screen for the user regarding advertisement screening. The user terminal 5 can display the operation screen using a general-purpose browser or a dedicated application. As shown in FIG. 9, when the manuscript submission tab 71 is selected on the operation screen, a manuscript submission screen for the user to submit a video to be reviewed (including pre-screened video) is displayed.

By pressing the file selection button 73 on the manuscript submission screen, the user can perform an operation of uploading the image to be screened stored in the user terminal 5 to the advertisement screening server 3 .

In the video information input area 75 of the submission screen, the user enters the purpose of submission (preliminary screening or final screening), advertisement title, industry name to which the product or service to be advertised belongs, advertiser name, video to be screened and other media broadcast schedule, broadcast performance of the video to be screened (whether or not it has been broadcast as an advertisement in the past), media company name, advertised product name (or service name), broadcast period, supplementary comments, etc. can be entered (or selected).

After the user has completed uploading the video and inputting necessary information, the user presses the submission button 77 to complete the video submission process. As a result, the advertisement screening server 3 starts preliminary screening processing or main screening processing for the video submitted from the user terminal 5 .

FIG. 10 is an explanatory diagram showing an example of an examination result display screen used when the user refers to examination results.

When the examination result tab 81 is selected on the operation screen similar to that of FIG. 9, an examination result display screen is displayed for the user to confirm the examination result (or examination status) of the video as shown in FIG.

In the search information input area 83 of the screening result display screen, the user enters a user ID, an advertisement title, an industry name, an advertiser name, a schedule for broadcasting the image to be screened on other media, a track record of broadcasting the video to be screened, a media company, etc. At least part of the search criteria including name, free keyword, status, etc. can be entered (or selected). As a result, the review status of the video corresponding to the search condition is displayed in the review status display area 85 .

By selecting a check box 86 corresponding to each video displayed in the review status display area 85, the user can display more detailed review results for the selected video. Such review results include, for example, a list of information about multiple video components that have been determined to be inappropriate for the advertisement and their corresponding information about the reason for the determination.

FIG. 11 is an explanatory diagram showing a final examination screen used by an examiner.

In the advertisement examination system 1 , the examiner accesses the advertisement examination server 3 from the examiner terminal 7 to display an operation screen for the examiner regarding advertisement examination. In the examiner terminal 7, it is possible to display the operation screen using a general-purpose browser or a dedicated application. As shown in FIG. 11, when the main examination tab 91 is selected on the operation screen, the main examination screen regarding the pre-examined video to be examined by the examiner is displayed. The judges can use the main judging screen to conduct the main judging of the pre-judged videos.

The main examination screen includes an image display area 92, an AI examination result display area 93 (first examination result display area), and a person in charge examination result display area 94 (second examination result display area).

In the image display area 92, a part of the image to be judged is displayed as a still image. This still image corresponds to a frame at a predetermined timing (that is, a predetermined elapsed time from the start of the video) in the video to be judged. Also, in the still image, symbols 200A to 203A (first symbols) for identifying each of the inappropriate elements (here, part of the image and characters) are superimposed and displayed. When the video component is an image region corresponding to a partial area of a still image, a frame of a predetermined shape (here, a rectangular frame) demarcating the image region is displayed, and a symbol ( Here, the symbol 200A) is attached.

The image display area 92 also includes an audio display area 95 . The audio display area 95 displays text information (characters) obtained by converting the audio output in a predetermined time period including the timing of the still image. If the text information in the audio display area 95 is inappropriate as an advertisement, a symbol for identifying it is superimposed and displayed in the same manner as the symbols 200A to 203A described above.

The image display area 92 is provided with an operation area 96 including a play button, a rewind button, a fast forward button, a fast rewind button, etc. for the judge to change the temporal position of the displayed still image in the video. ing.

Further, below the image display area 92, information related to the video to be reviewed (video ID, industry, company name, advertising company, advertised product/service, broadcast period, submission deadline, final submission date, etc.) is displayed. A video information display area 97 shown is arranged.

In the AI examination result display area 93, the result of the AI examination in the main examination is displayed. The results of the AI review include comments indicating the reason why each video component is inappropriate as an advertisement. These comments are marked with symbols 200B-202B (second symbols) corresponding to the symbols 200A-202A in the image display area 92, respectively. This allows the examiner to understand why each of the inappropriate elements in the image display area 92 is inappropriate from the associated comments.

The examiner refers to the comments in the AI examination result display area 93, and if he/she determines that they are appropriate, can adopt the examination result as it is (that is, maintain the display of the comment). In addition, comments from the examiner can be added to the comments as a result of AI examination. When an examiner's comment is added, the color or shape of the corresponding symbol changes, so that it can be distinguished from the symbol corresponding to the comment as a result of AI screening. On the other hand, if the examiner determines that at least part of the comments in the AI examination result display area 93 are inappropriate, the examiner deletes the comments and the corresponding symbols (including the symbols in the image display area 92). It is possible. In addition, if the reviewer's comment is not attached to the above-mentioned AI review result comment, it is possible to use it as a reference at the time of viewing confirmation without being subject to correction.

In addition, the examiner can perform an operation of attaching a symbol (here, symbol 203A) to a video component that he or she has judged to be inappropriate as an advertisement in the image display area 92 . In addition, the examiner can input a comment indicating the reason why each video component is inappropriate as an advertisement in the person-in-charge examination result display area 94 . A symbol (here, symbol 203B) corresponding to the symbol attached in the image display area 92 is attached to those comments.

The final examination screen includes a disapproval/approval decision area 98 for indicating the final judgment result (whether or not to approve the video) by the examiner as to whether the video to be reviewed is appropriate as an advertisement. be

The final examination screen also includes a supplementary comment area 99 for the examiner to input supplementary comments.

When the examiner finally finishes examining the video, the examiner presses the examination result transmission button 100 to store the examination result in the storage unit 13 and transmit it to the user (for example, the user terminal 5). be.

FIG. 12 is an explanatory diagram showing an example of an examination result display screen used by an examiner to refer to examination results.

When the examination result tab 101 is selected on the operation screen similar to that of FIG. 11, an examination result display screen is displayed for the examiner to confirm the examination result (or examination status) of the video.

The examiner can input (or select) at least part of the search conditions in the search information input area 102, as in FIG. As a result, the review status of the video corresponding to the search condition is displayed in the review status display area 103 .

The examiner can select either a tab 104 under preliminary examination or a tab 105 under final examination in the examination status display area 103 to display a list of videos under preliminary examination or under final examination. Also, the examiner can display more detailed examination results by selecting the check box 106 corresponding to each video displayed in the examination status display area 103 . Such review results include, for example, a list of information about multiple video components that have been determined to be inappropriate for the advertisement and their corresponding information about the reason for the determination.

FIG. 13 is an explanatory diagram showing an example of an examination history display screen used by an examiner to refer to an examination history.

When the examination history tab 110 is selected on the operation screen similar to that of FIG. 12, the examination history display screen for confirming the examination history of the video shown in FIG. 12 is displayed.

The examination history display screen includes a basic information display area 111 that displays basic information of the target video, and an examination progress display area 112 that displays information on the progress of the review of the target video.

In the basic information display area 111, the same information as in the video information display area 97 shown in FIG. manuscript date, etc.) is displayed.

In the examination progress display area 112, for each examination performed on the target video, examination date, examination request date, examination purpose (preliminary examination or main examination), examination result, indications (inappropriate as advertisement comments indicating the reason for this), etc. Further, the examiner can display a more detailed examination result (e.g., corresponding to the main examination screen in FIG. 11) by pressing a result button 113 related to the desired examination.

FIG. 14 is an overall configuration diagram showing a modification of the advertisement screening system 1 shown in FIG. Regarding this modified example, matters not specifically mentioned below are the same as those of the above-described advertisement screening system 1, so detailed description thereof will be omitted.

The advertisement screening system 1 further includes an API (Application Programming Interface) server 120 and an advertisement operation CMS (Contents Management System) 121 in addition to the advertisement screening server 3 described above.

The API server 120 has a function for calling the advertisement examination function by the advertisement examination server 3 and the examination-related data managed by the advertisement examination server 3 from the advertisement operation CMS 121 and making it available. The API server 120 suitably comprises known hardware such as a processor, memory, display, input device, network interface and storage. Note that the API server 120 may be implemented as part of the functions of the advertisement screening server 3 .

The advertisement management CMS 121 is a system used by users to manage advertisement video data, submit the video, and set broadcast conditions. By using the API server 120, the advertisement operation CMS 121 can appropriately execute acquisition of video examination status, acquisition of examination completion information, acquisition of examination results, acquisition of examination history, and the like. This allows the user to use the functions and data of the advertisement screening server 3 as needed.

As described above, the embodiments have been described as examples of the technology disclosed in the present application. However, the technology in the present disclosure is not limited to this, and can also be applied to embodiments with modifications, replacements, additions, omissions, and the like. Further, it is also possible to combine the constituent elements described in the above embodiments to create new embodiments.

The advertisement review system, advertisement review method, and advertisement review program according to the present disclosure make it possible for the reviewer to easily use the review results based on the learning model of the video for advertisement, thereby reviewing the video for advertisement. It has an effect of improving efficiency, and is useful as an advertisement examination system, an advertisement examination method, an advertisement examination program, etc. for examining the suitability of a video produced for advertisement as an advertisement.

1: Advertisement examination system 3: Advertisement examination server 5: User terminal 7: Examiner terminal 9: Communication network 21: Learning model 23: Examined video 31: Common learning model 33: Industry learning model 35: Media company learning model 41 : Image acquisition unit 43 : Preliminary examination unit 45 : Main examination unit 47 : Learning data generation unit 51 : First examination preprocessing unit 53 : First image examination unit 55 : First examination result output unit 57 : First Examination result notification unit 61 : Second examination preprocessing unit 63 : Second video examination unit 65 : Second examination result output unit 67 : Second examination result notification unit 92 : Image display area 93 : AI examination result Display area (first examination result display area)
94: Person in charge examination result display area (second examination result display area)
120: API server

Claims (7)

An advertisement screening system for screening a video produced for advertisement,
a video acquisition unit that acquires a video to be reviewed;
a storage unit that stores a learning model for examination regarding suitability of the video as an advertisement;
A pre-examination processing unit that extracts a plurality of video components included in the video;
a video examination unit that generates a judgment result regarding the video by judging whether each of the plurality of video components extracted by the pre-examination processing unit is appropriate as an advertisement based on the learning model;
an examination result output unit that outputs an examination result screen including the judgment result to an examiner involved in examination of the video advertisement;
with
The examination result screen is
an image display area for displaying a still image showing, in an identifiable form, the inappropriate elements, which are the video constituent elements determined to be inappropriate by the video examination section;
a first examination result display area displaying the judgment result for each of the inappropriate elements with respect to the still image;
a second examination result display area in which the examiner's comments on each of the inappropriate elements regarding the still image are input;
including
a first symbol associated with each of the inappropriate elements in the still image is superimposed and displayed in the image display area;
Advertisement screening , wherein third symbols corresponding to each of the first symbols are displayed in the second screening result display area in association with comments by the reviewer corresponding to each of the inappropriate elements. system. 2. The advertisement according to claim 1, wherein a still image superimposed with a frame for identifying the inappropriate element is displayed as a still image showing the inappropriate element in an identifiable form in the image display area. examination system. 2. A second symbol corresponding to each of said first symbols is displayed in said first examination result display area in association with said judgment result corresponding to each of said inappropriate elements. Or the advertisement screening system according to claim 2. The pre-examination processing unit converts the voice as the video component into text and extracts it,
The advertisement examination system according to any one of claims 1 to 3 , wherein the character converted from the voice is displayed in the image display area so as to be superimposed on the corresponding still image. The learning model is
A plurality of industry learning models generated for each industry based on industry rules respectively defined for a plurality of preset industries,
A media company learning model generated based on media company rules stipulated for media companies distributing said video, and a common learning model generated based on common rules stipulated regardless of said multiple industries and said media company including
The video examination unit generates the determination result based on the industry learning model, the media company learning model, and the common learning model corresponding to at least one industry to which the advertising target of the video belongs among the plurality of industries. The advertisement screening system according to any one of claims 1 to 4 , wherein An advertisement screening method by an advertisement screening system for screening a video produced for advertisement,
Acquire the video to be judged,
extracting each of a plurality of video components included in the video;
For each of the plurality of extracted video components, generating a determination result regarding the video by determining suitability as an advertisement based on a learning model for examination,
outputting an examination result screen including the judgment result to an examiner involved in judging the advertisement of the video;
The examination result screen is
an image display area for displaying a still image showing, in an identifiable form, the inappropriate element, which is the video component determined to be inappropriate;
a first examination result display area displaying the judgment result for each of the inappropriate elements with respect to the still image;
a second examination result display area in which the examiner's comments on each of the inappropriate elements regarding the still image are input;
including
a first symbol associated with each of the inappropriate elements in the still image is superimposed and displayed in the image display area;
Advertisement screening , wherein third symbols corresponding to each of the first symbols are displayed in the second screening result display area in association with comments by the reviewer corresponding to each of the inappropriate elements. Method. An advertisement screening program for screening a video produced for advertisement,
image acquisition processing for acquiring images to be reviewed;
Pre-examination processing for extracting a plurality of video components included in the video;
a video examination process for generating a judgment result regarding the video by judging whether each of the plurality of video constituent elements extracted by the pre-examination process is suitable as an advertisement based on a learning model for examination;
examination result output processing for outputting an examination result screen including the judgment result to an examiner involved in examination of the video advertisement;
on the computer, and
The examination result screen is
an image display area for displaying a still image showing, in an identifiable form, the inappropriate element, which is the video component determined to be inappropriate;
a first examination result display area displaying the judgment result for each of the inappropriate elements with respect to the still image;
a second examination result display area in which the examiner's comments on each of the inappropriate elements regarding the still image are input;
including
a first symbol associated with each of the inappropriate elements in the still image is superimposed and displayed in the image display area;
Advertisement screening , wherein third symbols corresponding to each of the first symbols are displayed in the second screening result display area in association with comments by the reviewer corresponding to each of the inappropriate elements. program.

Images