A system and method for blocking posts with visual contents on

Abstract

The invention relates to a system and method that enables posts with visual content shared on Facebook, Instagram, Twitter and similar social media platforms (30) to be blocked by a user (10), and all posts with the same content as the content in the blocked post are hidden from a user (10) feed screen.

Description

[0001] A SYSTEM AND METHOD FOR BLOCKING POSTS WITH VISUAL CONTENTS ON SOCIAL MEDIA PLATFORMS

[0002] Technological Field:

[0003] The invention relates to a system and method that enables posts with visual content shared on Facebook, Instagram, Twitter and similar social media platforms to be blocked by a user, and all posts with the same content as the content in the blocked post are hidden from a user feed screen.

[0004] State of the Art:

[0005] Social media platforms are environments where people can share posts such as text, photos and videos with their followers. Users socialize by following user accounts they are interested in or know. Also, users can block user accounts that they are not interested in, or do not like, block the posts published by these users, and prevent them from being published on the main feed screen. The invention of the description is related to social media platforms.

[0006] In the state of the art, users can only perform user-based blocking. In some cases, a user account followed may post something that some users don't like. Users need to block the user who shared those posts to prevent this post from falling on the main feed screen. However, the user who shared the post may be someone who is loved and wanted to be followed. An unsolicited status can be specific to only one post. Blocking the user is not a sufficient solution to solve such problems.

[0007] In the state of the art, it is possible to perform a text-based filtering by entering a number of keywords on social media platforms. For example, if the user selects keywords that are not of interest, such as "football", "politics" and the like, he / she can block the texts in which these words are mentioned. As explained above, although there is a method to filter text-based posts, a system is not used on social media platforms to prevent posts with visual content from falling on the main feed screen. In some cases, posts with disturbing or unwanted visual content can be published on social media platforms. For example, violent videos and sexual photos are found disturbing by most users. Also, such posts can be republished by anyone if they have gone viral. Users who are disturbed by these posts have to see the same images over and over again. Since the user cannot individually block all accounts that repost these posts, he / she will continue to be exposed.

[0008] In addition, if a post falls on the user's feed screen more than once, it causes the user to waste time while using the platform and use the platform inefficiently. It will be a feature that most users can prefer that a post containing a video or photo with the same image does not fall a user's feed screen a second time, even when it is shared again by someone else.

[0009] In the US patent document numbered US2014067833A1, which was encountered in the literature search, a system is disclosed, which allows users to filter the e-mails and messages they will receive according to predetermined parameters. Said system is actually used by most social media platforms as well. However, it is seen that the motivation of filtering a specific post with visual content, which is included in the invention of the description, independently of the user who shared it, if the user does not want to see it again, and hiding it forever or for a certain period of time from the main feed screen of the user who indicates that they do not want to see it, is not included in the US patent document mentioned in the paragraph.

[0010] In the PCT patent document numbered W02015006797A1 , which was encountered in the literature search, a method and system related to social media and a central management portal for social media is described. In the aforementioned patent, the validity of the posts is verified, and if the post does not pass the verification, it is blocked. However, the feature of blocking posts with visual content in the invention of the description independently of the users is not included in the patent document in the paragraph.

[0011] In the US patent document numbered US2020257742A1, which was encountered in the literature search, a system for configuring post filtering preferences via the dislike button is described. When the user presses the dislike button, the relevant post is hidden from the user main feed screen. However, if the post is shared again by another user, this may cause it to fall on the main account of the user. Because in the relevant patent, there is no analysis on the image of the post. In this case, users continue to be exposed to photos and videos that they do not want, especially those that are viral or frequently shared.

[0012] As a result, there is a need for a new visual content post blocking system and method in which the state of the art is gone beyond and the disadvantages thereof are eliminated.

[0013] Brief Description of the Invention:

[0014] The invention is a new system and method for blocking posts with visual content for use on social media platforms, in which the state of the art is gone beyond and the disadvantages thereof are eliminated.

[0015] The object of the invention is to introduce a system and method in which users can block posts with visual content that they do not want to see again, regardless of the user who shared it, and hide the image in said post from the main feed screen forever or for a certain period of time.

[0016] Another object of the invention is to reveal a system and method that provides users to use their social media accounts more efficiently and prevents them seeing the same posts or visual content repeatedly.

[0017] The invention identifies images in posts by a hash-based analysis or a machine learning-based analysis. According to these definitions, a comparison is made with the posts to be published on the user home screen, and posts with the same or very similar images are hidden from the user main feed screen.

[0018] The visual content filter performed in the invention can be selected manually, or as an automatic mode. Manually blocking a visual content can be summarized as follows. If the user blocks the post with visual content that he / she sees for the first time so that he does not see it anymore, an identification is made with machine learning algorithms or a hash-based analysis, and posts with the same or similar image are prevented from falling on the main user flow screen, regardless of who shares them. In the automatic mode, for each post with visual content that falls on the user's home screen, an identification is made with machine learning algorithms or a hash-based analysis, and posts with the same or similar images are prevented from falling on the main user feed screen for the second time, regardless of who shares them. Although the automatic mode has the disadvantage of slowing down the system, the usage efficiency of the social media platform is improved because users will not see the same posts twice.

[0019] The objects and features of the invention are summarized below;

[0020] • The user can block a post that contains a specific video or photo.

[0021] • The system automatically recognizes and hides all future posts that contain this visual content.

[0022] • The blocking process is content-based, not account- based. The content of users who repeatedly post (repost) the same video is also hidden in this way.

[0023] • The system improves the user experience and prevents disturbing content from appearing repeatedly.

[0024] Description of the Drawings:

[0025] The invention will be explained with reference to the attached figures, so that the features of the invention will be understood more clearly. However, this is not intended to limit the invention to these particular embodiments. On the contrary, it is intended to cover all alternatives, modifications and equivalents of the invention which may be included within the scope of the invention as defined by the appended claims. It should be understood that the details shown are only for illustration of the preferred embodiments of the present invention, and are presented for the purpose of providing the most useful and easy-to-understand description of both the embodiment of the methods and the rules and conceptual features of the invention. In these drawings:

[0026] Fig. - 1 is a schematic view of the system of the invention.

[0027] Fig. - 2 is a schematic view showing the process steps of the hash-based analysis used in the system of the invention.

[0028] Fig. - 3 is a schematic view showing the process steps of the machine learningbased analysis used in the system of the invention. The figures which will help to understand this invention are numbered as indicated in the attached drawings and are given below with their names.

[0029] Description of the references:

[0030] 10. User

[0031] 20. Device

[0032] 30. Social media platform

[0033] 31. Visual post

[0034] 32. Block button

[0035] 40. Server

[0036] 50. Database

[0037] Description of the Invention:

[0038] In this detailed description, the inventive visual content post blocking system and method are explained only for a better understanding of the subject matter, with examples that will not have any limiting effect.

[0039] The visual post (31) of the invention is a media content, which may be photographs or videos. In some cases, these posts may contain violence, sexuality, disturbing content or, independently of all, content that is not of interest to a user (10). In such cases, the user (10) can block only the visual post (31), not the sharing account, via the block button (32) and prevent the image in the relevant post from being published once again on the main screen on the social media platform (30). The image in the relevant post is quarantined for the blocking user (10), and the user (10) is hidden from the main feed screen forever or for a certain period of time. The device (20) mentioned in the invention is an element that provides access to social media platforms (30), which can be a phone, tablet or computer.

[0040] Figure 1 shows a schematic view of the visual post (31) blocking system of the invention. Accordingly, in the system, a user (10) can block a visual post on the social media platform (30) by pressing the block button (32) on the social media platform (31) via his device (20), which can be a phone, tablet or computer. In this case, the relevant post is quarantined in a database (50). Also, a server (40) performs a machine learning or hash-based analysis of the image in the relevant post and obtains an output. The output obtained is compared with all the posts to be published from the user (10) home screen, and a similarity score is obtained as a result of the comparison. If the similarity score is higher than a certain level (e.g. 80%), the relevant post is hidden from the main feed screen of the user (10).

[0041] Figure 2 shows the hash-based analysis performed by the inventive system by the server (40) thereof. A process of converting a string of any length into an algorithm and rendering it into an encrypted output is called hashing. Said output is also known as a hash value. In the invention, when the user (10) blocks a visual post (31), the image in the post is recorded in the database (50). Visual hash functions such as pHash and dHash are applied by the server (40) to the image in the blocked visual post (31), thus obtaining a hash value, which is an encrypted output. The hash value extracted by the server (40) is recorded in the database (50). Again, the server (40) in the invention instantly applies hash functions to the images in the visual posts (31) that are about to be published on the user (10) home screen and obtains the hash value. The hash value of the new visual post (31) to be published is compared with the hash value of the blocked image post (31) by the server (40), if the similarity rate between the hash values is more than XX%, the visual post (31) to be published is prevented from being published, and the posts containing the relevant image are hidden from the user (10) main feed screen.

[0042] In the invention, during the extraction of the hash value, certain key frames are selected from the videos and hash values are extracted from these frames. Hash values can be created directly for photos.

[0043] Figure 3 shows the machine learning algorithms used in the invention and the process of identifying and blocking the image in the visual post (31). ResNet, VGG Convolutional Neural Networks - CNN machine learning models are used as machine learning algorithms in the invention. In the invention, when the user (10) blocks a visual post (31), the image in the post is recorded in the database (50). Through said machine learning models, the videos or photos in the blocked visual post (31) in the database (50) are compared with the videos and photos in the newly published visual post (31), and if the images are the same or similar, it prevents the publication of the new visual post and hides the user (10) from the main feed screen. This process is carried out by the server (40) in the invention.

[0044] At least one, or in some cases, both of machine learning models or hash-based analysis can be used in the invention. The two techniques have advantages and disadvantages compared to each other. The hash-based analysis is fast and efficient, but robust to minor visual changes. The disadvantage is that large changes (e.g. cropping, resizing) can cause difficulty in detection. The machine learning based technique is more accurate and robust to various changes. The disadvantage is that it requires more computational power. Large data sets are needed for the models to work.

[0045] In the invention, the blocked visual posts (31) are stored in a database (50) as mentioned before. In order to keep the posts in the database (50) in an organised way, it is possible to tag them with data such as content ID, content type (video, photo), hash values, blocking users list. In the invention, indexing and caching methods can be used for efficient content recognition and tracking. Scalability can be achieved with distributed databases and parallel processing techniques.

[0046] Comparison with dozens of videos or photos that are blocked in the invention, i.e. quarantined, can lead to performance problems. Therefore, users (10) can be enabled to set the quarantine (block) time. Users (10) can specify how long the image will remain in quarantine (for example, 5 days, 10 days, etc.). Once this time has elapsed, the image can be automatically released from quarantine and made viewable again.

[0047] As mentioned before, in the system, by the automatic mode, a visual post (31) can be prevented from being seen twice by a user (10). For each visual post (31) that falls on the home screen of the user (10), an evaluation is made with machine learning algorithms or hash-based analysis, and posts with the same or similar images, regardless of who shares them, are prevented from falling on the main user (10) feed screen a second time. Although the automatic mode has the disadvantage of slowing down the social media platform (30), the usage efficiency of the social media platform (30) is improved as users (10) will not see the same posts twice.

Claims

CLAIMS1. A method for the operation of a system that enables visual posts (31) shared on social media platforms (30) to be blocked by a user (10), and all posts with the same content as the content in the blocked post are hidden from a user (10) feed screen, characterized in that it comprises the steps of:• when a user (10) blocks a visual post (31), saving the image in the post in a database (50),• obtaining a hash value, which is an encrypted output, by applying visual hash functions by a server (40) to the image in the blocked visual post (31),• recording the hash value extracted by a server (40) in the database (50),• obtaining, by the server (40), a hash value by applying hash functions to the images in the visual posts (31) that are about to be published on a user (10) home screen instantaneously,• comparing the hash value of a new visual post (31) to be published and the hash value of the blocked image post (31) by the server (40),• if the similarity rate between hash values is more than XX%, hiding the visual post (31) to be published from the user (10) main feed screen.

2. A method according to Claim 1 , characterized in that it comprises the steps of:• obtaining the hash value by evaluating each visual post (31) that falls on the user's (10) home screen with an automatic mode by a hash-based analysis,• preventing visual posts (31) with the same or similar hash value, regardless of who they are shared by, from falling a second time on the main user (10) feed screen.

3. A method for the operation of a system that enables visual posts (31) shared on social media platforms (30) to be blocked by a user (10), and all posts with the same content as the content in the blocked post are hidden from a user (10) feed screen, characterized in that it comprises the steps of:• when a user (10) blocks a visual post (31), saving the image in the post in a database (50),• comparing the videos or photos in the blocked visual post (31) in the database (50) with the videos and photos in the new visual post (31) to be published using machine learning models by the server (40),• if the images are the same or similar, hiding the new visual post (31) to be published from the user (10) main feed screen.

4. A method according to Claim 1 , characterized in that it comprises the steps of:• making an evaluation with machine learning algorithms-based analysis for each visual post (31) that falls on the user (10) home screen,• preventing posts with the same or similar images, regardless of who shares them, from appearing on the main user (10) feed screen a second time.

5. A system that enables visual posts (31) shared on social media platforms (30) to be blocked by a user (10), and all posts with the same content as the content in the blocked post are hidden from a user (10) feed screen, characterized in that it comprises:• a block button (32) allowing to block the visual post (31) on the social media platform (30) accessed by the user (10) via a device (20),• at least one database (50) in which the post blocked by said block button (32), outputs of the machine learning or hash-based analysis and the tag data of the image are stored,• at least one server (40) that performs machine learning or hash-based analysis on the images in the blocked visual post (31).

Images