Digital sex crime content monitoring system and method using artificial intelligence

Abstract

Disclosed are a digital sex crime content monitoring system and method using artificial intelligence. According to the present invention, illegal digital sex crime content posted online may be monitored to be tracked, searched, and deleted, thereby minimizing damage, improving work efficiency and reducing mental burden of damage support organizations, and illegal digital sex crime content targeting children / adolescents that has not been reported by victims may be preemptively deleted to prevent its spread, distribution, and re-distribution.

Description

Digital Sex Crime Content Monitoring System and Method Using Artificial Intelligence

[0001] The present invention relates to a system and method for monitoring digital sex crime content using artificial intelligence, and more specifically, to a system and method for monitoring digital sex crime content using artificial intelligence that can minimize damage by tracking, searching, and managing illegal digital sex crime content posted online so that it is deleted, and improve work efficiency and reduce mental burden on victim support officers.

[0002] The development of computer and mobile communications has enabled the widespread use of digital content through online web services and social networking services (SNS).

[0003] Although these web services and social networking services have made it easier for anyone to obtain information, a large amount of harmful content is being indiscriminately distributed through these services, which is becoming a social problem.

[0004] This harmful content may include digital sex crime content, and digital sex crime content that spreads rapidly in a short period of time through web services and social network services is emerging as a social problem by violating the human rights of victims and damaging public sentiment.

[0005] Recently, the anxiety of parents and citizens has been growing due to the increase in digital sex crimes targeting children and adolescents, such as Nth Room and Welcome to Video, and the related illegal digital sex crime content.

[0006] Moreover, given the nature of online crimes, which allow for unlimited copying, the secondary and tertiary suffering suffered by victims of digital sex crimes is also serious.

[0007] Additionally, with the advancement of digital technology, methods of sexual exploitation of children and adolescents are becoming more sophisticated and diverse, and the scale of damage is also rapidly increasing.

[0008] According to a Seoul City survey conducted in July 2021, approximately 21.3% (856 people) of 4,012 elementary, middle, and high school students have experienced digital sex crimes through chat or social media.

[0009] Additionally, according to the survey results, 56.4% of children / adolescents exposed to digital sex crimes stated that they had received sexual messages or photos, and 27.2% of them responded that they had received persistent online contact and requests to meet.

[0010] Among respondents, 4.8% reported having been exposed to sexual images or threats of distribution, and 4.3% reported being offered money for sexual acts.

[0011] This suggests that digital sex crimes are spreading rapidly and that a proactive response is needed, especially for crimes targeting children and adolescents.

[0012] However, as the incidence of digital sex crimes targeting adults and children / youth continues to rise, continuous online content monitoring technology is required to prevent these crimes in advance and minimize damage. However, there is a problem in that it is not possible to effectively detect and classify digital sex crime content from various online service providers (OSPs).

[0013] (Patent Document 1) Korean Patent Publication No. 10-2021-0098651, published on August 11, 2021 (Title of Invention: Device and Method for Monitoring the Distribution of Pornography)

[0014] In order to solve these problems, the present invention aims to provide a digital sex crime content monitoring system and method using artificial intelligence that can minimize damage by tracking, searching, and managing illegal digital sex crime content posted online so that it can be deleted, and improve the work efficiency and reduce the mental burden of victim support officers.

[0015] One embodiment of the present invention is a digital sex crime content monitoring system using artificial intelligence, comprising: a data collection unit that, when information on an online service server to be searched, arbitrary prompt data, and data on a video of a victim are received from a user terminal, embeds the prompt data to extract a plurality of keywords, and accesses the online service server to be searched based on the extracted keywords to collect arbitrary content; a data detection unit that extracts text data and image data from the collected content, and analyzes data of the content including the title, date of creation, frequency of creation, author, and frequency of distribution of a post based on the extracted text data and image data to classify harmful content; a data analysis unit that analyzes whether the text, voice, and images included in the harmful content are harmful video data based on a similarity value obtained by comparing the harmful content with the data on the video of a victim using an artificial intelligence-based analysis model; And it includes an evidence collection / deletion management unit that stores the analysis results of the harmful content in a database, and if the similarity value is higher than a certain value, collects and collects evidence of one or more of the image capture for the harmful content, the posting title, the damaging video, the URL, the standard time, the poster, and the posting date related to the harmful content, and generates report data and deletion request information for the harmful content based on the collected evidence and outputs them to the corresponding online service server.

[0016] In addition, the data collection unit according to the above embodiment extracts keywords by embedding the input prompt data, and is characterized in that it extracts multiple keywords through cluster data analysis and data embedding using a Generative Pre-trained Transformer (GPT).

[0017] In addition, the data analysis unit according to the above embodiment is characterized by recognizing the victim's face from the image of the harmful content, predicting the victim's age and gender through classification based on the facial shape, left-right symmetry characteristics of the eyes, nose, and mouth, and angle and inclination characteristics of the victim's face, and classifying the content as digital sex crime content targeting adults or digital sex crime content targeting children / youth based on the prediction result.

[0018] In addition, the evidence collection / deletion management unit according to the above embodiment is characterized by collecting and collecting evidence of image capture of the digital sex crime content targeting children / youth, and one or more of the posting title, the video of the damage, the URL, the standard time, the poster, and the posting date of the digital sex crime content targeting children / youth, and additionally generating report data and deletion request information for the digital sex crime content targeting children / youth based on the collected evidence and outputting them to the corresponding online service server.

[0019] In addition, the evidence collection / deletion management unit according to the above embodiment is characterized by performing additional monitoring on the deletion result, redistribution, and re-spread status of the harmful content and digital sex crime content targeting children / youth in response to the deletion request based on the deletion request information for the harmful content and digital sex crime content targeting children / youth.

[0020] In addition, one embodiment of the present invention is a method for monitoring digital sex crime content using artificial intelligence, comprising: a) a step in which a content monitoring server receives information on an online service server to be searched, arbitrary prompt data, and damaging video data about a victim from a user terminal; b) a step in which the content monitoring server embeds the received prompt data to extract a plurality of keywords, and accesses the online service server to be searched based on the extracted keywords to collect arbitrary content; c) a step in which the content monitoring server extracts text data and image data from the collected content, and analyzes data of the corresponding content, including the title, date of creation, frequency of creation, author, and frequency of distribution, based on the extracted text data and image data, to classify harmful content; d) a step in which, if the classification result is illegal sex crime content targeting adults, the content monitoring server analyzes whether the harmful content is a damaging video based on a similarity value obtained by comparing the text, voice, and image included in the harmful content with the damaging video data using an artificial intelligence-based analysis model; And e) the content monitoring server stores the analysis results of the harmful content in a database, and if the similarity value is higher than a certain value, captures an image of the harmful content and collects evidence of one or more of the posting title, damaging video, URL, standard time, poster, and posting date related to the harmful content, and generates report data and deletion request information for the harmful content based on the collected evidence and outputs them to the corresponding online service server.

[0021] In addition, step a) according to the above embodiment is characterized in that the content monitoring server extracts keywords by embedding the input prompt data, and extracts multiple keywords through cluster data analysis and data embedding using a GPT (Generative Pre-trained Transformer).

[0022] In addition, step c) according to the above embodiment is characterized by further including a step in which the content monitoring server recognizes the victim's face from the image of the classified harmful content, predicts the age and gender of the victim through classification based on the facial shape, left-right symmetry characteristics of the eyes, nose, and mouth, and angle and inclination characteristics of the victim's face, and classifies the content as digital sex crime content targeting adults or digital sex crime content targeting children / youth based on the prediction result.

[0023] In addition, if the harmful content according to the above example is classified as digital sex crime content targeting children / youth,

[0024] d) The step further comprises a step in which the content surveillance server collects and verifies image capture of the digital sex crime content targeting children / youth, and at least one piece of evidence from among the posting title, the video of the damage, the URL, the standard time, the poster, and the posting date of the digital sex crime content targeting children / youth, and additionally generates report data and deletion request information for the digital sex crime content targeting children / youth based on the collected evidence, and outputs the data to the corresponding online service server.

[0025] In addition, step d) according to the above embodiment is characterized by further including a step in which the content monitoring server performs additional monitoring of the deletion result, redistribution, and re-spread status of the corresponding harmful content and digital sex crime content targeting children / youth in response to the deletion request based on deletion request information for the harmful content and digital sex crime content targeting children / youth.

[0026] The present invention has the advantage of minimizing damage by tracking, searching, and managing the deletion of illegal digital sex crime content posted online, and improving the work efficiency and reducing the mental burden on victim support officers.

[0027] In addition, the present invention has the advantage of being able to prevent the spread, distribution, and redistribution of illegal digital sex crime content targeting children / youth without reports from victims by preemptively deleting such content.

[0028] Figure 1 is a block diagram schematically showing a digital sex crime content monitoring system using artificial intelligence according to one embodiment of the present invention.

[0029] FIG. 2 is a block diagram illustrating the configuration of a content surveillance server of a digital sex crime content surveillance system using artificial intelligence according to the embodiment of FIG. 1.

[0030] FIG. 3 is a block diagram illustrating the configuration of a data analysis unit of a content surveillance server according to the embodiment of FIG. 2.

[0031] FIG. 4 is a block diagram illustrating the configuration of the collection / deletion management unit of the content surveillance server according to the embodiment of FIG. 2.

[0032] FIG. 5 is a flowchart illustrating a method for monitoring digital sexual crime content using artificial intelligence according to one embodiment of the present invention.

[0033] FIG. 6 is a flowchart showing the process of evidence collection and deletion of the digital sex crime content monitoring method using artificial intelligence according to the embodiment of FIG. 5.

[0034] Hereinafter, the present invention will be described in detail with reference to the preferred embodiments of the present invention and the accompanying drawings, on the premise that the same reference numerals in the drawings refer to the same components.

[0035] Before describing the specific content for implementing the present invention, it should be noted that components not directly related to the technical gist of the present invention are omitted within the scope that does not scatter the technical gist of the present invention.

[0036] In addition, terms or words used in this specification and claims should be interpreted as meanings and concepts conforming to the technical idea of the invention based on the principle that the inventor can define the concept of appropriate terms to describe his invention in the best way.

[0037] The expression that a certain part "includes" a certain component in this specification means that it can further include other components rather than excluding other components.

[0038] In addition, terms such as "‥part", "‥device", "‥module" mean a unit that processes at least one function or operation, and this can be distinguished by hardware, software, or a combination of both.

[0039] In addition, the term "at least one" is defined as a term including the singular and the plural, and it is obvious that each component can exist in the singular or the plural and can mean the singular or the plural even if the term "at least one" does not exist.

[0040] In addition, whether each component is provided in the singular or the plural can be changed according to the embodiment.

[0041] Hereinafter, with reference to the attached drawings, a preferred embodiment of a digital sex crime content monitoring system and method using artificial intelligence according to an embodiment of the present invention will be described in detail.

[0042] FIG. 1 is a block diagram schematically showing a digital sex crime content surveillance system using artificial intelligence according to an embodiment of the present invention, FIG. 2 is a block diagram illustrating a content surveillance server configuration of a digital sex crime content surveillance system using artificial intelligence according to the embodiment of FIG. 1, FIG. 3 is a block diagram illustrating a data analysis unit configuration of a content surveillance server according to the embodiment of FIG. 2, and FIG. 4 is a block diagram illustrating a collection / deletion management unit configuration of a content surveillance server according to the embodiment of FIG. 2.

[0043] As shown in FIGS. 1 to 4, a digital sex crime content monitoring system using artificial intelligence according to an embodiment of the present invention can be configured to include a user terminal (100), a content search server (200), a web server (300) that is a search target, and an SNS server (400) to track, search, and delete illegal digital sex crime content posted online so that damage can be minimized.

[0044] The user terminal (100) may be configured as a mobile terminal such as a desktop PC, laptop PC, tablet PC, or smart phone capable of installing an application program to enable tracking, searching, and requesting deletion of illegal digital sex crime content posted online by connecting to a content search server (200) via a network.

[0045] Additionally, the user terminal (100) can transmit the reported video of the damage to the content search server (200) through the victim's report.

[0046] Additionally, the user terminal (100) can receive information from an online service server, including a web server (300) and an SNS server (400) that are search targets, and transmit the information to the content search server (200).

[0047] Additionally, the user terminal (1000) can input prompt data including the content desired to be searched in relation to the damaged video as text data.

[0048] Here, the prompt data may include, for example, objects, language, new words, chat data, website, SNS data, etc. related to the damaged video.

[0049] In addition, the user terminal (100) receives and displays search results and evidence information related to the video content transmitted from the content search server (200), search results and evidence information for digital sex crime content targeting adults and digital sex crime content targeting children / youth, and deletion-related processing information for the content.

[0050] The content search server (200) is a configuration that tracks, searches, deletes, and manages illegal digital sex crime content posted online, and may be configured to include a data collection unit (210), a data detection unit (220), a data analysis unit (230), an evidence collection / deletion management unit (240), and a database (250).

[0051] The data collection unit (210) can receive any search target online service server information, any prompt data, and damage video data about the victim from the user terminal (100).

[0052] The data collection unit (210) can collect content related to harmful content and damaging video material through web crawling targeting the received search target web server (300) and SNS server (400).

[0053] In addition, the data collection unit (210) extracts keywords by embedding the input prompt data, and can extract multiple keywords through cluster data analysis and data embedding using GPT (Generative Pre-trained Transformer).

[0054] The data collection unit (210) extracts keywords related to objects such as 'books', 'school uniforms', and 'dolls', as well as language and new words mainly used by teenagers, from among the prompt data if the data contains content related to children and teenagers, thereby enabling the collection of more harmful content and damage-related video content.

[0055] In addition, the data collection unit (210) can extract information on the search target online service server based on extracted keywords in addition to the search target online service server input from the user terminal (100), and collect harmful content and damage-related video content through web crawling by accessing the corresponding online service.

[0056] The data detection unit (220) can extract text data and image data from content collected through web crawling.

[0057] In addition, the data detection unit (220) can classify harmful content into digital sex crime content targeting adults and digital sex crime content targeting children / youth by analyzing the data of the content including the title, date of creation, frequency of creation, author, and frequency of distribution of the post based on the extracted text data and image data.

[0058] In addition, the data detection unit (220) can correct incorrectly written words or strings in collected content-related data based on preset correction rules, and classify the content into text data and image data.

[0059] Here, the correction rule can distinguish between typos and correct answers for extracted words or strings, and generate correction information by matching the distinguished typos and correct answers.

[0060] In addition, the data detection unit (220) can set words and strings used for each keyword and extract keyword information for the content using the BERT model, which is a natural language model, from the collected content-related data.

[0061] The data analysis unit (230) can compare the text, voice, and images included in the harmful content with the harmful video data using an artificial intelligence-based analysis model to calculate a similarity value.

[0062] In addition, the data analysis unit (230) can analyze whether the video is harmful based on a similarity value obtained by comparing harmful content and harmful video data, and can be configured to include a similarity analysis unit (231).

[0063] The similarity analysis unit (231) can recognize text data included in harmful content through optical character recognition (OCR), and can output an embedding vector that converts text values ​​of the recognized harmful content into vector values ​​by positioning them in a vector space.

[0064] In addition, the similarity analysis unit (231) can calculate a similarity value according to the similarity through matrix operation with keyword information based on the position of each vector and the produced embedding vector.

[0065] In addition, the similarity analysis unit (231) can analyze the patterns of posts, such as the frequency of posting, the post author, and the frequency of distribution, through similarity analysis of text data.

[0066] In addition, the similarity analysis unit (231) can analyze voices included in harmful content using a learned artificial intelligence-based analysis model, extract a feature vector of the analyzed voice, and calculate a similarity value based on the similarity with the damage video data based on acoustic and linguistic characteristics from the feature vector of the extracted voice.

[0067] In addition, the similarity analysis unit (231) can analyze images included in harmful content using a learned artificial intelligence-based analysis model, extract a feature vector of the analyzed image, and calculate a similarity value based on the similarity with the damage video data based on the image characteristics from the feature vector of the extracted image.

[0068] Additionally, the data analysis unit (230) can classify content into digital sex crime content targeting adults or digital sex crime content targeting children / youth.

[0069] To this end, the data analysis unit (230) may be configured to include a prediction analysis unit (232) to recognize the victim's face from an image of harmful content and to predict the victim's age and gender through classification based on the facial shape, left-right symmetry characteristics of the eyes, nose, and mouth, and angle and inclination characteristics of the recognized victim's face.

[0070] The predictive analysis unit (232) can extract facial recognition data that recognizes the victim's face from an image included in harmful content, and the facial recognition data can include the user's facial shape, left-right symmetry characteristics of the eyes, nose, and mouth, angles and inclinations, etc., and the characteristics of the extracted facial recognition data can be vectorized.

[0071] The predictive analysis unit (232) can use an artificial intelligence-based analysis model that learns to classify by age group and gender through machine learning using a CNN (Convolutional Neural Network) algorithm based on a labeled learning data set by acquiring facial appearances by age group and gender.

[0072] Here, the AI-based analysis model can learn age and gender classification from image-based learning data using a deep learning model based on a convolutional neural network (CNN).

[0073] Additionally, AI-based analysis models can be seen as a type of analysis model created through a method called deep learning among machine learning.

[0074] Therefore, an AI-based analysis model can also be used as a deep learning model or as a representation of a deep learning analysis model.

[0075] Additionally, machine learning is an application of artificial intelligence that enables complex systems to automatically learn and improve from experience without being explicitly programmed.

[0076] Additionally, the accuracy and effectiveness of machine learning models may partially depend on the data used to train them.

[0077] Additionally, the AI-based analysis model can further modify the analysis model by retraining the analysis model and verifying its performance using the classification data on the age and gender of the victims collected during the classification process.

[0078] That is, the predictive analysis unit (232) performs additional feature engineering, hyper parameter optimization, and auto machine learning using a champion model to improve the performance of the analysis model, thereby enabling analysis to be performed without analysis by an analyst and improving usability.

[0079] Here, feature engineering is a method of improving model performance by selecting meaningful variables that can be used to improve the model from existing data variables, transforming them, or creating new variables that affect predictive power.

[0080] In addition, hyperparameter optimization refers to a technique for finding hyperparameters that can produce the best performance when training an artificial neural network. Hyperparameters can include learning rate, learning flow scheduling method, loss function, number of training iterations, weight initialization method, regularization method, and number of layers to be stacked.

[0081] Additionally, the prediction analysis unit (232) can generate an additional learning data set through data augmentation and learn using the generated additional learning data set.

[0082] In addition, the prediction analysis unit (232) can obtain a result value predicting the age and gender of the victim through feature vector classification of the extracted facial recognition data.

[0083] Through this, the predictive analysis unit (232) can classify the content as digital sex crime targeting children / adolescents if the age of the victim is predicted to be, for example, a teenager or a child.

[0084] The evidence collection / deletion management unit (240) stores the analysis results of harmful content analyzed by the data analysis unit (230) in the database (250).

[0085] Additionally, the evidence collection / deletion management unit (240) may be configured to include an evidence extraction unit (241) and a deletion monitoring unit (242).

[0086] The evidence extraction unit (241) can capture an image of the harmful content if it is classified as illegal sexual crime content targeting adults and the similarity value between the harmful content analyzed by the data analysis unit (230) and the damage video data is greater than a preset value.

[0087] In addition, the evidence extraction unit (241) can collect evidence such as the posting title, the damaging video, URL, standard time, poster, and posting date related to the harmful content and perform evidence collection.

[0088] In addition, the evidence extraction unit (241) generates report data and deletion request information for the harmful content based on the evidence collected through evidence collection, and outputs the generated report data and deletion request information to the online service server that posted the harmful content and to the user terminal (100).

[0089] In addition, if the evidence extraction unit (241) is classified as digital sex crime content targeting children / youth by the data analysis unit (230), it can capture images of the digital sex crime content targeting children / youth.

[0090] In addition, the evidence extraction unit (241) can collect evidence such as the posting title, video of the victim, URL, standard time, poster, and posting date for digital sex crime content targeting children / youth, and perform evidence collection.

[0091] In addition, the evidence extraction unit (241) additionally generates report data on digital sex crime content targeting children / youth and request information for deletion of digital sex crime content targeting children / youth based on the evidence collected through evidence collection, and outputs the data to the online service server that posted the harmful content and to the user terminal (100).

[0092] The deletion monitoring unit (242) monitors the deletion results of adult digital sex crime content and child / youth digital sex crime content in response to deletion requests based on deletion request information for adult digital sex crime content and child / youth digital sex crime content, and monitors the status of redistribution and re-dissemination of the content.

[0093] Through this, the deletion monitoring unit (242) can operate to preemptively delete illegal digital sex crime content targeting children / youth that has not been reported by the victim, thereby preventing the spread, distribution, and redistribution of such content.

[0094] The database (250) stores search results and evidence information related to video content, search results and evidence information for digital sex crime content targeting adults and digital sex crime content targeting children / youth, and deletion-related processing information for the content.

[0095]

[0096] The following describes a method for monitoring digital sexual crime content using artificial intelligence according to one embodiment of the present invention.

[0097] FIG. 5 is a flowchart illustrating a method for monitoring digital sex crime content using artificial intelligence according to an embodiment of the present invention, and FIG. 6 is a flowchart illustrating an evidence collection and deletion process of a method for monitoring digital sex crime content using artificial intelligence according to the embodiment of FIG. 5.

[0098] Referring to FIGS. 1 to 6, a method for monitoring digital sexual crime content using artificial intelligence according to an embodiment of the present invention is such that a content monitoring server (200) receives information on an online service server for a random search, random prompt data, and damage video data on a victim from a user terminal (100) (S100).

[0099] The content monitoring server (200) embeds the received prompt data to extract multiple keywords, and connects to the search target online service server based on the extracted keywords to collect arbitrary content (S200).

[0100] In step S200, the content monitoring server (200) can collect content related to harmful content and damaging video material through web crawling targeting the received search target web server (300) and SNS server (400).

[0101] In addition, in step S200, the content surveillance server (200) extracts keywords by embedding prompt data, and by extracting multiple keywords through cluster data analysis and data embedding using a Generative Pre-trained Transformer (GPT), it is possible to collect more harmful content and damage-related video content.

[0102] In addition, in step S200, the content monitoring server (200) can extract information on the search target online service server based on extracted keywords in addition to the search target online service server input from the user terminal (100), and access the online service to collect harmful content and damage-related video content through web crawling.

[0103] Continuing, the content monitoring server (200) extracts text data and image data from the collected content, and analyzes the pattern of the content, including the frequency of posting, author, and distribution frequency, based on the extracted text data and image data, to detect (S300) adult digital sex crime content or child / youth digital sex crime content classified as harmful content.

[0104] That is, in step S300, the content monitoring server (200) extracts text data and image data from content collected through web crawling, and analyzes the pattern of the content, including the frequency of posting, author, and distribution frequency, based on the extracted text data and image data, to classify harmful content.

[0105] At step S300, the content monitoring server (200) recognizes the victim's face from an image of harmful content, and can predict the victim's age and gender through classification based on the facial shape, left-right symmetry characteristics of the eyes, nose, and mouth, and angle and inclination characteristics of the victim's face.

[0106] Through this, the content monitoring server (200) can classify content as digital sex crime content targeting adults or digital sex crime content targeting children / youth based on the prediction result.

[0107] Continuing, if the harmful content classified in step S300 is digital sex crime content targeting adults, the content monitoring server (200) can analyze whether the reported harmful video content is a harmful video content based on a similarity value obtained by comparing the text, voice, and image included in the harmful content with the harmful video data using an artificial intelligence-based analysis model (S400).

[0108] At step S400, the content monitoring server (200) recognizes text data included in harmful content through optical character recognition (OCR), and converts the text values ​​of the recognized harmful content's text data into vector values ​​by positioning them in a vector space, and calculates a similarity value according to the similarity through matrix operation with keyword information based on the position of each vector.

[0109] At step S400, the content monitoring server (200) can analyze the patterns of posts, such as the frequency of posting, the post author, and the frequency of distribution, through similarity analysis of text data.

[0110] At step S400, the content monitoring server (200) analyzes voices included in harmful content using a learned artificial intelligence-based analysis model, extracts a feature vector of the analyzed voices, and calculates a similarity value based on the similarity with the harmful video data based on acoustic and linguistic characteristics from the feature vector of the extracted voices.

[0111] At step S400, the content monitoring server (200) analyzes images included in harmful content using a learned artificial intelligence-based analysis model, extracts a feature vector of the analyzed image, and calculates a similarity value based on the similarity with the damage video data based on the image characteristics from the feature vector of the extracted image.

[0112] Continuing, the content monitoring server (200) stores the analysis results of harmful content in a database (250) and manages (S500) the collection of evidence and deletion of the harmful content.

[0113] At step S500, if the calculated similarity value is greater than a certain value, the content monitoring server (200) captures an image of the harmful content and collects evidence of one or more of the posting title, damaging video, URL, standard time, poster, and posting date related to the harmful content to collect evidence (S510).

[0114] In addition, the content monitoring server (200) generates report data and deletion request information for the harmful content based on evidence collected through evidence collection, and outputs the generated report data and deletion request information to the online service server that posted the harmful content and the user terminal (100) (S520).

[0115] In addition, the content monitoring server (200) monitors the deletion result of whether the deletion of the adult digital sex crime content in response to the deletion request was performed based on the deletion request information for the adult digital sex crime content, and the status of redistribution and re-dissemination of the content (S530).

[0116] In addition, the content monitoring server (200) can analyze the monitoring results and generate a result report in any format regarding the deletion status, redistribution, and redistribution status, and report it to the user terminal (100) (S540).

[0117] Meanwhile, in step S300, if the harmful content is classified as digital sex crime content targeting children / youth, the content monitoring server (200) can capture images of the digital sex crime content targeting children / youth.

[0118] In addition, the content monitoring server (200) can collect evidence such as the posting title, the video of the victim, URL, standard time, poster, and posting date for digital sex crime content targeting children / youth, and perform evidence collection.

[0119] In addition, the content monitoring server (200) additionally generates report data on digital sex crime content targeting children / youth and request information for deletion of digital sex crime content targeting children / youth based on evidence collected through evidence collection, and outputs the data to the online service server that posted the harmful content and to the user terminal (100).

[0120] In addition, the content monitoring server (200) monitors the deletion results for whether the digital sex crime content targeting children / youth has been deleted in response to the deletion request based on the deletion request information for the digital sex crime content targeting children / youth, and the redistribution and re-dissemination status of the content, and analyzes the monitoring results to generate a result report in any format for the deletion status, redistribution and re-dissemination status, and reports it to the user terminal (100).

[0121] Therefore, by tracking, searching, and managing the deletion of illegal digital sex crime content posted online, damage can be minimized, and the work efficiency of victim support officers can be improved and the mental burden reduced.

[0122] Additionally, by preemptively deleting illegal digital sex crime content targeting children and adolescents without reports from victims, the spread, distribution, and redistribution of such content can be prevented.

[0123] As described above, although the present invention has been described with reference to preferred embodiments, it will be understood by those skilled in the art that the present invention can be variously modified and changed within the scope and spirit of the present invention as set forth in the claims below.

[0124] In addition, the drawing numbers described in the patent claims of the present invention are not limited thereto and are described only for the sake of clarity and convenience of explanation, and in the process of explaining the embodiments, the thickness of lines or the size of components depicted in the drawings may be exaggerated for the sake of clarity and convenience of explanation.

[0125] In addition, the terms described above are terms defined in consideration of their functions in the present invention, and may vary depending on the intention or custom of the user or operator. Therefore, the interpretation of these terms should be based on the contents throughout this specification.

[0126] In addition, even if not explicitly illustrated or described, it is obvious that a person having ordinary skill in the art to which the present invention pertains can make various modifications including the technical ideas of the present invention from the description of the present invention, and this still falls within the scope of the present invention.

[0127] In addition, the above embodiments described with reference to the attached drawings are described for the purpose of explaining the present invention, and the scope of the present invention is not limited to these embodiments.

Claims

1. When information on an online service server for a search, arbitrary prompt data, and damage video data for a victim are received from a user terminal (100), a data collection unit (210) that embeds the prompt data to extract multiple keywords, and connects to the online service server for a search based on the extracted keywords to collect arbitrary content; A data detection unit (220) that extracts text data and image data from the collected content, and analyzes the data of the content, including the title, creation date, creation frequency, author, and distribution frequency of the post, based on the extracted text data and image data, to classify harmful content; A data analysis unit (230) that analyzes whether the text, voice, and image included in the harmful content are harmful video data based on a similarity value obtained by comparing the harmful content and the harmful video data using an artificial intelligence-based analysis model; and A digital sex crime content monitoring system using artificial intelligence, including an evidence collection / deletion management unit (240) that stores the analysis results of the harmful content in a database (250), and, if the similarity value is greater than a certain value, collects and collects evidence of one or more of image captures of the harmful content, the posting title, the damaging video, URL, standard time, the poster, and the posting date related to the harmful content, and generates report data and deletion request information for the harmful content based on the collected evidence and outputs them to the corresponding online service server.

2. In paragraph 1, The above data collection unit (210) extracts keywords by embedding the input prompt data, and is a digital sex crime content monitoring system using artificial intelligence characterized in that it extracts multiple keywords through cluster data analysis and data embedding using GPT (Generative Pre-trained Transformer).

3. In paragraph 2, The above data analysis unit (230) recognizes the victim's face from the image of the harmful content, and predicts the victim's age and gender through classification based on the facial shape, left-right symmetry characteristics of the eyes, nose, and mouth, and angle and inclination characteristics of the victim's face. A digital sex crime content monitoring system using artificial intelligence characterized by classifying content as digital sex crime content targeting adults or digital sex crime content targeting children / youth based on the above prediction results.

4. In paragraph 3, The above-mentioned evidence collection / deletion management unit (240) collects and collects evidence of one or more of image captures for the digital sex crime content targeting children / youth, the posting title, the damaging video, URL, standard time, the poster, and the posting date for the digital sex crime content targeting children / youth, and additionally creates report data and deletion request information for the digital sex crime content targeting children / youth based on the collected evidence and outputs them to the corresponding online service server. A digital sex crime content surveillance system using artificial intelligence.

5. In paragraph 4, The above-mentioned evidence collection / deletion management unit (240) is characterized in that it performs additional monitoring of the deletion results, redistribution, and re-spread status of the harmful content and digital sex crime content targeting children / youth in response to the deletion request based on the deletion request information for the harmful content and digital sex crime content targeting children / youth. A digital sex crime content monitoring system using artificial intelligence. 6.a) A step in which the content monitoring server (200) receives information on an online service server to be searched, arbitrary prompt data, and damage video data for the victim from a user terminal (100); b) A step in which the content monitoring server (200) embeds the received prompt data to extract a plurality of keywords, and connects to the search target online service server based on the extracted keywords to collect arbitrary content; c) A step of extracting text data and image data from the collected content by the content monitoring server (200), and analyzing the data of the content including the title, creation date, creation frequency, author, and distribution frequency of the post based on the extracted text data and image data to classify the harmful content; d) If the classification result is that the content is an illegal sexual crime targeting adults, the content monitoring server (200) analyzes whether the content is a harmful video based on a similarity value obtained by comparing the text, voice, and image included in the harmful content with the harmful video data using an artificial intelligence-based analysis model; and e) A method for monitoring digital sex crime content using artificial intelligence, comprising: a step of storing the analysis results of the harmful content in a database (250) by the content monitoring server (200), and, if the similarity value is greater than a certain value, collecting and collecting evidence of one or more of image captures of the harmful content, the posting title, the damaging video, URL, standard time, the publisher, and the posting date related to the harmful content, and generating report data and deletion request information for the harmful content based on the collected evidence and outputting them to the corresponding online service server; 7. In paragraph 6, The above step a) is a digital sex crime content surveillance method using artificial intelligence, characterized in that the content surveillance server (200) extracts keywords by embedding the input prompt data, and extracts multiple keywords through cluster data analysis and data embedding using a GPT (Generative Pre-trained Transformer).

8. In paragraph 7, The step c) above further includes a step of recognizing the face of the victim from the image of the classified harmful content by the content monitoring server (200), predicting the age and gender of the victim through classification based on the facial shape, left-right symmetry characteristics of the eyes, nose, and mouth, and angle and inclination characteristics of the victim's face, and classifying the content as digital sex crime content targeting adults or digital sex crime content targeting children / youth based on the prediction result; A method for monitoring digital sex crime content using artificial intelligence, characterized in that the step further includes a step of:

9. In paragraph 8, A method for monitoring digital sex crime content using artificial intelligence, characterized in that if the harmful content is classified as digital sex crime content targeting children / youth, the content monitoring server (200) collects and collects evidence of at least one of image capture for the digital sex crime content targeting children / youth, the posting title, the damaging video, the URL, the standard time, the poster, and the posting date for the digital sex crime content targeting children / youth, and additionally generates report data and deletion request information for the digital sex crime content targeting children / youth based on the collected evidence and outputs the data to the corresponding online service server.

10. In paragraph 9, A method for monitoring digital sex crime content using artificial intelligence, characterized in that the content monitoring server (200) further includes a step of performing additional monitoring on the deletion result, redistribution, and re-spread status of the corresponding harmful content and digital sex crime content targeting children / youth in response to the deletion request based on the deletion request information for the harmful content and digital sex crime content targeting children / youth.

Images