AI-Based Unstructured Data Processing System

Abstract

An AI-based unstructured data processing system comprises a data processing server including a plurality of user terminals having programs corresponding to a plurality of extensions, a server communication unit communicating with the plurality of user terminals, a server input unit for inputting a manager's command, and a data processing server including an artificial intelligence server control unit that generates a pixelation control signal to divide and transmit unstructured data into a plurality of pixels while displaying the unstructured data on a screen to at least one user terminal based on a user command that collects and organizes unstructured data among at least one of the plurality of user terminals input to the server input unit, controls the server communication unit to transmit the generated pixelation control signal to at least one user terminal, and generates screen data by concatenating the plurality of pixelated data received from at least one user terminal.

Description

Technology Field

[0001] The present invention relates to an AI-based unstructured data processing system, and more specifically, to an AI-based unstructured data processing system that reads and stores unstructured data in a standardized state. Background Technology

[0003] As we enter the full-scale era of AI, the importance of data is growing alongside it. Many companies are seeking to achieve business innovation by combining data and AI. They are also striving to utilize data strategically. In fact, most companies already possess vast amounts of unstructured data.

[0004] Data companies are also announcing various features for utilizing unstructured data. When we commonly speak of "data," what comes to mind is structured data. It is organized into distinct rows and columns, and information is recorded within a framework such as tables or fields. On the other hand, unstructured data refers to data that lacks such a fixed format or structure. Examples include emails, chat logs, work manuals, reports, customer consultation transcripts, and product images. If structured data is information that computers can understand, then unstructured data is information that humans can naturally comprehend.

[0005] As such, most of the data routinely generated by companies is unstructured. According to research, unstructured data accounts for over 80% of corporate data, yet it has largely been neglected due to the difficulty of utilizing it. This is because its diverse and flexible formats make it difficult for computers to process automatically, and the sheer volume makes it too massive for humans to analyze manually. However, with the advancement of generative AI and natural language processing technologies, avenues for utilizing unstructured data have now opened up and are receiving significant attention.

[0006] However, corporate files—including Word, PDF, Excel, Hangul, images, and emails—diverse in format, extension, and structure, making it very difficult for AI to learn. AI cannot directly understand various extensions such as docx, xlsx, pdf, and hwp. Most unstructured data is a dormant asset; since it is largely siloed and not systemically connected, it poses a problem in that it is difficult to process and store as data. Furthermore, large language models generate results only after processing all tokens included in the prompt. This leads to a problem where processing time and resource consumption increase as the number of input tokens increases. Additionally, inputting unnecessary code blocks or entire folders directly poses a risk of exceeding the context window limit, which can result in wasted hardware resources. Prior art literature

[0008] Republic of Korea Published Patent No. 10-2024-0067541 AI-based classification system and method for unstructured data (Publication Date: May 17, 2024) The problem to be solved

[0009] Therefore, the objective of the present invention is to provide an AI-based unstructured data processing system capable of reading and storing unstructured data in a standardized state, regardless of the type of unstructured data. means of solving the problem

[0011] An AI-based unstructured data processing system according to the present invention for achieving the above objective comprises: a plurality of user terminals having programs corresponding to a plurality of file extensions; and a data processing server comprising a server communication unit communicating with the plurality of user terminals, a server input unit for inputting a manager's command, and a data processing server comprising an artificial intelligence server control unit that generates a pixelation control signal to divide and transmit unstructured data into a plurality of pixels while displaying the unstructured data on a screen to at least one user terminal based on a user command that collects and organizes unstructured data among at least one of the plurality of user terminals input to the server input unit, controls the server communication unit to transmit the generated pixelation control signal to the at least one user terminal, and generates screen data by concatenating a plurality of pixelated data received from the at least one user terminal. The artificial intelligence server control unit of the data processing server transmits a pixelation control signal to each of the multiple user terminals, and in response to the transmission, the multiple received pixelated data are connected to create and store screen data. Since the data is stored in a screen state regardless of the extension of the unstructured data, unstructured data that could not be utilized can be practically utilized, thereby improving the storability and usability of unstructured data.

[0012] Here, the pixelation control signal includes a control command for distinguishing multiple object regions while the unstructured data is displayed on a screen, and the pixelation control signal includes a control command for assigning important weights to each of the distinguished multiple object regions. This is desirable because it allows for different pixelation of important and less important parts within the unstructured data, thereby improving the efficiency of data storage.

[0013] Here, it is desirable that the pixelation control signal includes a control command that differentially assigns important weights based on the main field of view and the peripheral field of view, as this allows the resolutions of the important and non-important parts to be stored differently, thereby significantly reducing the data storage space.

[0014] And the pixelation control signal is desirable because if it includes a control command that divides the resolution of the object in the main viewing area to be at least four times higher than the resolution of the object in the peripheral viewing area, the amount of data in non-important areas other than the most important area within the screen area can be significantly reduced, thereby increasing the data storage speed, utilization speed, and data search speed.

[0015] Here, it is desirable that the pixelation control signal includes metadata containing data items and corresponding data values ​​for each of the separated plurality of object regions, so that omissions or errors in data storage can be minimized.

[0016] Furthermore, it is desirable that the pixelation control signal includes a control command that classifies the separated plurality of object areas into one of text, table, button, icon, and image, as this simplifies the method of distinguishing objects and thus increases the data processing speed. Effects of the invention

[0018] According to the present invention, an artificial intelligence server control unit of a data processing server transmits a pixelation control signal to each of a plurality of user terminals, and in response to the transmission, a plurality of received pixelation data are connected to create and store screen data. Since the data is stored in a screen state regardless of the extension of the unstructured data, unstructured data that could not be utilized can be practically utilized, thereby improving the storability and usability of unstructured data.

[0019] In addition, by distinguishing between important and less important parts within unstructured data and pixelating them differently, there is an effect of improving the efficiency of data storage.

[0020] In addition, since the resolution of key and non-key parts can be stored differently, it has the effect of significantly reducing data storage space.

[0021] In addition, since the amount of data in non-critical areas other than the most important areas within the screen can be significantly reduced, there is an effect of increasing data storage speed, utilization speed, and data search speed.

[0022] In addition, it has the effect of minimizing omissions or errors in data storage.

[0023] In addition, since the method of distinguishing objects can be simplified, there is an effect of increasing data processing speed. Brief explanation of the drawing

[0025] FIG. 1 is an overall configuration diagram and control block diagram of an AI-based unstructured data processing system according to the present invention. FIG. 2 is a flowchart of a first embodiment of an AI-based unstructured data processing method using an AI-based unstructured data processing system according to the present invention. Figure 3 is a flowchart of a second embodiment of an AI-based unstructured data processing method using an AI-based unstructured data processing system. Specific details for implementing the invention

[0026] Hereinafter, an AI-based unstructured data processing system (1) according to a preferred embodiment of the present invention will be described in detail with reference to the attached drawings.

[0027] FIG. 1 is an overall configuration diagram and control block diagram of an AI-based unstructured data processing system (1) according to the present invention.

[0028] The AI-based unstructured data processing system (1) includes a user terminal (10) and a data processing server (20).

[0029] The user terminal (10) is a plurality of components having programs corresponding to multiple extensions. The user terminal (10) includes a user communication unit (11), a user display unit (12), a user input unit (13), a user camera (14), a user speaker (15), a user microphone (16), and a user control unit (17).

[0030] The user communication unit (11) can perform wireless communication, and the wireless communication includes at least one of infrared communication, RF, Zigbee, and Bluetooth. The user communication unit (11) receives a video signal and transmits it to the user control unit (17) described later, and can be implemented in various ways corresponding to the specifications of the received video signal and the implementation form of the user terminal. For example, the user communication unit (11) can wirelessly receive an RF (radio frequency) signal transmitted from a broadcasting station (not shown), or receive a video signal according to composite video, component video, super video, SCART, HDMI (high definition multimedia interface) specifications, etc. via a wired connection. If the video signal is a broadcast signal, the user communication unit (11) may include a tuner that tunes the broadcast signal by channel.

[0031] The user display unit (12) displays an image based on an image signal processed by image processing. The implementation method of the user display unit (12) is not limited and can be implemented in various display methods such as liquid crystal, plasma, light-emitting diode, organic light-emitting diode, surface conduction electron-emitter, carbon nanotube, nano-crystal, etc.

[0032] The user display unit (12) may additionally include additional configurations depending on the implementation method. For example, if the user display unit (12) is a liquid crystal type, the user display unit (12) includes a liquid crystal display panel (not shown), a backlight unit (not shown) that supplies light thereto, and a panel driving board (not shown) that drives the panel (not shown). The user display unit (12) may display a voice recognition result as information regarding the recognized voice. Here, the voice recognition result can be displayed in various forms such as text, graphics, and icons, and the text includes characters and numbers. The user display unit (12) may further display candidate commands and application information based on the voice recognition result. The user can check whether the voice has been correctly recognized by the voice recognition result displayed on the user display unit (12), and can select a command corresponding to the voice spoken by the user from among the displayed candidate commands by operating the user input unit (13) provided on the remote control, or select and check information related to the voice recognition result.

[0033] The user input unit (13) may be composed of an input means that allows the user to input user commands. The user input unit (13) may receive the user's touch input or remote input using a remote controller and transmit it to the corresponding user control unit (17). Additionally, the user input unit (13) may receive voice input spoken by the user and transmit the voice signal to the user control unit (17). In that case, the user input unit (13) may be implemented, for example, as a microphone. The user input unit (13) may also perform signal processing on the received voice signal itself. However, the form of user input that the user input unit (13) can receive is not limited to this, and, for example, user input through motion recognition, etc., may also be received.

[0034] The user camera (14) can photograph the outside and the front including the user.

[0035] The user speaker (15) can output a total of voices including the voice of the video, the voice of the application, and the voice of the user.

[0036] The user microphone (16) can receive a whole of voices including the voice of the video, the voice of the application, and the voice of the user.

[0037] When the user control unit (17) receives a pixelation control signal from the data processing server (20) to divide and transmit unstructured data into multiple pixels while it is displayed on the screen, it generates multiple pixelated data by dividing the unstructured data into multiple pixels while it is displayed on the screen, and controls the server communication unit (21) to transmit the generated multiple pixelated data to the data processing server (20).

[0038] When the user control unit (17) receives a pixelation control signal from the data processing server (20) that includes a control command for distinguishing multiple object areas while unstructured data is displayed on the screen, it distinguishes multiple object areas while unstructured data is displayed on the screen, divides each of the distinguished multiple object areas into multiple pixels to generate multiple pixelated data, and controls the server communication unit (21) to transmit the generated multiple pixelated data to the data processing server (20).

[0039] When the user control unit (17) receives a pixelation control signal from the data processing server (20) that includes a control command for assigning important weights to each of the separated multiple object areas, assigns important weights to each of the separated multiple object areas, divides each of the multiple object areas assigned important weights into multiple pixels to generate multiple pixelation data, and controls the server communication unit (21) to transmit the generated multiple pixelation data to the data processing server (20).

[0040] When the user control unit (17) receives a pixelation control signal from the data processing server (20) that includes a control command for differentially assigning important weights based on the main view area and the peripheral view area, it differentially assigns important weights based on the main view area and the peripheral view area for each of the separated multiple object areas, divides each of the multiple object areas to which important weights are differentially assigned into multiple pixels to generate multiple pixelation data, and controls the server communication unit (21) to transmit the generated multiple pixelation data to the data processing server (20).

[0041] When the user control unit (17) receives a pixelation control signal from the data processing server (20) that includes a control command to divide the resolution of the object in the main viewing area to be at least 4 times higher than the resolution of the object in the peripheral viewing area, the user control unit (17) generates multiple pixelated data by dividing the resolution of the object in the main viewing area to be at least 4 times higher than the resolution of the object in the peripheral viewing area, and controls the server communication unit (21) to transmit the generated multiple pixelated data to the data processing server (20).

[0042] When the user control unit (17) receives a pixelation control signal from the data processing server (20) that includes meta-information containing data items and corresponding data values ​​for each of the separated multiple object areas, it controls each of the separated multiple object areas to include data items and corresponding data values, thereby dividing each of the multiple object areas into multiple pixels to generate multiple pixelation data, and controls the server communication unit (21) to transmit the generated multiple pixelation data to the data processing server (20).

[0043] When the user control unit (17) receives a pixelation control signal from the data processing server (20) that includes a control command to classify a plurality of separated object areas into one of text, a table, a button, an icon, and an image, it classifies the plurality of separated object areas into one of text, a table, a button, an icon, and an image, divides each of the classified plurality of object areas into a plurality of pixels to generate a plurality of pixelated data, and controls the server communication unit (21) to transmit the generated plurality of pixelated data to the data processing server (20).

[0044] The data processing server (20) includes a server communication unit (21), a server input unit (22), a server display unit (23), and an artificial intelligence server control unit (24).

[0045] The server communication unit (21) communicates with a plurality of user terminals (10). The server communication unit (21) can perform wireless communication, and the wireless communication includes at least one of infrared communication, RF, Zigbee, and Bluetooth. The server communication unit (21) receives a video signal and transmits it to the artificial intelligence server control unit (24) described later, and can be implemented in various ways corresponding to the specifications of the received video signal and the implementation form of the user terminal (10). For example, the server communication unit (21) can wirelessly receive an RF (radio frequency) signal transmitted from a broadcasting station (not shown), or receive a video signal according to composite video, component video, super video, SCART, HDMI (high definition multimedia interface) specifications, etc. via a wired connection. If the video signal is a broadcast signal, the server communication unit (21) may include a tuner that tunes the broadcast signal by channel.

[0046] The server input unit (22) is configured to input commands from an administrator. The server input unit (22) can receive touch input from an administrator or remote input from an administrator using a remote controller and transmit it to the corresponding artificial intelligence server control unit (24). Additionally, the server input unit (22) can receive voice input spoken by an administrator and transmit the voice signal to the artificial intelligence server control unit (24). In that case, the server input unit (22) can be implemented, for example, as a microphone. The server input unit (22) may also perform signal processing on the received voice signal itself. However, the form of administrator input that the server input unit (22) can receive is not limited to this, and, for example, administrator input through motion recognition, etc., can also be received.

[0047] The server display unit (23) displays an image based on an image signal processed by image processing. The implementation method of the server display unit (23) is not limited and can be implemented in various display methods such as liquid crystal, plasma, light-emitting diode, organic light-emitting diode, surface conduction electron-emitter, carbon nanotube, nano-crystal, etc.

[0048] The server display unit (23) may additionally include additional configurations depending on the implementation method. For example, if the server display unit (23) is a liquid crystal type, the server display unit (23) includes a liquid crystal display panel (not shown), a backlight unit (not shown) that supplies light thereto, and a panel driving board (not shown) that drives the panel (not shown). The server display unit (23) may display a voice recognition result as information regarding the recognized voice. Here, the voice recognition result can be displayed in various forms such as text, graphics, and icons, and the text includes characters and numbers. The server display unit (23) may further display candidate commands and application information based on the voice recognition result. The server can check whether the voice has been correctly recognized based on the voice recognition result displayed on the server display unit (12), and may select a command corresponding to the voice spoken by the user from among the displayed candidate commands by operating the server input unit (22) provided on the remote control, or select and check information related to the voice recognition result.

[0049] The artificial intelligence server control unit (24) generates a pixelation control signal based on a user command to collect and organize unstructured data from at least one of the multiple user terminals (10) input to the server input unit (21), and transmits the unstructured data to at least one user terminal (10) by dividing it into multiple pixels while displaying the unstructured data on the screen, and controls the server communication unit (21) to transmit the generated pixelation control signal to at least one user terminal (10), and generates screen data by connecting multiple pixelated data received from at least one user terminal (10).

[0050] The pixelation control signal may include a control command that distinguishes multiple object regions while unstructured data is displayed on the screen.

[0051] The pixelation control signal may include a control command that assigns important weights to each of the separated multiple object regions.

[0052] The pixelation control signal may include a control command that assigns important weights differentially based on the main field of view and the peripheral field of view.

[0053] The pixelation control signal may include a control command that divides the resolution of objects in the main field of view area to be at least four times higher than the resolution of objects in the peripheral field of view area.

[0054] The pixelation control signal may include meta-information containing data items and corresponding data values ​​for each of the separated multiple object regions.

[0055] Each pixel chunk can be created to embed meta-information describing itself. For example, when chunking, summary meta-information such as "This block is a table, contains time information, and has an importance of 0.8" can be included in the image.

[0056] Previously, the sequence of image - OCR - text extraction - interpretation was inefficient; however, by simultaneously including semantic labels, contextual tags, and inference weights during chunk generation, and since the AI ​​interprets while already knowing the meaning, it can result in a completely different information flow.

[0057] The pixelation control signal above may include a control command that classifies a plurality of separated object areas into one of text, a table, a button, an icon, and an image.

[0058] FIG. 2 is a flowchart of a first embodiment of an AI-based unstructured data processing method using an AI-based unstructured data processing system (1) according to the present invention.

[0059] A user command to collect and organize unstructured data from at least one of the multiple user terminals (10) is input to the server input unit (22) (S1).

[0060] A pixelation control signal is generated to divide and transmit unstructured data into multiple pixels from a data processing server (20) to at least one user terminal (10) while the data is displayed on the screen (S2).

[0061] The server communication unit (21) is controlled to transmit a pixelation control signal generated by the data processing server (20) to at least one user terminal (10) (S3).

[0062] A data processing server (20) generates screen data by connecting multiple pixelated data received from at least one user terminal (10) (S4).

[0063] The generated screen data is saved (S5).

[0064] FIG. 3 is a flowchart of a second embodiment of an AI-based unstructured data processing method using an AI-based unstructured data processing system (1).

[0065] A user command to collect and organize unstructured data from at least one of the multiple user terminals (10) is input to the server input unit (22) (S11).

[0066] A data processing server (20) generates and transmits a control command for distinguishing multiple object areas while displaying unstructured data on a screen to at least one user terminal (10), a control command for assigning important weights to each of the distinguished multiple object areas, a control command for assigning important weights differentially based on the main viewing area and the peripheral viewing area, a control command for dividing the resolution of objects in the main viewing area to be at least 4 times higher than the resolution of objects in the peripheral viewing area, meta information including data items and corresponding data values ​​for each of the distinguished multiple object areas, and a control command for classifying the distinguished multiple object areas into text, a table, a button, an icon, and an image (S12).

[0067] At least one user terminal (10) displays unstructured data on a screen (S13).

[0068] At least one user terminal (10) displays unstructured data on a screen and distinguishes multiple object areas (S14).

[0069] At least one user terminal (10) displays unstructured data on a screen and assigns important weights to each of the separated multiple object areas (S15).

[0070] At least one user terminal (10) displays unstructured data on a screen and assigns important weights based on the main field of view area and the peripheral field of view area (S16).

[0071] With unstructured data displayed on a screen on at least one user terminal (10), the resolution of the object in the main viewing area is divided to be at least 4 times higher than the resolution of the object in the peripheral viewing area (S17).

[0072] At least one user terminal (10) displays unstructured data on a screen and classifies a plurality of separated object areas into text, table, button, icon, and image (S18).

[0073] At least one user terminal (10) divides the unstructured data into multiple pixels to generate multiple pixelated data (S19).

[0074] At least one user terminal (10) transmits a plurality of pixelated data generated to a data processing server (S20).

[0075] A data processing server generates screen data by connecting multiple pixelated data received from at least one user terminal (10) (S21).

[0076] The generated screen data is saved (S22).

[0077] Here, data cleaning methods are explained.

[0078] Data preprocessing pipeline for AI analyzes files of various formats and different extensions based on pixelated bitmap images captured via screen capture, processes files of different extensions into a unified file such as PDF in the first stage, processes them into visualized images in the second stage, and then processes them into files such as json and md in the third stage.

[0079] Here, data types such as hwp, png, and composite PDF require preprocessing due to their low AI affinity, whereas txt, csv, json, and md can be utilized immediately due to their high AI affinity. Until now, humans have standardized data before feeding it into AI, but going forward, AI will be able to directly perform the interpretation, organization, semantic linking, and judgment of unstructured data.

[0080] Explain the sequence of data cleaning methods.

[0081] First, the resolution range is set to 1024X768 to 4K, and screen capture is performed at intervals of 0.5 to 2 seconds.

[0082] Second, it performs color, border, and shape-based segmentation, classifies objects such as text, icons, tables, and buttons, and conducts pixel-level analysis.

[0083] Third, the minimum unit size is set to 16x16 to 128x128 pixels, and semantic unit chunking is performed by assigning an object type, location coordinates, and importance score to each chunk.

[0084] Fourth, chunks with an importance score of 0.8 or higher are maintained at high resolution (original size), and the remaining chunks are processed at low resolution (reduced to 1 / 4) to perform focus area selection.

[0085] Fifth, input the Global View (full screen) and Local View (important chunks) in parallel, and perform context modeling by assigning positional relationships as coordinates and tokens during LLM input.

[0086] Sixth, determine task execution commands, including button clicks, text input, and menu selections.

[0087] Seventh, reflect the execution results back into the screen analysis and provide feedback on the action execution.

[0088] Modifiable embodiments other than the above embodiments are described.

[0089] The AI ​​server control unit can collect information from the network, including SNS, by utilizing log records from multiple user terminals. This information can be stored in a separate storage space, and information where hazardous elements are detected can be transmitted to a pre-configured administrator terminal.

[0090] Here, the risk factor may be situational information regarding the company's technology leakage, and may be stored in multiple stages depending on the degree of progress. The AI ​​server control unit may control the relevant user terminal to be monitored 24 hours a day when the degree of progress is determined to be at a dangerous level, and may restrict the external transmission and copying of data to prevent technology leakage.

[0091] After classifying the entire area of ​​unstructured data into areas where meaningful data exists and areas where meaningless data exists, the meaningless areas are masked, and the meaningful areas are converted into multiple pixelated data that are displayed on the screen and stored.

[0092] For example, unstructured data documents may consist only of areas with text and areas without text; in this case, only the text areas may be classified as meaningful regions, while the remaining areas may be classified as empty or meaningless regions. In such cases, the meaningless regions can be excluded from being converted into pixelated data, while only the meaningful regions are converted.

[0093] Each of the multiple pixelated data includes location information data of a pixel indicating its position on the screen and display data including the color and brightness of the pixel, and the artificial intelligence server control unit can generate screen data using the location information data and the display data.

[0094] Here, among multiple pixelated data, pixelated data having the same shape and color as the previous pixelated data may be configured to store only the location information data of the pixelated data identical to the previous pixelated data and the current location information data. In this case where there are only two location information data, the first location information data refers to the pixelated data identical to the previous pixelated data.

[0095] The AI ​​server control unit can minimize storage space by storing representative display data of the pixel as a labeling number, and then storing the location information data of the pixelated data along with the labeling number. The AI ​​server control unit can gradually increase the representative display data of the pixel to store up to a preset capacity, and by utilizing this, it can store a larger amount of unstructured data, thereby ensuring efficient storage and operational efficiency.

[0096] Due to the above-described AI-based unstructured data processing system (1) and AI-based unstructured data processing method, the AI ​​server control unit of the data processing server transmits a pixelation control signal to each of the multiple user terminals, and in response to the transmission, the multiple pixelated data received are connected to create and store screen data, so that the data is stored in a screen state regardless of the extension of the unstructured data, thereby making it possible to practically utilize unstructured data that could not be utilized, thus improving the storability and usability of unstructured data.

[0097] In addition, the efficiency of data storage can be improved by distinguishing between important and less important parts within unstructured data and pixelating them differently.

[0098] In addition, since the resolution of key and non-key parts can be stored differently, the data storage space can be significantly reduced.

[0099] In addition, since the amount of data in non-critical areas other than the most important areas within the screen area can be significantly reduced, data storage speed, utilization speed, and data search speed can be increased.

[0100] In addition, it is possible to minimize omissions or errors in data storage.

[0101] In addition, since the method of distinguishing objects can be simplified, the data processing speed can be increased. Explanation of the symbols

[0103] 1: AI-based unstructured data processing system 10: User terminal 11: User Communication Unit 12: User Display Unit 13: User Input Section 14: User Camera 15: User Speaker 16: User Microphone 17: User Control Unit 20: Data processing server 21: Server Communication Unit 22: Server Input Unit 23: Server Display Unit 24: Artificial Intelligence Server Control Unit

Claims

Claim 1 In an AI-based unstructured data processing system, a plurality of user terminals having programs corresponding to a plurality of extensions; The system includes a data processing server comprising: a server communication unit communicating with the plurality of user terminals; a server input unit for inputting a manager's command; and an artificial intelligence server control unit that generates a pixelation control signal to divide and transmit unstructured data into multiple pixels while displaying the unstructured data on a screen to at least one user terminal based on a user command that collects and organizes unstructured data from at least one of the plurality of user terminals input to the server input unit, controls the server communication unit to transmit the generated pixelation control signal to the at least one user terminal, and generates screen data by concatenating multiple pixelation data received from the at least one user terminal. The pixelation control signal includes a control command that distinguishes multiple object areas while the unstructured data is displayed on a screen. The artificial intelligence server control unit classifies the entire area of ​​the unstructured data into an area where meaningful data exists and a meaningless area, masks the meaningless area, converts the meaningful area into multiple pixelation data displayed on a screen and stores it, and each of the multiple pixelation data includes pixel location information data indicating the position on the screen and An AI-based unstructured data processing system characterized by including display data including the color and brightness of pixels, generating screen data using the location information data and the display data, and storing only the location information data of the pixelated data identical to the previous pixelated data and the current location information data for pixelated data having the same shape and color as the previous pixelated data among a plurality of pixelated data. Claim 2 delete Claim 3 An AI-based unstructured data processing system according to claim 1, wherein the pixelation control signal includes a control command that assigns important weights to each of the separated plurality of object regions. Claim 4 In claim 3, the AI-based unstructured data processing system is characterized in that the pixelation control signal includes a control command that differentially assigns important weights based on the main field of view area and the peripheral field of view area. Claim 5 An AI-based unstructured data processing system according to claim 4, wherein the pixelation control signal includes a control command that divides the resolution of an object in the main viewing area to be at least four times higher than the resolution of an object in the peripheral viewing area. Claim 6 In claim 5, the above pixelation control signal is characterized by including meta-information including data items and corresponding data values ​​for each of the separated plurality of object regions, in an AI-based unstructured data processing system. Claim 7 In claim 6, the above pixelation control signal is characterized by including a control command that classifies the separated plurality of object areas into one of text, table, button, icon, and image, in an AI-based unstructured data processing system.

Images