Electronic book authoring method using collaborative generative ai based on cloud environment, and apparatus therefor

Abstract

Disclosed are an electronic book authoring method using collaborative generative AI based on a cloud environment, and an apparatus therefor. An electronic book authoring apparatus according to an embodiment of the present invention comprises: an electronic book authoring processing unit that generates an electronic book by applying an electronic book authoring model generated through model training; and a model training processing unit that performs model training for generating the electronic book authoring model for initial model construction or model update for electronic book authoring.

Description

A method for authoring e-books using collaborative generative AI based on a cloud environment and a device therefor.

[0001] The present invention relates to an e-book authoring method using collaborative generative AI based on a cloud environment and a device therefor.

[0002] The material described in this section merely provides background information on embodiments of the present invention and does not constitute prior art.

[0003] An e-book is a digital book that records information, such as text or images, on an electronic medium, allowing it to be used like a book. An e-book is created by converting a previously published or potential published work into a digital format, storing it on an electronic recording medium or storage device, and then enabling it to be read or listened to on a computer, smartphone, tablet, or other mobile device via a wired or wireless information and communications network.

[0004] Compared to traditional paper books, e-books can efficiently store large amounts of data, making them highly portable and allowing for easy content updates. Furthermore, they utilize multimedia elements like audio, video, and animation to deliver more vivid and three-dimensional information to users. Thanks to these features, e-books are increasingly being utilized in a variety of fields, and the rapid growth of the digital publishing market is driving a steady increase in demand for e-book production.

[0005] However, the e-book production process is complex and requires a high level of technical expertise. Each stage of production, from text writing to image editing to layout composition, requires different technologies and tools, posing a significant barrier to entry, particularly for non-experts and small publishers. For this reason, e-book production is becoming increasingly time-consuming and costly, hindering access to the digital publishing market for small publishers and independent authors.

[0006] The main purpose of the present invention is to provide an e-book authoring method and a device therefor utilizing a collaborative generative AI based on a cloud environment, which performs model learning to create an e-book authoring model and applies the e-book authoring model created through model learning to create an e-book.

[0007] According to one aspect of the present invention, in a device for authoring an e-book using a generative AI based on a cloud environment to achieve the above purpose, the e-book authoring device may include an e-book authoring processing unit for creating an e-book by applying an e-book authoring model created through model learning; and a model learning processing unit for performing model learning to create the e-book authoring model for initial construction or model update of a model for e-book authoring.

[0008] In addition, according to another aspect of the present invention, in a method for authoring an e-book using a generative AI based on a cloud environment by an e-book authoring device for achieving the above purpose, the e-book authoring method may include an e-book authoring processing step of creating an e-book by applying an e-book authoring model created through model learning; and a model learning processing step of performing model learning to create the e-book authoring model for initial construction or model update for e-book authoring.

[0009] As described above, the present invention utilizes generative AI technology to automate content creation and layout optimization, thereby simplifying complex tasks occurring in the e-book production process and minimizing repetitive tasks.

[0010] Furthermore, the present invention can reduce the burden on creators and maximize efficiency by providing advanced features that were lacking in existing tools, thereby providing an environment in which even non-experts can easily create e-books, thereby expanding the diversity of the publishing market and significantly improving the accessibility of self-publishing.

[0011] Figure 1 is a block diagram schematically showing an e-book authoring system according to an embodiment of the present invention.

[0012] FIG. 2 is a block diagram schematically showing a model learning processing unit of an e-book authoring device according to an embodiment of the present invention.

[0013] FIG. 3 is a block diagram schematically showing an e-book authoring processing unit of an e-book authoring device according to an embodiment of the present invention.

[0014] Figures 4 and 5 are flowcharts for explaining an e-book authoring method according to an embodiment of the present invention.

[0015] FIG. 6 is a diagram schematically illustrating the engine configuration of a cloud environment and an e-book authoring device according to an embodiment of the present invention.

[0016] FIGS. 7 to 9 are diagrams for explaining the processing operation of a model learning processing unit according to an embodiment of the present invention.

[0017] FIG. 10 is a drawing for explaining a collaborative processing unit of an e-book authoring device according to an embodiment of the present invention.

[0018] FIG. 11 is a drawing showing a one-stop e-book service using an e-book authoring device according to an embodiment of the present invention.

[0019] Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the attached drawings. In describing the present invention, if a detailed description of a related known structure or function is judged to obscure the gist of the present invention, such detailed description will be omitted. Furthermore, although preferred embodiments of the present invention will be described below, it should be understood that the technical spirit of the present invention is not limited thereto and can be modified and implemented in various ways by those skilled in the art. Hereinafter, with reference to the drawings, a method for authoring an e-book using collaborative generative AI based on a cloud environment and a device therefor proposed in the present invention will be described in detail.

[0020] Figure 1 is a block diagram schematically showing an e-book authoring system according to an embodiment of the present invention.

[0021] The e-book authoring system (10) according to the present embodiment includes a user terminal (20), a cloud environment (30), and an e-book authoring device (40). The e-book authoring system (10) of FIG. 1 is according to one embodiment, and not all blocks illustrated in FIG. 1 are essential components, and some blocks included in the e-book authoring system (10) in other embodiments may be added, changed, or deleted.

[0022] The e-book authoring system (10) refers to a system that can produce e-books using an e-book authoring device (40) that utilizes generative AI based on a cloud environment.

[0023] Below, each component included in the e-book authoring system (10) will be described.

[0024] A user terminal (20) refers to a terminal carried by a user who wishes to create an e-book. Here, the user terminal (20) may be a communication-capable device such as a smartphone, pad, or PC.

[0025] The user terminal (20) is connected to a cloud environment (30) and inputs authoring request data through the cloud environment (30). Here, the authoring request data may include at least one request information related to an image, text, and layout input by the user for authoring an e-book.

[0026] A cloud environment (30) is a distributed IT infrastructure that provides computing resources (e.g., servers, storage, databases, software, etc.) via wireless communication, and can efficiently support all stages of the e-book authoring process.

[0027] The cloud environment (30) provides a foundation for effectively performing data storage, processing, access, and sharing.

[0028] Cloud environments (30) can provide data accessibility and storage space. Cloud environments (30) centrally store large amounts of data, including text, images, layouts, and multimedia files generated during the e-book authoring process, and support access to this data from any Internet-connected device. This allows users to continue working regardless of location or device, and even enables real-time collaboration.

[0029] The cloud environment (30) can provide integration with generative AI by linking with an e-book authoring device (40). Here, generative AI technology can also be integrated and operated within the cloud environment (30). AI models deployed in the cloud can process various tasks in real time, such as text generation, image creation and editing, and layout automation, allowing users to utilize AI technology without the need for separate high-performance equipment or complex setup.

[0030] The cloud-based e-book authoring environment (30) provides collaborative features that allow multiple users to participate in a project simultaneously. Multiple users, such as text writers, designers, and editors, can work simultaneously, sharing and reflecting changes in real time.

[0031] The e-book authoring device (40) performs model learning to create the e-book authoring model for initial construction or model update of an e-book authoring model, and applies the e-book authoring model created through model learning to create an e-book.

[0032] The e-book authoring device (40) according to the present embodiment includes a model learning processing unit (42), an e-book authoring processing unit (44), and a collaboration processing unit (46).

[0033] The model learning processing unit (42) performs model learning to create an e-book authoring model for initial model construction or model update for e-book authoring.

[0034] The model learning processing unit (42) collects a dataset for model learning. The model learning processing unit (42) refines the collected data and converts it into a dimensional vector suitable for generative AI learning through a preset preprocessing algorithm. In addition, the model learning processing unit (42) generates images, layouts, and templates necessary for producing an e-book using generative AI based on the dimensional vectors. The model learning processing unit (42) learns the sample structure and sample layout of the e-book from the generated results and uses the generated e-book prototype to create an e-book authoring model.

[0035] A detailed description of the model learning processing unit (42) is described in Fig. 2.

[0036] The e-book authoring processing unit (44) creates an e-book by applying an e-book authoring model (45) generated through model learning.

[0037] The e-book authoring processing unit (44) obtains authoring request data input through a cloud environment and generates an e-book using the authoring request data as input through an e-book authoring model (45). The e-book authoring processing unit (44) provides the e-book authoring results for the generated e-book to an external device. Here, the external device may be at least one terminal connected to the cloud environment (30).

[0038] A detailed description of the e-book authoring processing unit (44) is described in Fig. 3.

[0039] The collaboration processing unit (46) obtains collaboration request data based on the cloud environment (30) and transmits collaboration information generated based on the collaboration request data to at least one of the model learning processing unit (42) and the e-book authoring processing unit (44).

[0040] The collaboration processing unit (46) acquires collaboration request data input through a cloud environment in real time or at preset intervals while creating an e-book through an e-book authoring model or performing model learning to create an e-book authoring model.

[0041] The collaboration processing unit (46) analyzes the collaboration request data and transmits the collaboration information corresponding to all or part of the collaboration request data to the e-book authoring processing unit (44) that creates the e-book or the model learning processing unit (42) that performs model learning so that it can be applied.

[0042] FIG. 2 is a block diagram schematically showing a model learning processing unit of an e-book authoring device according to an embodiment of the present invention.

[0043] The model learning processing unit (42) of the e-book authoring device (40) according to the present embodiment includes a data collection unit (210), a data preprocessing unit (220), and an e-book authoring model generation unit (230). The model learning processing unit (42) of FIG. 2 is according to one embodiment, and not all blocks illustrated in FIG. 2 are essential components, and some blocks included in the model learning processing unit (42) in other embodiments may be added, changed, or deleted.

[0044] The model learning processing unit (42) performs model learning to create an e-book authoring model for initial model construction or model update for e-book authoring.

[0045] Below, each component included in the model learning processing unit (42) will be described.

[0046] The data collection unit (210) collects a dataset for model learning. Here, the dataset may include various data related to e-books, such as images, text, and layouts.

[0047] The data collection unit (210) collects datasets using Creative Commons licenses or public domain content. Here, the data collection unit (210) can collect datasets by adding customized data through user surveys or expert collaboration.

[0048] Additionally, the data collection unit (210) can collect a data set by applying a data augmentation technique for at least one of changing the sentence structure of text data, replacing words, rotating images, adjusting the image size, and converting colors.

[0049] The data preprocessing unit (220) refines the collected data and converts it into a dimensional vector suitable for generative AI learning through a preset preprocessing algorithm.

[0050] The data preprocessing unit (220) analyzes the tags and components of the collected data set, extracts preset features for creating an e-book based on the analysis results, and preprocesses the refined data. For example, the data preprocessing unit (220) can analyze HTML DOM tags and CSS components and extract key features necessary for creating an e-book. Here, the data preprocessing unit (220) can remove unnecessary data except for key features and convert the refined data into a form suitable for generative AI learning.

[0051] The data preprocessing unit (220) performs preprocessing by searching for the optimal number of dimensions for application to generative AI through a preprocessing algorithm and converting the refined data into a dimensional vector corresponding to the searched number of dimensions. The data preprocessing unit (220) can perform preprocessing by applying an algorithm for converting into a fixed dimensional vector and an algorithm for converting into a dimensional vector corresponding to the number of dimensions by searching for the optimal number of dimensions so as to maximize the learning performance of the e-book production system.

[0052] The e-book authoring model generation unit (230) uses generative AI based on dimensional vectors to generate images, layouts, and templates necessary for producing e-books. The e-book authoring model generation unit (230) learns the sample structure and sample layout of e-books from the generated results, and uses the generated e-book prototype to generate an e-book authoring model.

[0053] The e-book authoring model generation unit (230) transmits the generated e-book authoring model to the e-book authoring processing unit (44) so ​​that it is stored.

[0054] The e-book authoring model generation unit (230) includes a probabilistic denoising model optimized for the structural characteristics of e-books. The e-book authoring model generation unit (230) generates an e-book layout based on web standards through the probabilistic denoising model, and generates an e-book prototype that automatically adjusts at least one design element among image size, sentence structure, and font size according to user input within the generated layout.

[0055] The e-book authoring model generation unit (230) can generate high-quality images and promote harmony between text and design elements through a probabilistic denoising model. For example, the e-book authoring model generation unit (230) can generate an e-book layout based on web standards such as HTML, CSS, and JavaScript, and can automatically adjust design elements such as image size, sentence structure, and font size within the generated layout based on user input.

[0056] Meanwhile, the e-book authoring model generation unit (230) converts the structural characteristics of the e-book into a graph-based data structure and processes it to be suitable for generative AI. For example, the e-book authoring model generation unit (230) analyzes the hierarchical structure of HTML, CSS, and JavaScript through a graph neural network-based generative model and converts it into graph data, and expresses the connectivity and hierarchical relationship between components as nodes and edges of the graph. In addition, the e-book authoring model generation unit (230) can convert e-book data reconstructed in the form of a DOM Tree into graph data through a graph neural network-based generative model.

[0057] Meanwhile, the e-book authoring model generation unit (230) learns the structural characteristics of e-book images, layouts, and templates based on a preset graph neural network, and generates new design elements based on the learning results. Here, the algorithm for generating new design elements is designed based on a diffusion process, and generates an e-book prototype with new design elements added by combining a graph transformer and a DDPM (Diffusion Denoising Probability Model), thereby generating an e-book authoring model. The e-book authoring model generation unit (230) can improve the visual quality of e-books through the algorithm for generating new design elements, and can automatically generate various structures and designs requested by users.

[0058] FIG. 3 is a block diagram schematically showing an e-book authoring processing unit of an e-book authoring device according to an embodiment of the present invention.

[0059] The e-book authoring processing unit (44) of the e-book authoring device (40) according to the present embodiment includes an authoring request data acquisition unit (310), an e-book generation unit (320), and an e-book authoring result providing unit (330). The e-book authoring processing unit (44) of FIG. 3 is according to one embodiment, and not all blocks illustrated in FIG. 3 are essential components, and some blocks included in the e-book authoring processing unit (44) in other embodiments may be added, changed, or deleted.

[0060] The e-book authoring processing unit (44) creates an e-book by applying an e-book authoring model generated through model learning.

[0061] Below, each component included in the e-book authoring processing unit (44) will be described.

[0062] The copyright request data acquisition unit (310) acquires copyright request data input through a cloud environment (30).

[0063] The authoring request data acquisition unit (310) acquires authoring request data including at least one request information related to an image, text, and layout input by a user for authoring an e-book in a cloud environment.

[0064] The e-book generation unit (320) generates an e-book using the e-book creation model (45) as input for creation request data. Here, the e-book creation model (45) is a model generated from the model learning processing unit (42).

[0065] The e-book generation unit (320) generates images, text, and layout corresponding to the authoring request data through the e-book authoring model (45), and performs optimization processing to integrate all of the images, text, and layout to generate the e-book.

[0066] The e-book authoring result providing unit (330) provides the e-book authoring result for the generated e-book to an external device. Here, the external device may be at least one terminal connected to a cloud environment (30).

[0067] The e-book authoring result providing unit (330) can provide an e-book authoring result including e-book data for the created e-book, e-book creation history data, and user data that requested authoring.

[0068] Figures 4 and 5 are flowcharts for explaining an e-book authoring method according to an embodiment of the present invention.

[0069] Referring to Figure 4, the model learning processing unit (42) collects training data required for e-book authoring (S410). The collected data includes various data related to e-books, such as images, text, and layouts. It can utilize Creative Commons licenses (CCL) or public domain content, or additionally collect customized data. Furthermore, data augmentation techniques (e.g., sentence structure modification, image rotation, and resizing) are applied to enhance the diversity and quality of the data.

[0070] The model learning processing unit (42) refines the collected data and converts it into dimensional vectors suitable for generative AI learning through a preset preprocessing algorithm (S420). In the preprocessing stage, data tags and components are analyzed to extract key characteristics necessary for e-book production, and unnecessary data is removed.

[0071] The model learning processing unit (42) uses generative AI based on preprocessed data to generate images, layouts, and templates necessary for e-book production. During this process, a probabilistic denoising model (DDPM) and a graph neural network are utilized to learn the structural characteristics of the e-book. The generated e-book prototype automatically adjusts design elements such as image size, sentence structure, and font size to suit user needs, ultimately generating an e-book authoring model (S430).

[0072] The model learning processing unit (42) transfers the generated e-book authoring model to the e-book authoring processing unit (44) for storage. The generated e-book authoring model is directly utilized in e-book production and enhances the visual quality of e-book content by including high-quality images, text, and layouts. The e-book authoring model includes an algorithm that can automatically generate various structures and design elements requested by the user, thereby simultaneously enhancing the efficiency and quality of e-book production.

[0073] Referring to Figure 5, the e-book authoring processing unit (44) obtains authoring request data entered by the user through a cloud environment (S510). The authoring request data includes information related to images, text, layout, and other information for e-book authoring. The user inputs this data in the cloud environment to submit an e-book production request, and the authoring request data acquisition unit collects this data and utilizes it for subsequent processing.

[0074] The e-book authoring processing unit (44) generates an e-book using a pre-generated e-book authoring model based on the acquired authoring request data (S520). The e-book authoring processing unit (44) generates content using the e-book authoring model based on the image, text, and layout information contained in the authoring request data, and completes the e-book by integrating the images, text, and layout into an optimized form. During this process, the user can generate a high-quality e-book based on the input data.

[0075] The e-book authoring processing unit (44) transmits the e-book authoring results for the generated e-book to an external device (S530). The external device may be a terminal connected to a cloud environment (e.g., a user's PC, tablet, smartphone, etc.). The e-book authoring processing unit (44) provides the results, including not only the generated e-book data but also e-book creation history data and authoring request user data. This allows the user to view not only the completed e-book but also information related to the production process.

[0076] Although FIGS. 4 and 5 each describe the steps as being executed sequentially, this is not necessarily the case. In other words, the steps described in FIGS. 4 and 5 may be modified and executed, or one or more steps may be executed in parallel. Therefore, FIGS. 4 and 5 are not limited to a chronological order.

[0077] The e-book authoring method according to the present embodiment described in FIGS. 4 and 5 may be implemented as an application (or program) and recorded on a recording medium readable by a terminal device (or computer). The recording medium on which the application (or program) for implementing the e-book authoring method according to the present embodiment is recorded and readable by a terminal device (or computer) includes all types of recording devices or media that store data that can be read by a computing system.

[0078] FIG. 6 is a diagram schematically illustrating the engine configuration of a cloud environment and an e-book authoring device according to an embodiment of the present invention.

[0079] Referring to (a) of Fig. 6, the authoring tool and engine configuration of the e-book authoring device (40) can be composed of a platform, a foundation, a core, and a controller.

[0080] A platform can include an OS, kernel, native platform, etc. The platform is responsible for communicating with the device's basic operating system and hardware, and serves as the foundation for all higher layers.

[0081] Foundation can include Base Class, XML Parser, Font Graphic, Memory, Image, G-AI, etc.

[0082] Foundation supports basic functions related to data processing, graphics rendering, and AI, and can provide XML parsing, font rendering, memory management, image processing, and generative AI functions.

[0083] Core can include Editor Importer / Exporter, Rendering Engine, Generative AI, etc.

[0084] Editor Importer / Exporter provides the ability to import or export files such as ePUB, Office documents (e.g. docx, xlsx, pptx, etc.).

[0085] The Rendering Engine renders HTML, ePUB, Office documents, images, PDFs, etc. and displays them on the screen.

[0086] Generative AI uses generative AI to automatically create and edit text, images, and other content.

[0087] A Controller can contain an Event Manager, Task Manager, Function Executor, Callback, Browser (Flutter UI), etc.

[0088] Event Manager, Task Manager, Function Executor, and Callback manage the interaction between the user interface and the engine, and provide a flexible user experience through event handling, task management, function execution, and callbacks.

[0089] Browser (Flutter UI) provides an intuitive and responsive interface by leveraging browser-based UI (Flutter technology).

[0090] Referring to (b) of Fig. 6, the cloud service module for the cloud environment is linked with the e-book authoring device and is composed of a Contents Publishing Layer, a Hosting Management Layer, a Contents Management Layer, and a Contents Archiving Layer.

[0091] In the Content Publishing Layer, Static (X)HTML and Dynamic (X)HTML generate and distribute static and dynamic HTML content, while Video / Audio Based on HTML5 manages HTML5-based video and audio content. Additionally, Mobile Page and Other Format support mobile pages and other formats of content.

[0092] In the Hosting Management Layer, Domain N performs content hosting and management in a multi-domain environment.

[0093] In the Contents Management Layer, (X)HTML Authoring, Workflow, Widget & Template support content authoring, workflow management, and tasks utilizing widgets and templates, while Media Transcoding and Search Engine provide media file conversion and search engine functions. Additionally, User Management and J2EE Application Server operate as user management and Java-based application servers.

[0094] In the Contents Archiving Layer, the Database and File System provide stable storage of content and link the database and file system.

[0095] In Figure 6, the e-book authoring device (authoring tool and engine) and the cloud service module interact via an API, enhancing the efficiency of data processing and content management. This allows for active utilization of the cloud's scalability and data collaboration capabilities during the e-book production process.

[0096] FIGS. 7 to 9 are diagrams for explaining the processing operation of a model learning processing unit according to an embodiment of the present invention.

[0097] Figure 7 shows a process of processing document data composed of HTML, CSS, and JavaScript in an e-book authoring device (40) and analyzing and converting it through graph-based representation.

[0098] Figure 7 (a) shows HTML, CSS, and JavaScript-based document processing and graph generation.

[0099] Referring to (a) of Fig. 7, document data composed of HTML, CSS, and JavaScript is processed hierarchically. Each node of the HTML document (e.g., , , , , (etc.) represents a hierarchical relationship, and the structure of a document is identified by analyzing the parent-child relationship and the connection structure between nodes. The attributes and data of each node are extracted as features of the node, and these features are converted into vector format.

[0100] The extracted vector data is converted into a graph-based representation. During this process, each element within the HTML document is represented as a node in the graph, and relationships between nodes are defined as edges. This graph represents the overall hierarchical structure of the document and serves as the basis for data used in the e-book production process. The graph includes the characteristics of all nodes and serves as input data for graph learning and transformation.

[0101] Figure 7 (b) shows graph representation learning and optimization.

[0102] Referring to Figure 7 (b), the graph-based representation extracts and optimizes the information necessary for e-book creation through a learning process. This learning process utilizes a Transformer architecture and a multi-head attention mechanism. Multi-head attention calculates the relationships and importance between nodes based on the query, key, and value vectors of each node, and the learned results are used to optimize the document's graph representation.

[0103] During this process, the Transformer structure normalizes the features of each node in the graph and computes the final representation of the node through a Feed-Forward Neural Network (FFN). Through this learning process, the graph is transformed into a form that comprehensively understands the document's structure and contextual information. This allows for the automatic design of content required for e-book production by optimizing text and graphic elements, producing visually enhanced results.

[0104] Figure 8 is a diagram illustrating the operating principle of a diffusion-based probabilistic denoising model (DDPM).

[0105] Referring to Figure 8, the two main processes of DDPM, the Forward Process and the Reverse Process, can be explained.

[0106] The Forward Process, which is the step of adding noise, starts from the original image (x0) and gradually adds noise. This process proceeds by using a probability distribution q(x t |x t-1 ) and noise is gradually added at each step. By repeating this process, the final image is completely noised (x t ) are generated. This process allows the model to learn from data at different noise levels.

[0107] Conversely, the Reverse Process returns a completely noised image state (x) from the Forward Process. t ) to restore the original image (x0). This process is performed by gradually removing noise from the learned model using the probability distribution p. θ (x t-1 |x t ) to remove noise, and at each step the image is restored to a form that is closer and closer to the original. Ultimately, a clean image can be generated, or a new image can be generated under new conditions.

[0108] Figure 9 shows the operation process of a probabilistic denoising model (DDPM: Denoising Diffusion Probabilistic Model) based on a diffusion process for image generation.

[0109] The probabilistic denoising model can be utilized to generate visual content based on text in an e-book authoring device (40).

[0110] A probabilistic denoising model is a model that probabilistically processes the process of adding and removing noise to create or restore images, which can be used to create high-quality images or produce new content.

[0111] Figure 9 illustrates a text-to-image generation process, which can be utilized to generate visual content based on text input into an e-book authoring device (40). This process includes a series of steps that gradually convert text input by a user into an image using a trained model.

[0112] First, the user inputs text, such as "Milky way of the space." This text defines the subject of the image to be generated and is used to generate images based on specific sentences or keywords during the e-book production process. The input text is tokenized and converted into a sequence of numeric tokens. For example, the text "Milky way of the space" is converted into a numeric sequence such as {33, 38, 71, 814, 1321}, which is a format the model can process.

[0113] The numeric tokens generated by the tokenizer are then converted into high-dimensional vectors through an embedding process. Each token has a unique vector representation, and these vectors numerically represent the meaning of the text. The Text Transformer learns these vectors and analyzes the contextual meaning and patterns of the input text. This process transforms the implicit meaning of the text into information necessary for image generation, such as expressions like "Milky Way" being converted into visual elements depicting the Milky Way.

[0114] The image generation process begins with an initial state consisting of random noise. This noise sample is gradually refined by a trained model and transformed into an image that visually represents the meaning of the text. The initial noisy sample is gradually enriched with details and structure, forming an image that matches the text. Ultimately, a high-quality image generated based on the input text is output. For example, the text "Milky way of the space" is transformed into a visual image depicting the Milky Way.

[0115] The diffusion process-based probabilistic denoising model can be utilized in various ways in the e-book production process. It can automatically generate illustrations based on specific phrases or paragraphs to enhance the visual appeal of e-books. It can also create customized illustrations based on reader interests or requests. This allows for real-time image generation without the need for a separate designer or image search, reducing production time. Furthermore, it can enhance the visual appeal of e-books by providing images in various styles (e.g., realistic, cartoonish, and abstract) based on text.

[0116] FIG. 10 is a drawing for explaining a collaborative processing unit of an e-book authoring device according to an embodiment of the present invention.

[0117] Figure 10 shows the structure of the collaborative processing unit (46) of the e-book authoring device (40).

[0118] The collaborative processing unit (46) is divided into a user interface (UI) layer (1010) and an engine layer (1020), and each layer divides its role in the e-book authoring process to maximize work efficiency.

[0119] The Collaborate UI Framework manages user interaction and provides various visual interfaces and functions required for e-book production. The UI Framework of the present invention includes the following components. The Annotation View provides a function to display annotation content on the screen, allowing users to add, modify, and delete annotation data. The Annotation Manager centrally manages annotation data and efficiently processes the creation, deletion, and editing of annotation-related tasks. The Document View visually displays the contents of an e-book document and provides an environment for users to view and edit the document. The History View displays a work history, allowing users to check the history of their work and revert to specific work steps.

[0120] The Collaborate Engine Framework processes and stores data generated during the e-book production process and is responsible for the core logic of tasks performed in the UI layer. The Engine Framework of the present invention includes the following components: The Annotation Data Encoder / Decoder encodes and decodes annotation data, converts it for storage or communication, and thereby maintains data consistency and compatibility. The Office Engine is the core engine for editing and processing various elements of e-book documents, such as text, images, and layout. It supports multiple formats, including ePUB, PDF, and HTML, enabling the creation and management of various e-book formats.

[0121] For example, the collaborative processing structure of the collaborative processing unit (46) can operate in the following manner. The user creates or edits e-book content through the UI Framework, and the annotation data created through the Annotation Manager is encoded and stored in the Annotation Data Encoder / Decoder, and decoded and displayed on the screen when necessary. The entire document can be checked and worked on in real time through the Document View, and the work history can be checked and reverted to a specific point in time through the History View. The Office Engine plays a central role in document editing and processing, and supports conversion and editing of complex data and file formats.

[0122] FIG. 11 is a drawing showing a one-stop e-book service using an e-book authoring device according to an embodiment of the present invention.

[0123] Figure 11 details the e-book authoring, distribution, and utilization ecosystem, depicting the entire flow from content creation to distribution and utilization, centered around e-book authoring devices and collaborative AI technology. E-book services are designed for individual, institutional, and business users.

[0124] Individual users include writers, professors, and general users, and can create their own content using e-book authoring tools (40).

[0125] The e-book authoring tool (40) simplifies the complex e-book production process, maximizes efficiency, and enhances user convenience. Institutional and business users include public institutions, public publishers, and private companies. It converts existing PDF or audiobook content to the ePUB 3.0 format, enabling efficient internal and external publishing.

[0126] The e-book authoring tool (40) plays a key role in the e-book authoring process. Based on ePUB 3.0, it supports HTML5 and ePUB standards, ensuring compatibility across multiple platforms. Collaboration technologies include video conferencing for real-time collaboration between teams, a whiteboard for sharing ideas and designs, and an authoring history feature for tracking edits and changes. Generative AI technology automatically generates content, including text, clip art, images, and layouts, by learning from existing data, and provides the ability to automate content using templates and designs.

[0127] The ePUB 3.0-based e-book service supports content upload, distribution, and sales, including e-book registration, ISBN issuance, sales, and settlement management. Users can purchase or subscribe to e-books and receive personalized learning experiences through recommended learning models. Furthermore, it analyzes purchasing patterns and user interests to provide even more customized services.

[0128] The above description is merely an example of the technical idea of ​​the embodiment of the present invention, and those skilled in the art to which the embodiment of the present invention pertains can make various modifications and variations without departing from the essential characteristics of the embodiment of the present invention. Therefore, the embodiment of the present invention is not intended to limit the technical idea of ​​the embodiment of the present invention, but to explain it, and the scope of the technical idea of ​​the embodiment of the present invention is not limited by these embodiment. The protection scope of the embodiment of the present invention should be interpreted by the following claims, and all technical ideas within a scope equivalent thereto should be interpreted as being included in the scope of the rights of the embodiment of the present invention.

[0129] <Explanation of symbols>

[0130] 10: E-book authoring system

[0131] 20: User terminal

[0132] 30: Cloud environment

[0133] 40: E-book authoring device

[0134] 42: Model learning processing unit

[0135] 44: E-book authoring processing unit

[0136] 46: Collaborative Processing Unit

Claims

1. In a device that creates e-books using generative AI based on a cloud environment, An e-book authoring processing unit that creates an e-book by applying an e-book authoring model generated through model learning; and A model learning processing unit that performs model learning to create the e-book authoring model for initial model construction or model update for e-book authoring. An e-book authoring device characterized by including:

2. In paragraph 1, The above e-book authoring processing unit, A copyright request data acquisition unit that acquires copyright request data entered through a cloud environment; An e-book generation unit that generates the e-book through the e-book creation model by inputting the above creation request data; and An e-book authoring result providing unit that provides the e-book authoring result for the e-book generated above to an external device. An e-book authoring device characterized by including:

3. In paragraph 2, The above copyright request data acquisition unit, An e-book authoring device characterized in that it obtains authoring request data including at least one request information related to an image, text, and layout input by a user for authoring the e-book in the cloud environment.

4. In paragraph 2, The above e-book generation unit, An e-book authoring device characterized in that it generates an image, text, and layout corresponding to the authoring request data through the e-book authoring model, and performs optimization processing to integrate all of the image, the text, and the layout to generate the e-book.

5. In paragraph 2, The above e-book authoring result provider is, An e-book authoring device characterized in that it provides the e-book authoring result including e-book data for the generated e-book, e-book creation history data, and user data that requested authoring.

6. In paragraph 1, The above model learning processing unit, Data collection unit that collects datasets for model learning; A data preprocessing unit that refines the collected data and converts it into a dimensional vector suitable for generative AI learning through a preset preprocessing algorithm; and An e-book authoring model generation unit that generates images, layouts, and templates necessary for producing the e-book using the generative AI based on the above-mentioned dimensional vector, and generates the e-book authoring model using the e-book prototype generated by learning the sample structure and sample layout of the e-book based on the generated result. An e-book authoring device characterized by including:

7. In paragraph 6, The above data collection unit, Collect the above datasets using Creative Commons License or public domain content, An e-book authoring device characterized in that the dataset is collected by adding customized data through user surveys or expert collaboration.

8. In paragraph 6, The above data collection unit, An e-book authoring device characterized in that the data set is collected by applying a data augmentation technique for at least one of changing the sentence structure of text data, replacing words, rotating images, resizing images, and converting colors.

9. In paragraph 6, The above data preprocessing unit, An e-book authoring device characterized by analyzing tags and components of the collected data set, extracting preset features for producing the e-book based on the analysis results, and preprocessing the refined data.

10. In paragraph 9, The above data preprocessing unit, An electronic book authoring device characterized in that the preprocessing is performed to search for an optimal number of dimensions to be applied to the generative AI through the preprocessing algorithm and to convert the refined data into the dimension vector corresponding to the searched number of dimensions.

11. In paragraph 6, The above e-book authoring model generation unit, Includes a probabilistic denoising model optimized for the structural features of the above e-book, An e-book authoring device characterized in that it generates an e-book layout based on web standards through the probabilistic denoising model, and generates an e-book prototype that automatically adjusts at least one design element among image size, sentence structure, and font size according to a user's input within the generated layout.

12. In paragraph 6, The above e-book authoring model generation unit, An e-book authoring device characterized in that it converts the structural features of the e-book into a graph-based data structure and processes it to be suitable for the generative AI.

13. In paragraph 6, The above e-book authoring model generation unit, Based on a pre-configured graph neural network, it learns the structural characteristics of e-book images, layouts, and templates, and generates new design elements based on the learning results. An e-book authoring device characterized in that the algorithm for generating new design elements is designed based on a diffusion process, and generates the e-book prototype with new design elements added by combining a graph transformer and a DDPM (Diffusion Denoising Probability Model) to generate the e-book authoring model.

14. In paragraph 1, A collaboration processing unit that acquires collaboration request data input through a cloud environment in real time or at preset intervals while generating the e-book through the e-book authoring model or performing model learning to generate the e-book authoring model, analyzes the collaboration request data, and transmits collaboration information corresponding to all or part of the collaboration request data so that it can be applied to generating the e-book or learning the model. An e-book authoring device characterized by additionally including:

15. In a method for authoring an e-book using a cloud-based generative AI, An e-book authoring processing step for creating an e-book by applying an e-book authoring model generated through model learning; and A model learning processing step for performing model learning to create the e-book authoring model for initial model building or model update for e-book authoring. An e-book authoring method characterized by including:

Images