Generate output documents through a machine learning model based on document templates

Abstract

A computing device for generating an output document comprises one or more memories for storing instructions and one or more processors for executing instructions to perform operations, wherein the operations include: an operation of receiving a plurality of source documents; an operation of receiving an input associated with a request to generate an output document based on a plurality of source documents and a specific document template; and an operation of generating an output document having a specific document template through one or more machine-learned models.

Description

Technology Field

[0001] The present disclosure relates to generating content through one or more machine learning models based on source content generally selected (identified) by a user. For example, the present disclosure relates to a method and a computing device for generating content (output document) by implementing a document template generated by one or more machine learning models for source content, thereby assisting a user in managing, organizing, and creating content. Background Technology

[0002] According to current computing systems, large-scale language models (LLMs) can interact with text content. For example, a user can query an LLM about content by copying the content of a document and pasting it into a chat box. The LLM can provide output about the content (e.g., a summary).

[0003] Aspects and advantages of the embodiments of the present disclosure may be partially presented in the following description, learned from the description, or learned through the practice of exemplary embodiments.

[0004] In one or more exemplary embodiments, a computing device for creating, organizing, managing, and writing content is provided. For example, the computing device includes one or more memories configured to store instructions; and one or more processors configured to execute instructions to perform operations, wherein the operations include receiving a plurality of source documents, receiving an input associated with a request to generate an output document based on a plurality of source documents and a specific document template, and generating an output document having a specific document template through one or more machine-learned models.

[0005] In some implementations, the operations further include providing a graphical user interface configured to receive a selection of multiple source documents to be displayed on a display device.

[0006] In some implementations, the actions further include an action that determines the intent associated with the output document based on the content associated with each of the multiple source documents by one or more machine-learned models.

[0007] In some implementations, the operations further include operations that determine the style associated with the output document based on the content of each of the multiple source documents by one or more machine-learned models.

[0008] In some implementations, the operations further include determining the format associated with the output document based on the content of each of the multiple source documents by one or more machine-learned models.

[0009] In some implementations, the operation of determining a format associated with an output document based on the content of each of a plurality of source documents by one or more machine-learned models includes the operation of identifying a plurality of headings for individual sections of the output document.

[0010] In some implementations, the operations include an operation for displaying a graphical user interface comprising a plurality of user interface elements on a display device, wherein each of the plurality of user interface elements corresponds to a different style that can be applied by one or more machine-learned models to generate an output document having a specific document template; and further include an operation for receiving a selection for one or more of the plurality of user interface elements, wherein the operation for generating an output document having a specific document template through one or more machine-learned models includes an operation for applying one or more styles corresponding to the selection for one or more of the plurality of user interface elements.

[0011] In some implementations, the operations include an operation for displaying a graphical user interface comprising a plurality of user interface elements on a display device, wherein each of the plurality of user interface elements corresponds to a different heading that can be applied by one or more machine-learned models to generate a section of an output document having a specific document template; and further include an operation for receiving a selection for one or more of the plurality of user interface elements, wherein the operation for generating an output document having a specific document template through one or more machine-learned models includes an operation for applying one or more headings corresponding to the selection for one or more of the plurality of user interface elements to a section of the output document.

[0012] In some implementations, the computing device further includes one or more databases configured to store a plurality of generative machine learning models associated with a plurality of different document types, and the operations further include an operation to search among the plurality of generative machine learning models for at least one generative machine learning model associated with a document type represented by the common content of a plurality of source documents.

[0013] In some implementations, the operations further include receiving a plurality of training documents associated with a document type—each of the plurality of training documents contains one or more common features associated with a specific document template—; and training one or more machine learning models to learn at least one of the intent, style, or format of a specific document template for a document type.

[0014] In one or more exemplary embodiments, a computer-implemented method for organizing, managing, and creating content is provided. The computer-implemented method comprises the steps of: receiving a plurality of source documents by a computing system comprising one or more processors and one or more machine-learned models; receiving an input associated with a request to generate an output document based on the plurality of source documents and a specific document template by the computing system; and generating an output document having a specific document template through one or more machine-learned models.

[0015] In some implementations, the method further includes the step of determining at least one of a style, intent, and format associated with an output document based on the content of each of a plurality of source documents by one or more machine-learned models.

[0016] In some implementations, the method comprises the step of providing a graphical user interface including a plurality of user interface elements for display on a display device of a computing system, wherein each of the plurality of user interface elements corresponds to a style, intent, or format that can be applied by one or more machine-learned models to generate an output document having a specific document template; and further comprises the step of receiving a selection for one or more of the plurality of user interface elements, wherein the step of generating an output document having a specific document template through one or more machine-learned models includes the step of applying a style, intent, or format corresponding to the selection for one or more of the plurality of user interface elements.

[0017] In some implementations, the method further comprises: storing a plurality of generative machine learning models associated with a plurality of different document types, respectively, in one or more databases; and searching among the plurality of generative machine learning models for at least one generative machine learning model associated with a document type represented by the common content of a plurality of source documents.

[0018] In some implementations, the method further comprises the step of receiving a plurality of training documents associated with a document type, wherein each of the plurality of training documents includes one or more common features associated with a specific document template; and the step of training one or more machine learning models to learn at least one of the intent, style, or format of a specific document template for a document type.

[0019] In one or more exemplary embodiments, a computing device for creating, organizing, managing, and writing content is provided. For example, the computing device includes one or more memories configured to store instructions; and one or more processors configured to execute instructions to perform operations, wherein the operations include receiving a selection of a plurality of sample documents; receiving input for generating a document template based on the selection of a plurality of sample documents; and in response to receiving the input, implementing one or more machine learning models to generate a document template based on at least one of a style, format, or intent associated with a plurality of sample documents.

[0020] In some implementations, the operations further include receiving multiple source documents, receiving additional inputs associated with a request to generate an output document based on multiple source documents and a document template, and generating an output document according to the document template through one or more machine-learned models.

[0021] In some implementations, the operations further include an operation that determines at least one of a style, format, and intent associated with a plurality of sample documents based on the content and structure of each of a plurality of sample documents by one or more machine-learned models.

[0022] In some implementations, the operations include: an operation of providing a graphical user interface comprising a plurality of user interface elements for display on a display device, wherein at least one of the plurality of user interface elements is selectable to change at least one of a style, format, and intent associated with a plurality of sample documents determined by one or more machine learning models; and further including an operation of receiving a selection for one or more of the plurality of user interface elements to change at least one of the style, format, and intent associated with a plurality of sample documents determined by one or more machine learning models, and an operation of generating an output document according to a document template through one or more machine learning models, which includes an operation of applying at least one of the style, format, and intent changed according to the selection for one or more of the plurality of user interface elements.

[0023] In some implementations, the computing device further includes one or more databases configured to store a plurality of generative machine learning models associated with a plurality of different document types, and the operations further include an operation to search among the plurality of generative machine learning models for at least one generative machine learning model associated with a document type represented by the common content of a plurality of sample documents.

[0024] In one or more exemplary embodiments, a computer-implemented method for organizing, managing, and creating content is provided. The computer-implemented method comprises: receiving a selection of a plurality of sample documents by a computing system comprising one or more processors; receiving an input by the computing system for generating a document template based on the selection of the plurality of sample documents; and, in response to receiving the input, implementing one or more machine-learned models of the computing system to generate a document template based on at least one of a style, format, or intent associated with the plurality of sample documents.

[0025] In one or more exemplary embodiments, a computer-readable medium (e.g., a non-transient computer-readable medium) is provided for storing instructions executable by one or more processors of a computing system. In some implementations, the computer-readable medium stores instructions that may include instructions for one or more processors to perform one or more operations (e.g., operations of a server computing system and / or operations of a computing device) associated with any of the methods described herein. The computer-readable medium may store additional instructions for executing other aspects of a server computing system and a computing device as described herein and corresponding methods of operation.

[0026] These and other features, aspects, and advantages of the various embodiments of the present disclosure will be better understood by referring to the following description, drawings, and appended claims. The appended drawings, which are incorporated into and form part of this specification, serve to illustrate exemplary embodiments of the present disclosure and, together with the description, explain the relevant principles. Brief explanation of the drawing

[0027] Detailed descriptions of exemplary embodiments intended for those skilled in the art are provided in the specification with reference to the attached drawings, which are as follows. FIGS. 1a and FIGS. 1b illustrate an exemplary system according to one or more exemplary embodiments of the present disclosure. FIG. 2 illustrates a flowchart of an exemplary, non-limiting computer implementation method according to one or more exemplary embodiments of the present disclosure. FIG. 3 illustrates an exemplary block diagram of a notebook application according to one or more exemplary embodiments of the present disclosure. FIGS. 4a through 4h illustrate exemplary user interface screens of a notebook application according to one or more exemplary embodiments of the present disclosure. FIGS. 5a and 5b illustrate exemplary user interface screens of a notebook application according to one or more exemplary embodiments of the present disclosure. FIGS. 6a and 6b illustrate exemplary user interface screens of a notebook application according to one or more exemplary embodiments of the present disclosure. FIG. 7 illustrates an exemplary notebook or project that may be represented in a particular manner according to one or more exemplary embodiments of the present disclosure. FIGS. 8a and 8b illustrate flowcharts of an exemplary, non-limiting computer implementation method according to one or more exemplary embodiments of the present disclosure. FIG. 9 illustrates an exemplary block diagram of a document extractor application according to one or more exemplary embodiments of the present disclosure. FIGS. 10a through 10e illustrate exemplary actions associated with a document extractor application according to one or more exemplary embodiments of the present disclosure. FIG. 11a illustrates a block diagram of an exemplary computing system for organizing, managing, and creating content by implementing one or more machine-learned models in relation to training content and / or source content selected by a user, according to one or more exemplary embodiments of the present disclosure. FIG. 11b illustrates a block diagram of an exemplary computing device for organizing, managing, and creating content by implementing one or more machine-learned models in relation to training content and / or source content selected by a user, according to one or more exemplary embodiments of the present disclosure. FIG. 11c illustrates a block diagram of an exemplary computing device for organizing, managing, and creating content by implementing one or more machine-learned models in relation to training content and / or source content selected by a user, according to one or more exemplary embodiments of the present disclosure. Specific details for implementing the invention

[0028] We will now refer to embodiments of the present disclosure, which have one or more examples illustrated in the drawings, wherein similar reference numerals denote similar elements. Each example is provided through the description of the present disclosure and is not intended to limit the present disclosure. In practice, it will be apparent to those skilled in the art that various modifications and variations may be made to the present disclosure without departing from the scope or spirit of the present disclosure. For example, features illustrated or described as part of one embodiment may be used in conjunction with other embodiments to yield another embodiment. Accordingly, the present disclosure is intended to cover such modifications and variations as those within the scope of the appended claims and their equivalents.

[0029] The terms used herein are for describing exemplary embodiments and are not intended to limit or / or restrict the disclosure. The singular forms ('one', 'one', and 'above') are intended to include the plural forms unless the context clearly indicates otherwise. In the disclosure, terms such as 'including', 'having', 'comprising', etc., are used to specify features, numbers, steps, actions, elements, components, or combinations thereof, but do not exclude the presence or addition of one or more of features, elements, steps, actions, elements, components, or combinations thereof.

[0030] While terms such as first, second, third, etc. may be used herein to describe various elements, it will be understood that the elements are not limited by these terms. Instead, these terms are used to distinguish one element from another. For example, without departing from the scope of this disclosure, a first element may be named a second element, and similarly, a second element may be named a first element.

[0031] The term 'and / or' includes a combination of multiple items listed in relation to each other or any of the multiple items listed in relation to each other. For example, the scope of the expression or phrase 'A and / or B' includes item 'A', item 'B', and a combination of items 'A and B'.

[0032] Additionally, the scope of the expression or phrase ‘at least one of A or B’ is intended to include all of the following: (1) at least one of A, (2) at least one of B, and (3) at least one of A and at least one of B. Likewise, the scope of the expression or phrase ‘at least one of A, B, or C’ is intended to include all of the following: (1) at least one of A, (2) at least one of B, (3) at least one of C, (4) at least one of A and at least one of B, (5) at least one of A and at least one of C, (6) at least one of B and at least one of C, and (7) at least one of A, at least one of B, and at least one of C.

[0033] Large-scale Language Models (LLMs) can generate general text content. However, if a user wishes to utilize an LLM to generate content of a specific type or style, the user must experiment with multiple different prompting strategies or retrain (fine-tune) the LLM on a large number of training samples displaying content of that specific type or style. Experimenting with different prompting strategies can lead to unnecessary and redundant processing and content generation, as users attempt to write the desired content multiple times until the appropriate style is achieved. Retraining a model is computationally expensive. Therefore, all of these approaches result in a waste of computational resources.

[0034] According to an example of the present disclosure, a computing system is configured to automatically extract a type or style from one or more source documents (e.g., notes) and then generate a template for creating a new (output) document by applying the extracted type or style. For example, the computing system may be trained to learn a specific template based on a plurality of documents of a specific type. For example, a user may upload a plurality of Product Requirements Documents (PRDs), and the computing system may be configured to learn the document structure and style. Subsequently, the user (or another user) may upload a plurality of source documents (e.g., a plurality of User Experience Research (UXR) documents) and provide a request to generate a document having a specific format (e.g., PRD format). The computing system may be configured to generate a PRD based on the plurality of source documents and based on the learned document structure and style.

[0035] In some implementations, a user may be provided with a graphical user interface to change or modify the style, format, and / or intent of a document template. For example, if the document template has a document format including a heading section, a description section, a background section, and a conclusion section provided in a specific order, a graphical user interface may be provided to the user that includes multiple user interface elements that can be selected to modify the order or arrangement of the sections. For example, if the document template has a specific style (e.g., an opinion-based style), a graphical user interface may be provided to the user that includes multiple user interface elements that can be selected to change the style of the document template to another style (e.g., an everyday language style).

[0036] In some implementations, the computing system may be configured to train one or more machine learning models to learn document types by learning the attributes of multiple training documents and / or to learn the intent, style, and format of multiple training documents. Multiple training documents may share one or more common features (e.g., identical or similar document types, identical or similar document structures, identical or similar styles, etc.).

[0037] In some implementations, the computing system includes one or more databases configured to store a plurality of generative machine learning models associated with a plurality of different document types, each respectively. The computing system may be configured to search among the plurality of generative machine learning models for at least one generative machine learning model associated with a document type represented by the common content of a plurality of source documents. For example, a plurality of source documents may each contain content associated with a specific type of document (e.g., resume, PRD, research paper, etc.).

[0038] According to current computing systems, Large Language Models (LLMs) can interact with text content. However, in current computing systems, significant effort is required to construct the specific prompts that the LLM will process. For example, a user may need to copy and paste content from a document into a chat box to query the LLM about the content. This switching between multiple windows or applications wastes a significant amount of computation time and resources (e.g., processor cycles).

[0039] According to an example of the present disclosure, a computing system (computing platform, computing device) is configured to generate a new type of output (e.g., an overview, report, summary, etc.) through one or more machine learning models based on source content provided to the computing system (e.g., provided by a user). For example, the computing system may be configured to receive source content selected by a user and to generate a summary of the source content, including the identification of one or more topics related to the source content, through one or more machine learning models.

[0040] For example, a user can identify and select a subset of documents (e.g., four documents) from a plurality of documents (a large document corpus) related to a specific topic (e.g., modern American history in the 1990s) provided to a computing system. The computing system may include one or more machine learning models configured to receive the selected documents as input and provide as output a summary (or report, paper, outline, etc.) related to the selected documents and identification results of key topics (e.g., via a document guide).

[0041] In some implementations, the computing system implements semantic search methods (e.g., clustering) and one or more machine learning models (e.g., one or more LLMs) to generate summaries, key topics, and suggested queries (e.g., questions) to produce a document guide for content identified or specified by the user (e.g., based on text bodies found in the content).

[0042] For example, in some implementations, the computing system may be configured to receive source content from a user. For example, the user may upload source content (e.g., documents, images, sound files, websites, videos, presentations, PDFs, etc.). In some implementations, the computing system may be configured to automatically generate information containing a summary of the source content, generate key themes found in the source content, and generate suggested topics and questions to help the user explore the source content further, in response to the user uploading the source content. The information may be displayed through a user interface. The user interface may be configured to receive input from the user on user interface elements corresponding to themes, questions, etc. (e.g., via touch input, mouse clicks, etc.). In response to receiving input from the user, the computing system may be configured to respond to the input by providing answers to questions or theme queries through one or more machine learning models based on the source content, for example.

[0043] In some implementations, the computing system may be configured to automatically generate information including reports, overviews, or rewrites of original content to create new content based on source content identified (selected) by the user in response to the user uploading source content. For example, the user may request the computing system to identify a specified number of themes from one or more documents, summarize client interactions occurring over a specified period (e.g., at least two weeks), and generate a specified number of ideas based on source documents, etc.

[0044] In some implementations, the source content that the LLM uses as a basis or references may be selected (e.g., curated) by the user. For example, the user may consider or indicate that the selected source content is trustworthy (e.g., trusted source content, authoritative source content, etc.) or has a higher priority than other content that does not have such designation. Accordingly, one or more machine learning models are configured to improve the accuracy and reliability of information and data provided to the user by generating summaries of content or creating new content based on trusted source content. Additionally, one or more machine learning models are configured to respond to questions about source content based on trusted source content, thereby improving the accuracy and reliability of information and data provided as responses to questions raised by the user.

[0045] In some implementations, the computing system may be configured to search, add, or remove source content. For example, a user may add or remove source content. For example, a user may provide input requesting the computing system to search for source content (e.g., by searching for academic papers on a specific topic), and may add the searched source content as part of selected source content that the user (and / or the computing system) deems trustworthy.

[0046] In some implementations, the computing system may be configured to receive additional source content when a user creates a new note, uploads source content to the computing system, or adds source content via a website. The computing system may be configured to generate or receive metadata regarding the added source content. For example, the metadata may include one or more of the following: title, author, upload date, date associated with the creation of the source content, and URL associated with the source content.

[0047] In some implementations, the computing system may be configured to delete or remove source content by having the user select source content and provide input requesting that the source content be deleted (e.g., from a notepad application). In some implementations, the source content may be deleted as a source used as a basis for generating summaries, key topics, etc., in the notepad application, but an original copy of the source content may be retained in another location.

[0048] In some implementations, the computing system may be configured to receive user input through a text input box (e.g., an open text input box). For example, user input may be in the form of a question (e.g., 'What did Nixon say about car usage in his speech?'). For example, user input may be in the form of a theme or idea (e.g., 'Nixon car crisis' or 'What is this document about?').

[0049] A computing system may be configured to provide a response to user input based on selected source content through one or more machine-learned models. In some implementations, the computing system is configured to display the number of sources (citations) used as grounds to provide the response. In some implementations, the computing system is configured to provide a display of the sources (citations) used as grounds for specific passages in the response. In some implementations, the computing system is configured to provide additional context regarding the sources (citations) used as grounds for specific passages in the response. For example, the computing system may indicate the passage (e.g., sentence or paragraph) within the source that formed the basis of a part of the response, and may additionally indicate preceding and / or subsequent passages within the source to provide additional context regarding the specific passage.

[0050] In some implementations, the computing system may be configured to store one or more phrases (e.g., snippets) from the response (answer) generated for a query (question) entered by the user. For example, one or more phrases may be stored in a designated area of ​​a notepad application. The designated area may be referred to as a scratchpad, and each piece of information stored in the scratchpad may be referred to as a note. One or more phrases may be selected by the user to be stored as a first note within the scratchpad. In some implementations, a quotation may be stored in the scratchpad as a second note. In some implementations, the user may select (e.g., highlight) specific phrases from a quotation (source content) to be stored as a third note in the scratchpad. In some implementations, the user may store their own writing or comments as a written note (fourth note).

[0051] According to examples of the present disclosure, a computing system may be configured to provide a notepad application configured to generate output (e.g., an outline, a report, a summary, etc.) through one or more machine learning models based on source content provided to the notebook application (e.g., by a user). The notebook application may be configured to allow the user to create various projects to complete various tasks. Each project may be configured to function in a manner similar to a folder, where the user can store various information in each project. In some implementations, individual scratchpads may correspond to a specific project or be used exclusively for a specific project. In some implementations, the notebook application may be configured to receive source content specified by the user. The notebook application may be configured to add, delete, or modify projects based on input received from the user. Each project may be provided with a name generated by the notebook application (e.g., based on source content), a name provided by the user, or a default name (e.g., through one or more machine learning models) based on information stored in the project.

[0052] In some implementations, in response to source content being provided to the notebook application, the notebook application may be configured to automatically generate a graphic image (e.g., emoji, icon, etc.) or graphic animation corresponding to or representing the source content (e.g., using one or more machine learning models, one or more generative machine learning models, etc.). In some implementations, the graphic image or graphic animation may be overlaid on the folder provided as a user interface element that causes the folder to open and the folder's contents to be displayed to the user when selected. Additionally or alternatively, in some implementations, in response to source content being provided to the notebook application, the notebook application may be configured to automatically generate a text description (name) corresponding to or representing the source content (e.g., using one or more machine learning models, one or more generative machine learning models, etc.). The text description may be overlaid on the folder provided as a user interface element that causes the folder to open and the folder's contents to be displayed to the user when selected.

[0053] One or more technical advantages of the present disclosure include generating content through one or more machine learning models based on specific content items selected by a user. Current methods for having a large language model (LLM) generate output require the user to copy content from one document and paste it into a chat box to query the LLM regarding the content. A significant amount of computational time and resources (e.g., processor cycles) are wasted due to switching between multiple windows or applications. Unlike current methods, in response to the user uploading content items (e.g., source content), a summary regarding the content items selected by the user can be automatically generated through a notebook application and one or more machine learning models. Thus, the user does not need to switch between applications or windows or provide prompts.

[0054] Another technical advantage of the present disclosure includes one or more machine-learned models providing a proposed query based on a content item selected by a user, a proposed key topic based on a content item selected by the user, and a selectable chip based on the output of the response. The user may select a proposed query, and one or more machine-learned models may be configured to provide a response to the query based on the content item selected by the user. Providing a proposed query automatically saves computing resources (e.g., networking resources including bandwidth, processor cycles, etc.) by not requiring the user to input the proposed query.

[0055] Another technical advantage of the present disclosure is that one or more machine-learned models generate content based on content items selected by the user and / or notes selected by the user. Generating content saves time and computing resources because the user does not need to cut and paste content from multiple sources to generate new content (e.g., outlines, essays, reports, etc.) based on multiple content items.

[0056] Another technical advantage of the present disclosure is that one or more machine-learned models generate a graphic image or animation to be displayed in a manner overlaid on a folder to represent the content stored in the folder. The graphic image or animation can improve search capabilities and save computing resources that might otherwise be consumed opening and closing folders that do not contain the content the user is actually looking for.

[0057] Another technical advantage of the present disclosure includes one or more machine-learned models providing a proposed query based on a content item selected by a user, a proposed key topic based on a content item selected by the user, and a selectable chip based on the output of the response. The user may select a proposed query, and one or more machine-learned models may be configured to provide a response to the query based on the content item selected by the user. Providing a proposed query automatically saves computing resources (e.g., networking resources including bandwidth, processor cycles, etc.) by not requiring the user to input the proposed query.

[0058] Another technical advantage of the present disclosure is that one or more machine-learned models generate document templates based on training documents that may be provided by a user. For example, a user may not be familiar with the common or standard formats of a specific document type (e.g., resume, PRD, legal opinion, etc.) and may not possess the knowledge or expertise to provide appropriate prompts for generating document templates that have a format suitable for that specific document type. Accordingly, the document extractor application described herein can improve the efficiency of one or more machine-learned models by generating document templates with appropriate structures (e.g., style, intent, format, etc.), and said document templates can be used to generate output documents based on the correct document templates. Accordingly, reducing interaction with the user to generate document templates can save or conserve bandwidth, network resources, computing processing power, etc. Additionally, reducing the number of inferences performed by one or more machine-learned models to generate document templates can also save or conserve bandwidth, network resources, computing processing power, etc.

[0059] Another technical advantage of the present disclosure is that one or more machine-learned models generate an output document based on a source document that may be provided by a user and a document template that may be selected by the user or a computing device. For example, a user may not be familiar with the common or standard formats of a specific document type (e.g., resume, PRD, legal opinion, etc.) and may not possess the knowledge or expertise to provide appropriate prompts for generating an output document with a format suitable for that specific document type. Accordingly, the document extractor application described herein can improve the efficiency of one or more machine-learned models by generating an output document with an appropriate structure (e.g., style, intent, format, etc.) based on a document template suitable for the document type. Accordingly, reducing the interaction with the user to generate an output document based on a document template generated by one or more machine-learned models can save or conserve bandwidth, network resources, computing processing power, etc. Additionally, reducing the number of inferences performed by one or more machine-learned models to generate an output document can also save or conserve bandwidth, network resources, computing processing power, etc.

[0060] Now, referring to the drawings, FIG. 1a is an exemplary system according to one or more exemplary embodiments of the present disclosure. FIG. 1a illustrates an example of a system (1000) comprising a computing device (100), an external computing device (200), a server computing system (300), and external content (500) that can communicate with each other through a network (400). For example, the computing device (100) and the external computing device (200) may include a personal computer, a smartphone, a tablet computer, a laptop, a global positioning service device, a smartwatch, etc. The network (400) may include any type of communication network including a wired or wireless network or a combination thereof. The network (400) may include a local area network (LAN), a wireless local area network (WLAN), a wide area network (WAN), a personal area network (PAN), a virtual private network (VPN), etc. For example, wireless communication between elements of the exemplary embodiment may be performed via wireless LAN, Wi-Fi, Bluetooth, ZigBee, Wi-Fi Direct (WFD), Ultra-Wideband (UWB), Infrared Data Communication (IrDA), Low Power Bluetooth (BLE), Near Field Communication (NFC), radio frequency (RF) signals, etc. For example, wired communication between elements of the exemplary embodiment may be performed via pair cable, coaxial cable, fiber optic cable, Ethernet cable, etc. Communication through the network (400) may use various communication protocols (e.g., TCP / IP, HTTP, SMTP, FTP), encoding or format (e.g., HTML, XML), and / or protection schemes (e.g., VPN, secure HTTP, SSL).

[0061] As described in more detail below, in some implementations, the computing device (100) and / or server computing system (300) may form part of an application system capable of providing a tool for a user to create, manage, or organize information (e.g., documents, images, etc.) through, for example, one or more machine-learned models.

[0062] In some exemplary embodiments, the server computing system (300) may obtain data from one or more of a content data store (350), a user data store (360), and a machine learning model data store (370) to implement various operations and aspects of the application system disclosed herein. The content data store (350), the user data store (360), and the machine learning model data store (370) may be provided integrally with the server computing system (300) (e.g., as part of one or more memory devices (320) of the server computing system (300)) or provided separately (e.g., remotely). Additionally, the content data store (350), the user data store (360), and the machine learning model data store (370) may be combined into a single data store (database) or may include a plurality of individual data stores. Data stored in one data store (e.g., content data store (350)) may overlap with some data stored in another data store (e.g., user data store (360)). In some implementations, one data store (e.g., machine learning model data store (370)) may reference data stored in another data store (e.g., user data store (360)).

[0063] In some examples, the content data store (350) may store any type of information or content. For example, the content data store (350) may include books, product manuals, resumes, legal opinions, academic papers, proprietary data files, patent documents, web pages, emails, forum posts, social media posts, videos, images, geographic information, or any other type or form of content that can be stored or accessed in digital form (e.g., content stored in a database, memory device, etc.). In some implementations, information may be stored in the content data store (350) by the user selecting specific documents, images, or other content to be stored in the content data store (350).

[0064] In some examples, the user data store (360) may contain information about one or more user profiles, including various user data such as user preference data, user demographic data, user calendar data, user social network data, user past travel data, etc. For example, the user data store (360) may contain email data including text content, images, calendar information or contact information associated with email; social media data including comments, reviews, check-ins, likes, invitations, contacts or reservations; calendar application data including dates, times, events, descriptions or other content; virtual wallet data including purchase history, e-tickets, coupons or discounts; scheduling data; location data; SMS data; or other suitable data associated with a user account, but is not limited thereto. According to one or more examples of the present disclosure, data may be analyzed to determine user preferences regarding the creation, management, and / or organization of content, for example, to automatically generate a summary of a document in a particular manner or style, to automatically provide customized features regarding the content, to provide suggestions, recommendations, and / or questions related to specific content identified by the user as source content, to automatically generate document templates, and to automatically generate documents based on document templates.

[0065] A user data store (360) is provided to exemplify potential data that may be analyzed by a computing device (100) and / or a server computing system (300) for purposes such as identifying user preferences, providing recommendations, and creating, managing, and / or organizing content, in some embodiments. However, such user data may not be collected, used, or analyzed unless the user consents after being notified of what data is collected and how such data is used. Additionally, in some embodiments, a tool for revoking or modifying the scope of permissions may be provided to the user (e.g., in a notebook application or through a user account). Additionally, certain information or data may be processed in one or more ways before being stored or used so that personally identifiable information is removed or stored in an encrypted form. Accordingly, certain user information stored in the user data store (360) may or may not be accessed by the computing device (100) and / or the server computing system (300) depending on the permissions granted by the user, and such data may not be stored in the user data store (360) at all.

[0066] The machine learning model data repository (370) may store machine learning models that can be used to generate distilled or fine-tuned machine learning models (e.g., distilled or fine-tuned generative machine learning models) which can be retrieved and implemented by the server computing system (300) and provided to the computing device (100) in some implementations. The machine learning model data repository (370) may also store distilled or fine-tuned machine learning models (e.g., distilled or fine-tuned generative machine learning models) which can be retrieved and implemented by the computing device (100). In some implementations, the computing device (100) may retrieve and implement machine learning models that are large parameter models that are not fine-tuned or distilled. The machine learned models stored in the machine learned model data repository (370) (including large parameter models and distilled or fine-tuned models) may include generative machine learned models associated with different types of content (e.g., different genres or topics, different kinds of content including images, videos and text, different types of content or documents (e.g., outlines, reports, spreadsheets, etc.), documents having different styles (e.g., everyday conversation style, opinion-presenting style, expert style, etc.), documents having different formats (e.g., outline format, PRD format, resume format, etc.), and documents having different intentions (e.g., informative, persuasive, comparative, contrasting, etc.). The machine learned models may include large language models (e.g., BERT (Bidirectional Encoder Representations from Transformers) large language models) and general multimodal models (e.g., Gemini). Machine-learned models may include generative artificial intelligence (AI) models (e.g., Bard) capable of implementing Generative Adversarial Networks (GANs), Transformers, Variational Autoencoders (VAEs), Neural Radiation Fields (NeRF), etc.

[0067] External content (500) may be any form of external content including news articles, web pages, video files, audio files, written descriptions, reviews, game content, social media content, photos, commercial offers, transportation methods, weather conditions, sensor data acquired by various sensors, or other suitable external content. A computing device (100), an external computing device (200), and a server computing system (300) may access the external content (500) through a network (400). The external content (500) may be searched by the computing device (100), the external computing device (200), and the server computing system (300) according to a known search method, and the search results may be ranked according to relevance, popularity, or filtering by location, or other suitable attributes including promotion.

[0068] Now, with reference to FIG. 1b, an exemplary block diagram of a system (1000') including a computing device (100) and a server computing system (300) according to one or more exemplary embodiments of the present disclosure will be described. Although the computing device (100) is shown in FIG. 1b, the features of the computing device (100) described herein are applicable to an external computing device (200).

[0069] A computing device (100) may include one or more processors (110), one or more memory devices (120), an application system (130), a position determination device (140), an input device (150), a display device (160), an output device (170), and a capture device (180). A server computing system (300) may include one or more processors (310), one or more memory devices (320), and an application system (330).

[0070] For example, one or more processors (110, 310) may be any suitable processing device that can be included in a computing device (100) or a server computing system (300). For example, one or more processors (110, 310) may include one or more of a processor, a processor core, a controller and an arithmetic logic unit, a central processing unit (CPU), a graphics processing unit (GPU), a digital signal processor (DSP), an image processor, a microcomputer, a field programmable array, a programmable logic unit, an application-specific integrated circuit (ASIC), a microprocessor, a microcontroller, etc., and a combination thereof, and may include any other device capable of responding to and executing instructions in a defined manner. One or more processors (110, 310) may be a single processor or, for example, a plurality of processors connected to operate in parallel.

[0071] One or more memory devices (120, 320) may include volatile memory devices including read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM) and flash memory, USB drives, random access memory (RAM), hard disks, floppy disks, Blu-ray disks, or optical media such as CD-ROM disks and DVDs, and combinations thereof. However, examples of one or more memory devices (120, 320) are not limited to the description above, and one or more memory devices (120, 320) may be realized by various other devices and structures understood by those skilled in the art.

[0072] For example, one or more memory devices (120) may also include data (122) and instructions (124) that can be retrieved, manipulated, generated, or stored by one or more processors (110). In some exemplary embodiments, such data may be accessed and used as input to execute instructions for performing operations to implement a notebook application (132) and provide a user interface including a first part and a second part as described according to examples of the present disclosure, wherein the first part includes a text summary generated through one or more machine learning models based on a plurality of documents selected by the user, and the second part includes a plurality of user interface elements for performing operations regarding the text summary. In some exemplary embodiments, such data may be accessed and used as input to execute instructions to implement a document extractor application (136) and perform the following operations as described according to examples of the present disclosure: an operation to receive a plurality of source documents; an operation to receive an input associated with a request to generate an output document based on a plurality of source documents and a specific document template; and an operation to generate an output document having a specific document template through one or more machine learning models.

[0073] One or more memory devices (320) may also include data (322) and instructions (324) that can be retrieved, manipulated, generated, or stored by one or more processors (310). In some exemplary embodiments, such data may be accessed and used as input to execute instructions for performing operations to implement a notebook application (332) and provide a user interface including a first part and a second part as described according to examples of the present disclosure, wherein the first part includes a text summary generated through one or more machine learning models based on a plurality of documents selected by the user, and the second part includes a plurality of user interface elements for performing operations regarding the text summary. In some exemplary embodiments, such data may be accessed and used as input to execute instructions to implement a document extractor application (336) and perform the following operations as described according to examples of the present disclosure: an operation to receive a plurality of source documents; an operation to receive an input associated with a request to generate an output document based on a plurality of source documents and a specific document template; and an operation to generate an output document having a specific document template through one or more machine learning models.

[0074] In some exemplary embodiments, the computing device (100) includes an application system (130). For example, the application system (130) may include a notebook application (132), a document application (134) (e.g., a word processing application, a spreadsheet application, a presentation application, an image application, etc.), and a document extractor application (136). The application system (130) may include various other applications, such as a text messaging application, an email application, a dictation application, a virtual keyboard application, a browser application, a map application, a social media application, a navigation application, etc.

[0075] According to an example of the present disclosure, a notebook application (132) may be executed by the computing device (100) to provide a user of the computing device (100) with a way to organize, manage, and create content, particularly content selected or chosen by the user, and to interact with said content. In some implementations, the notebook application (132) may be part of a document application (134) or may be a standalone application. The notebook application (132) may be configured to interact dynamically according to various user inputs. While exemplary implementations of the notebook application (132) are described herein, the present disclosure is not limited to these examples, as various modifications may be made to the embodiments described herein.

[0076] In some examples, one or more aspects of the notebook application (132) may be implemented by a notebook application (332) of a server computing system (300) that may be located remotely to organize, manage, and create content and interact with content in response to receiving input from a user. In some examples, one or more aspects of the notebook application (332) may be implemented by a notebook application (132) of a computing device (100) to organize, manage, and create content and interact with content in response to receiving input from a user.

[0077] According to an example of the present disclosure, a document application (134) may be executed by the computing device (100) to provide a user of the computing device (100) with a way to organize, manage, and create content, particularly content selected or chosen by the user, and to interact with said content. The document application (134) may be any type of application related to a document (e.g., a document in a text or visual format) and may include a word processing application, a spreadsheet application, a presentation application, a visual application, a PDF file application, etc. In some implementations, a notebook application (132), a document application (134), and a document extractor application (136) may interact with each other. For example, content from a document created through the document application (334) may be uploaded or stored for use with the notebook application (132). In some implementations, the notebook application (132) may be configured to generate a document (e.g., a report, outline, presentation, spreadsheet) that is compatible with the document application (134) (that can be opened in the document application (134) or exported to the document application (134)). In some implementations, the notebook application (132) may be configured to generate a document (e.g., a report, outline, presentation, spreadsheet) that has a specific style, format, and / or intent based on a document template generated through the document extractor application (136).

[0078] In some examples, the document application (134) may be a dedicated application specifically designed to provide a specific service. In other examples, the document application (134) may be a general-purpose application (e.g., a web browser) and may provide access to various different services through the network (400).

[0079] According to an example of the present disclosure, a document extractor application (136) may be executed by the computing device (100) to provide a document template to the user of the computing device (100) for organizing, managing, and creating content, particularly content selected or chosen by the user, and for interacting with said content. In some implementations, the document extractor application (136) may be part of a document application (134) and / or a notebook application (132), or it may be a standalone application. The document extractor application (136) may be configured to interact dynamically according to various user inputs. While exemplary implementations of the document extractor application (136) are described herein, the present disclosure is not limited to these examples, as various modifications may be made to the embodiments described herein.

[0080] In some examples, one or more aspects of the document extractor application (136) may be implemented by the document extractor application (336) of a server computing system (300) that may be located remotely, so as to provide document templates for organizing, managing, and creating content and interacting with content in response to receiving input from a user. In some examples, one or more aspects of the document extractor application (336) may be implemented by the document extractor application (136) of a computing device (100), so as to provide document templates for organizing, managing, and creating content and interacting with content in response to receiving input from a user.

[0081] In some exemplary embodiments, the computing device (100) includes a position determining device (140). The position determining device (140) can determine the current geographic location of the computing device (100) and communicate this geographic location with a server computing system (300) via a network (400). The position determining device (140) may be any device or circuit for analyzing the position of the computing device (100). For example, the position determining device (140) may determine the actual or relative position by using a satellite navigation positioning system (e.g., a GPS system, a Galileo positioning system, a GLONASS (Global Navigation Satellite System), a BeiDou satellite navigation and positioning system), an inertial navigation system, or a dead reckoning system, or by using an IP address, or by using triangulation and / or proximity with a cellular tower or Wi-Fi hotspot, or by using other suitable techniques for determining the position of the computing device (100).

[0082] The computing device (100) may include an input device (150) configured to receive input from a user, and the input device (150) may include one or more of, for example, a keyboard (e.g., a physical keyboard, a virtual keyboard, etc.), a mouse, a joystick, a button, a switch, an electronic pen or a stylus, a gesture recognition sensor (e.g., a sensor for recognizing a user’s gestures including movements of a part of the body), a sound input device or a speech recognition sensor (e.g., a microphone for receiving voice input such as voice commands or voice queries), a trackball, a remote controller, a mobile phone (e.g., a cellular phone or a smartphone), a tablet PC, a pedal or foot switch, a virtual reality device, etc. The input device (150) may also be embodied as a touch-sensitive display having, for example, a touchscreen. For example, the input device (150) may be configured to receive user input associated with the input device (150), such as selecting content to be configured or managed, selecting a query or action regarding content selected or chosen by the user, uploading multiple training documents to create a document template that can be applied to multiple source documents, and selecting multiple source documents to create an output document based on the document template.

[0083] A computing device (100) may include a display device (160) that displays information that a user can view (e.g., a user interface screen). For example, the display device (160) may be a non-touch-sensitive display or a touch-sensitive display. For example, the display device (160) may include a liquid crystal display (LCD), a light-emitting diode (LED) display, an organic light-emitting diode (OLED) display, an active matrix organic light-emitting diode (AMOLED), a flexible display, a 3D display, a plasma display panel (PDP), a cathode ray tube (CRT) display, etc. However, the present disclosure is not limited to these exemplary displays and may include other types of displays. The display device (160) may be used by an application system (130) provided to the computing device (100) to display information regarding input (e.g., information regarding documents, notes, projects, document templates, user interface screens having user interface elements that the user can select) to the user.

[0084] The computing device (100) may include an output device (170) for providing output to a user, and may include one or more of, for example, an audio device (e.g., one or more speakers), a tactile device for providing tactile feedback to a user (e.g., a vibration device), a light source (e.g., one or more light sources such as an LED for providing visual feedback to a user), a thermal feedback system, etc.

[0085] According to various examples of the present disclosure, a computing device (100) may include a capture device (180) capable of capturing media content. For example, the capture device (180) may include an image capturer (182) (e.g., a camera) configured to capture an image (e.g., a photograph, video, etc.). For example, the capture device (180) may include a sound capturer (184) (e.g., a microphone) configured to capture sound or audio (e.g., an audio recording). Media content captured by the capture device (180) may be transmitted, for example, via a network (400) to one or more of a server computing system (300), a content data store (350), a user data store (360), and a machine learning model data store (370). For example, in some implementations, media content captured by the capture device (180) may be selected by a user as source content to be used for writing notes about a project. Media content can be provided as input to one or more machine-learned models to generate notes, for example.

[0086] According to an exemplary embodiment of the present disclosure, the server computing system (300) may include one or more processors (310) and one or more memory devices (320) as described herein. The server computing system (300) may also include an application system (330) similar to the application system (130) described herein.

[0087] For example, the application system (330) may include a notebook application (332) that performs functions similar to those described herein for the notebook application (132), a document application (334) that includes an application similar to the application described above for the document application (134), and a document extractor application (336) that performs functions similar to those described herein for the document extractor application (136). In some implementations, one or more machine learning models (e.g., generative machine learning models, large language models, etc.) associated with the application system (330) may be configured to organize, manage, and create content and interact with content based on source content selected or curated by the user. For example, one or more machine learning models (e.g., generative machine learning models, large language models, etc.) associated with the application system (330) may be configured to perform a first action (e.g., an action to generate a summary or document guide regarding source content selected by the user), while the computing device (100) may be configured to perform a second action (e.g., an action to generate a proposed action or an action to generate an outline or learning guide based on a plurality of notes stored in a scratchpad). For example, one or more machine learning models (e.g., generative machine learning models, large language models, etc.) associated with the application system (130) may be configured to perform a first action (e.g., an action to upload source content selected by the user), while the server computing system (300) may be configured to perform a second action (e.g., an action to generate a document template based on uploaded source content). For example, specific actions to be performed by the application system (330) may vary depending on network conditions (e.g., available bandwidth, channel usage status, latency, throughput, etc.).In some implementations, one or more machine learning models associated with the application system (330) may be configured to process user input to generate information (e.g., semantic information), and this information may then be provided as input to one or more other machine learning models associated with the application system (330) (e.g., generative machine learning models, large language models, etc.) to generate content to be utilized in relation to the notebook application (132) and / or the project for the notebook application (332).

[0088] Examples of the present disclosure also relate to a computer implementation method for providing a user interface for organizing, managing, and creating content by implementing one or more machine-learned models with respect to source content selected by a user. FIG. 2 illustrates a flowchart of an exemplary, non-limiting computer implementation method according to one or more exemplary embodiments of the present disclosure. FIG. 3 illustrates a block diagram of a notebook application according to one or more exemplary embodiments of the present disclosure.

[0089] The flowchart of FIG. 2 illustrates a method (2000) for providing a user interface for organizing, managing, and creating content by implementing one or more machine learning models with respect to source content selected by the user. Although illustrated in a specific sequence or order, the order of the processes may be modified unless otherwise specified. Accordingly, the illustrated embodiments should be understood merely as examples, and the illustrated processes may be performed in a different order, and some processes may be performed in parallel. Additionally, one or more processes may be omitted in various embodiments. Accordingly, not all processes are required in all embodiments. Other process flows are also possible.

[0090] Referring to FIG. 2, in operation (2100), the method (2000) includes the step of the computing device receiving input from the user regarding the selection of source content. As described herein, the computing device may be implemented as a computing device (100), a server computing system (300), or a combination thereof. For example, the input may be provided by the user through an input device (150). For example, the input may be provided by selecting a specific file or document uploaded to the computing device for use by a notebook application (132). In some implementations, the content may be uploaded from local memory, another application (e.g., a PDF application), copied text, or a website. The selected file, text, document, etc. may be referred to as source content. In some implementations, the source content may be a subset of a larger content corpus. The input may be provided to or input into, for example, the notebook application (132) or the notebook application (332).

[0091] In some implementations, the response to an input selecting source content may be processed at the computing device (100) without involving the server computing system (300). In some implementations, the input selecting source content may be transmitted from the computing device (100) to the server computing system (300), and at least a portion of the response to the input may be processed by the server computing system (300). For example, the input related to the selection of source content may be provided at the computing device (100), and the server computing system (300) may be configured to perform an action in response to receiving an indication of the input.

[0092] In operation (2200), the computing device may be configured to implement one or more machine learning models regarding source content selected to generate a document guide. In some implementations, the document guide (source guide) generated by one or more machine learning models may include a summary of the source content and key topics regarding the source content. In some implementations, the document guide may further include one or more proposed queries (e.g., questions) that may be provided in the form of selectable user interface elements.

[0093] For example, a computing device may obtain information indicating that a user has selected source content. The computing device may process the source content with one or more machine-learned models (e.g., one or more large-scale language models) to obtain language output. Then, the computing device may generate summary output using one or more machine-learned models (e.g., one or more large-scale language models). In particular, machine-learned large-scale language models may be trained to process various outputs to generate language output. For example, machine-learned large-scale language models may process embeddings generated by a machine-learned embedding generation model, parts of source content identified using an embedding generation model, language output generated using the machine-learned large-scale language model or some other model, etc.

[0094] In operation (2300), the computing device may be configured to receive input for performing an action on the document guide. In operation (2400), the computing device may be configured to perform an action in response to receiving the input. For example, the input may be a selection for a proposed query, and the action may include providing an answer to the question by implementing one or more machine learning models on the source content. For example, the input may be a text input asking a question, and the action may include providing an answer to the question by implementing one or more machine learning models on the source content. For example, the input may be a selection for a portion of a summary, and the action may include providing an output representing specific sources among the source content used as a basis for generating text associated with the selected portion of the summary.

[0095] Referring to FIG. 3, a notebook application (3100) (corresponding to notebook application (132) and / or notebook application (332)) may include a conditioning parameter generator (3110), one or more sequence processing models (3120), one or more large-scale language models (3130), and one or more generative machine learning models (3140). The notebook application (3100) may receive input (3200) from a user as described above with respect to the operation (2100) and operation (2300) of FIG. 2. The conditioning parameter generator (3110) may be configured to generate conditioning parameters based at least partially on the input (3200), the conditioning parameters provide values ​​for one or more conditions associated with the content to be generated, and the content to be generated is at least partially associated with the input (3200) and the source content (3400) selected by the user.

[0096] For example, source content (3400) may include any type of document (e.g., a document in digital form) and may include books, product manuals, legal opinions, academic papers, proprietary data files, patent documents, web pages, emails, forum posts, social media posts, videos, images, geographic information, or any other type or form of content that can be stored or accessed in digital form (e.g., content stored in a database, memory device, etc.). In some implementations, source content (3400) may be stored in the content data store (350) by the user selecting specific documents, images, or other content to be stored in the content data store (350). In some implementations, source content (3400) may be stored in a computing device (100) or a server computing system (300).

[0097] To generate a conditioning parameter, the conditioning parameter generator (3110) may be configured to retrieve values ​​for one or more conditions associated with the input. For example, to generate a conditioning parameter, the conditioning parameter generator (3110) may be configured to extract values ​​for one or more conditions from the input. The input may contain information indicating the user's intent or requirements. In some implementations, the conditioning parameter generator (3110) (or one or more sequence processing models (3120) or one or more large language models (3130)) may be configured to extract information from the input (3200) to identify values ​​for one or more conditions, and the conditioning parameter generator (3110) may be configured to generate a conditioning parameter based on the extracted values. For example, the input itself can identify attributes or features that can be used to generate a color to be used for the heading of the generated document (e.g., 'blue font for heading') or conditioning parameters to generate a document related to the source content (e.g., 'circular bullet points').

[0098] To generate a conditioning parameter, the conditioning parameter generator (3110) may be configured to infer a value for one or more conditions from the input. The input may include information indicating a user's intent or requirement. In some implementations, the conditioning parameter generator (3110) (or one or more sequence processing models (3120) or one or more large language models (3130)) may be configured to infer information from the input (3200) to identify a value for one or more conditions, and the conditioning parameter generator (3110) may be configured to generate a conditioning parameter based on the inferred value. For example, the input may include a reference to the length ('short', 'long', etc.) of a summary to be generated or another document to be generated based on source content, and the conditioning parameter generator (3110) (or one or more sequence processing models (3120) or one or more large language models (3130)) may be configured to infer a value based on the input. For example, an input requesting the notebook application (3100) to generate a 'short' essay can infer a value of about 500 words, whereas a 'long' essay can be associated with a value of about 2000 words. For example, the notebook application (3100) can be configured to check the inferred value based on information from external content (3300).

[0099] In some implementations, the conditioning parameter generator (3110) may be configured to infer values ​​for one or more conditions from the input by providing input to one or more sequence processing models (3120), wherein the one or more sequence processing models (3120) are configured to respond to a query or output values ​​for one or more conditions based on a query. The one or more sequence processing models (3120) may include one or more machine learning models configured to process and analyze sequential data and to process data occurring in a specific order or sequence, including time-series data, natural language text, or any other data having a temporal or sequential structure.

[0100] One or more sequence processing models (3120) receive an input containing text and can tokenize the input by breaking down the sequence of text into small units (tokens) to provide a structured representation of the input sequence. One or more sequence processing models (3120) can represent the tokens as vectors in a continuous vector space by mapping each token to a high-dimensional vector, where the relationship between the tokens (words) is reflected in the geometric relationship between the corresponding vectors. For example, one or more sequence processing models (3120) receive an input containing the text 'How did the Cold War end?' and can provide a structured representation of the input sequence by tokenizing the input by breaking down the sequence of text into small units (tokens) (e.g., 'how', 'Cold War', and 'end'). In word embeddings, semantically similar words are closer to each other in the vector space. For example, vectors for 'war' and 'combat' can be close to each other due to their semantic relationship, whereas vectors for 'war' and 'peace' can be far apart compared to vectors for 'war' and 'combat'.

[0101] One or more large language models (3130) may be models trained on a large corpus of language training data to enable the one or more large language models (3130) to perform multiple language tasks, or otherwise may include such models. For example, one or more large language models (3130) may be trained to perform summarizing tasks, conversation tasks, simplification tasks, opposing viewpoint tasks, etc. In particular, one or more large language models (3130) may be trained to generate language output by processing various outputs. For example, one or more large language models (3130) may process embeddings generated by a machine learning embedding generation model, parts of source content identified using an embedding generation model (e.g., document chunks), language output generated using one or more large language models (3130) or some other model, etc.

[0102] One or more generative machine learning models (3140) may include a deep neural network or a generative adversarial network (GAN), a variational autoencoder, a stable diffusion machine learning model, a visual transformer, a neural radiation field (NeRF), etc., to generate content (e.g., a summary, a response to a query, etc.) having a value of a condition associated with one or more features. For example, a computing device may include a database (e.g., a machine learning model data store (370)) configured to store a plurality of generative machine learning models associated with a plurality of different types of content (e.g., different genres or topics, different kinds of content including images, videos, and text, different styles of content including outlines, reports, spreadsheets, etc.). In some implementations, the computing device may be configured to search for a generative machine learning model associated with a specific type of content related to the input among one or more generative machine learning models (3140).

[0103] In some implementations, one or more generative machine learning models (3140) may be trained on a large dataset of content having corresponding information on conditions associated with the content (e.g., a large corpus of language training data). During training, one or more generative machine learning models (3140) learn the relationships between elements within the output (e.g., content) and the conditions affecting them. This may include the computing device adjusting the internal parameters of each generative machine learning model to generate realistic or accurate content (e.g., grammatically correct content, consistent content, etc.) based on the training data. One or more generative machine learning models (3140) may be trained on one or more training datasets containing multiple reference images of a location. One or more training datasets may contain values ​​for one or more conditions.

[0104] In some implementations, one or more generative machine learning models (3140) are configured to generate a response content (3600) corresponding to the content generated in response to an input for performing an action on the document guide, etc., based on conditioning parameters (and corresponding values ​​for one or more conditions) to make a decision to generate the document guide (3500) in response to receiving a selection for source content (3400) and / or to generate content.

[0105] In some implementations, the server computing system (300) may provide (transmit) content or part of the generated content to the computing device (100), or the server computing system (300) may provide access to the generated content to the computing device (100). For example, a document guide (3500) may be generated in the server computing system (300) and stored in one or more computing devices (e.g., one or more of the computing device (100), external computing device (200), server computing system (300), external content (500), content data storage (350), user data storage (360), etc.).

[0106] In some implementations, after a document guide is generated and / or an action is performed on the document guide, the user may provide feedback or additional input regarding the generated content based on the provided source content and / or the query provided through the user, and one or more of the actions (2100 to 2400) may be repeated.

[0107] Examples of the present disclosure also relate to user-facing modes that enable a user to manage, organize, create, etc. content through a notebook application configured to implement one or more machine-learned models for source content selected by the user. For example, FIGS. 4a through 4h illustrate examples of actions that can be implemented for a project in which a document guide is generated through one or more machine-learned models based on source content selected by the user according to one or more exemplary embodiments of the present disclosure.

[0108] For example, FIG. 4a illustrates a first user interface screen of a notebook application (e.g., a start user interface screen, a start graphical user interface, etc.) according to one or more exemplary embodiments of the present disclosure.

[0109] In FIG. 4a, the first user interface screen (4100) illustrates a user interface (e.g., an execution screen) that provides information regarding a notebook application (3100). In particular, the notebook application (3100) is configured to display the first user interface screen (4100) which includes various information (4110) regarding functions available in the notebook application (3100).

[0110] As illustrated in FIG. 4b, the notebook application (3100) is further configured to display a second user interface screen (4200) comprising a first user interface element (4210). For example, the first user interface element (4210) is associated with enabling the user to create a new notebook (new project) in which the user can manage, organize, and create content based on source content that the user can select or choose.

[0111] As illustrated in FIG. 4c, the notebook application (3100) is further configured to display a third user interface screen (4300) in response to a user providing input to create a new notebook (e.g., through the selection of a first user interface element (4210). The third user interface screen (4300) includes a first part (4310) having a plurality of selectable user interface elements corresponding to a location where source content can be uploaded. For example, the first user interface element (4312) corresponds to a storage space or other storage device (e.g., a portable storage device) that may be associated with a local computing device or a remote server system (e.g., a cloud server). For example, the second user interface element (4314) corresponds to a PDF file, the third user interface element (4316) corresponds to copied text, and the fourth user interface element (4318) corresponds to content that can be uploaded from a specific website or URL.

[0112] As illustrated in FIG. 4d, the notebook application (3100) is further configured to display a fourth user interface screen (4400) in response to the selection of one of a plurality of selectable user interface elements corresponding to a location where source content can be uploaded, as described in relation to FIG. 4c. The fourth user interface screen (4400) includes a first part (4410) having a plurality of selectable source content items (e.g., a plurality of documents, images, videos, etc.). For example, a first user interface element (4412) corresponds to a selected first document, a second user interface element (4414) corresponds to a selected second document (e.g., a PDF file), and a third user interface element (4416) corresponds to a selected third document. FIG. 4d illustrates that a user can select or choose specific source content items that can be used to create a notebook or project, and that one or more machine-learned models can use as input data for content organization, content management, content creation, etc.

[0113] As illustrated in FIG. 4e, the notebook application (3100) is further configured to display a fifth user interface screen (4500) in response to the selection of one or more of the source content items described in relation to FIG. 4d. The fifth user interface screen (4500) includes a first part (4510), a second part (4520), a third part (4530), and a fourth part (4540). Each part of the fifth user interface screen (4500) may correspond to a section or panel of the fourth user interface screen (4400) and may be associated with other functions.

[0114] For example, the first part (4510) corresponds to a document guide (also called a source guide) comprising a summary section (4512) and a core topic section (4514). The notebook application (3100) may be configured to generate content associated with the summary section (4512) (e.g., text description) by implementing one or more machine learning models as described in the specification in relation to FIG. 3, based on selected source content (e.g., as described in relation to FIG. 4d). For example, the summary section (4512) may provide a brief summary associated with one or more of the content items constituting the selected source content. Likewise, the notebook application (3100) may be configured to generate content associated with the core topic section (4514) (e.g., text description) by implementing one or more machine learning models as described in the specification in relation to FIG. 3, based on selected source content (e.g., as described in relation to FIG. 4d). For example, the core topic section (4514) may include one or more user interface elements that identify a theme or important topic associated with one or more of the content items constituting the selected source content. Additionally, the notebook application (3100) may be configured to generate output in response to a selection of one of the user interface elements within the core topic section (4514). The output may be, for example, a text summary or text description regarding the core topic corresponding to the selected user interface element. The output may be provided on a separate user interface screen or on another part of a fifth user interface screen configured to be generated in response to a selection of one of the user interface elements in the core topic section (4514).

[0115] For example, the second part (4520) corresponds to a source content section (e.g., a context window) containing information (4522) from at least a portion of the content items included in the source content. The notebook application (3100) may be configured to reproduce at least a portion of the content items from the source content in the second part (4520). In some implementations, the content within the source content section may correspond to a portion of the content items that were used as a basis to generate the summary section (4512).

[0116] For example, the third part (4530) corresponds to a note section (e.g., a scratchpad) that may include one or more notes that can be generated through various methods described in the present specification (e.g., notes automatically generated by the notebook application (3100), notes manually entered by the user, notes automatically generated by the notebook application (3100) in response to the selection of a user interface element corresponding to an action to be performed, etc.).

[0117] For example, the fourth part (4540) corresponds to a query section that may include one or more user interface elements for submitting or providing a query to the notebook application (3100) regarding the source content. For example, the fourth part (4540) includes a plurality of user interface elements (4542) corresponding to a proposed question or action related to the source content. For example, the notebook application (3100) may be configured to generate a proposed question or action based on information contained in the source content. For example, the notebook application (3100) may be configured to generate a proposed question or action based additionally on conversation history (e.g., previous questions or queries), user data (e.g., user preferences, user attributes, etc.), and other contextual information. The fourth part (4540) may further include a text input box (4544) that enables the user to provide input (e.g., via keyboard or voice input) to query the notebook application (3100). The fourth part (4540) may further include a user interface element (4546) indicating the number of content items constituting the source content. For example, in FIG. 4e, the user interface element (4546) indicates that three sources were used as a basis by the notebook application (3100) to generate the summary section (4512).

[0118] Referring to FIG. 4f, an exemplary user interface screen illustrates an input question and an output response related to source content. For example, in FIG. 4f, the notebook application (3100) is further configured to display a sixth user interface screen (4600) in response to receiving a query (e.g., a text query entered through the text input box (4544) in FIG. 4e). For example, the sixth user interface screen (4600) includes a first part (4610) corresponding to a conversation section, a second part (4620) corresponding to a source section, and a third part (4630) corresponding to a note section (e.g., a scratchpad).

[0119] For example, the first part (4610) includes a prompt area (4612) corresponding to a text query and a response area (4614) corresponding to a response to the text query. In some implementations, the notebook application (3100) is configured to generate a response by implementing one or more machine-learned models in response to receiving a text query as input and by referencing source content (3400). For example, when a user enters a question (e.g., 'What impact did the Cold War have on U.S. foreign policy?') through a text input box (4544) as described in relation to FIG. 4e, the notebook application (3100) may be configured to provide a sixth user interface screen (4600) and generate a response as indicated in the response area (4614). As indicated in the response area (4614), the number of references (source content items) used as a basis by one or more machine-learned models to generate the response may be indicated by the first user interface element (4616). In the example of FIG. 4f, three references were used to generate a response. The response area (4614) further includes a selectable second user interface element (4618) that, when selected, causes the response to be saved as a note in a third part (4630) corresponding to a note section (e.g., a scratchpad), and the note section may include one or more notes that can be generated through various methods described herein (e.g., notes automatically generated by the notebook application (3100) in response to the selection of the second user interface element (4618).

[0120] In some implementations, one or more parts of the response area may contain selectable information, and when such information is selected, additional information related to the selected information may be displayed. For example, in FIG. 4f, the text 'blocking policy' may be displayed in a visually distinct manner, such as by being highlighted, bolded, underlined, or otherwise, to indicate that the text is selectable (e.g., a clickable chip) and that additional information related to the text is available. The notebook application (3100) may be configured to provide additional information regarding the text in response to a selection of the text (e.g., by implementing one or more machine learning models based on the selected source content).

[0121] The second part (4620) may correspond to a source section and may include content items (4622) that constitute the source content. In some implementations, the content items (4622) may correspond to content items used as a basis by one or more machine-learned models to generate a response. In some implementations, the notebook application (3100) may be configured to dynamically modify or regenerate a response in the response area (4614) in response to receiving additional content items to be added as source content via a user interface element (4624). Additionally or alternatively, in some implementations, the notebook application (3100) may be configured to dynamically modify or regenerate a response in the response area (4614) in response to receiving the deselection of a content item from the list of content items in the second part (4620) via a user interface element (4626) (e.g., deselecting a checkbox for one or more content items in the second part (4620)).

[0122] Referring to FIG. 4g, an exemplary user interface screen includes an exemplary note section (scratch pad) for a project according to the examples of the present disclosure. For example, in FIG. 4g, the notebook application (3100) is further configured to display a seventh user interface screen (4700) in response to receiving a selection of a second user interface element (4618) (e.g., as shown in FIG. 4f). When the second user interface element (4618) is selected, the response is stored as a note (4712) in a third part (4710) corresponding to the note section (e.g., scratchpad), and the note section may include one or more notes that can be generated through various methods described herein (e.g., notes automatically generated by the notebook application (3100) in response to the selection of the second user interface element (4618). In FIG. 4g, the user interface element (4714) represents the number of content items used as a basis by one or more machine-learned models to generate a response to the note (4712). Additionally, the user interface element (4714) may be configured to be selectable, and accordingly, a list of content items (citation information) of the source content used to generate the response may be displayed in response to the selection of the user interface element (4714).

[0123] Referring to FIG. 4h, an exemplary user interface screen includes a note section (scratchpad) for a project according to the examples of the present disclosure. For example, in FIG. 4h, the notebook application (3100) is further configured to display an eighth user interface screen (4800) in response to receiving a selection of a content item (4816) from a list of content items (citation information) of source content used by one or more machine-learned models to generate a response stored in a note (4812) provided to be displayed in a first part (4810). The eighth user interface screen (4800) further includes a second part (4820) corresponding to the source section. In FIG. 4h, the notebook application (3100) is configured to display information (4822) related to the selected item (4816) in the second part (4820) in response to receiving a selection of a content item (4816) from a list (4814) of content items (citation information) of source content used by one or more machine-learned models to generate a response stored in the notebook (4812).

[0124] In some implementations, the information (4822) may include information from the content item (4816) that was used to generate the response. For example, the notebook application (3100) may be configured to refer to metadata associated with the response to refer back to the information (4822). The metadata may indicate the location of the information from the content item used to generate the response. Additionally, the information (4822) may correspond to or include a specific phrase of the content item used as the basis for generating the response. For example, the notebook application (3100) may cause a specific phrase to be displayed in the second part (4820) in a visually distinguishable manner (e.g., highlighted, bolded, enlarged font size, underlined, italicized, etc.). For example, the notebook application (3100) may be configured to display additional phrases preceding and / or following a specific phrase in the second part (4820). This additional information may provide the user with additional context regarding the information used as the basis for generating the response. For example, the notebook application (3100) may be configured not only to mark specific content items used as the basis for generating a response in the note (4812), but also to mark specific phrases of specific content items used as the basis for generating the response. Thus, the user can easily visually identify the basis for the response in the content items.

[0125] In some implementations, information (4822) from a selected item (4816) of content used to generate a response may be displayed in part or in whole. For example, if the information (4822) is below a threshold, the full text of the selected content item (4816) may be displayed in a second part (4820) and may be used by one or more machine-learned models to generate a response (e.g., a response to a text query). For example, if the information (4822) exceeds a threshold, the notebook application (3100) may be configured to implement a semantic search method to determine specific phrases related to the user query (e.g., a text query) from the full content of the selected content item (4816). In this example, rather than the full information of the content item, the related phrases are used as a basis by one or more machine-learned models to generate a response to the user query (e.g., a text query).

[0126] Examples of the present disclosure also relate to user-facing modes that enable a user to manage, organize, create, etc. content through a notebook application configured to implement one or more machine-learned models for source content selected by the user. For example, FIGS. 5a and 5b illustrate examples of actions that can be implemented for a project in which notes are generated through one or more machine-learned models based on source content selected by the user, according to one or more exemplary embodiments of the present disclosure.

[0127] For example, FIG. 5a illustrates a first user interface screen of a notebook application according to one or more exemplary embodiments of the present disclosure. For example, in FIG. 5a, the first user interface screen (5100) includes a first part (5110), a second part (5120), and a third part (5130). The first part (5110) corresponds to a note section (e.g., a scratchpad) that may include one or more notes (5112) that can be generated through various methods described herein (e.g., notes automatically generated by the notebook application (3100), notes manually entered by the user, notes automatically generated by the notebook application (3100) in response to the selection of a user interface element corresponding to an action to be performed).

[0128] The second part (5120) corresponds to a source content section (e.g., a source guide or context window) that may include one or more sources (5122) (e.g., content items constituting source content (3400) used as a basis by one or more machine-learned models to generate information contained in one or more notes (5112)). In some implementations, the notebook application (3100) may be configured to generate a note that is stored in the first part (5110) as a note based on a selection of at least some of the information from the content items provided in the second part (5120). For example, FIG. 5a illustrates selected text (5124) (e.g., highlighted text) selected by the user.

[0129] For example, the third part (5130) corresponds to a query section that may include one or more user interface elements for submitting or providing a query to the notebook application (3100) regarding the source content. For example, the third part (5130) includes a plurality of user interface elements (5132) corresponding to a proposed question or action related to the source content. For example, the notebook application (3100) may be configured to generate a proposed question or action based on information contained in the source content and / or information displayed in the second part (5120). The third part (5130) may further include a text input box (5134) that allows a user to provide input (e.g. via a keyboard or voice input) to query the notebook application (3100). The third part (5130) may further include a user interface element (5136) indicating the number of content items constituting the source content. For example, in FIG. 5a, the user interface element (5136) indicates that three sources were used as a basis by the notebook application (3100) to generate one or more notes (5112).

[0130] In some implementations, a plurality of user interface elements (5132) may be configured to change dynamically based on actions on the first user interface screen (5100). For example, a notebook application (3100) may be configured to change, modify, delete, or add user interface elements of the third part (5130) based on actions related to source content (e.g., actions related to content items provided to be displayed in the second part (5120). In FIG. 5a, the notebook application (3100) may be configured to change user interface elements of the third part (5130) dynamically based on (in response to) the selection of text from one or more sources (5122) (e.g., selected text (5124)). For example, as shown in FIG. 5a, actions may include summarizing the selected text into a note, adding a quote to the note, requesting additional information regarding the selected text (5124), or suggesting related ideas. For example, the notebook application (3100) may be configured to generate a note summarizing selected text in response to receiving a selection of a user interface element (5132a) corresponding to an action of summarizing selected text into a note. For example, the notebook application (3100) may be configured to add content corresponding to the selected text to an existing note in response to receiving a selection of a user interface element (5132b) corresponding to an action of adding a quote to the note.

[0131] For example, FIG. 5b illustrates a second user interface screen of a notebook application according to one or more exemplary embodiments of the present disclosure. For example, in FIG. 5b, the second user interface screen (5200) includes a first part (5210), a second part (5220), and a third part (5230), which may correspond to the first part (5110), the second part (5120), and the third part (5130) of FIG. 5a, respectively.

[0132] As described in relation to FIG. 5a, the notebook application (3100) may be configured to generate a note summarizing selected text in response to receiving a selection of a user interface element (5132a) corresponding to an action of summarizing selected text into a note. FIG. 5b illustrates a generated note (5214) stored in a first part (5210) comprising one or more notes (5212). Additionally, in some implementations, after the generated note (5214) is stored in the first part (5210), the plurality of user interface elements (5132) of FIG. 5a may be configured to dynamically return to a previous state corresponding to the plurality of user interface elements (5232) illustrated in FIG. 5b.

[0133] Examples of the present disclosure also relate to user-facing modes that enable a user to manage, organize, create, etc. content through a notebook application configured to implement one or more machine-learned models for source content selected by the user. For example, FIGS. 6a and 6b illustrate examples of actions that can be implemented for a project in which notes are generated through one or more machine-learned models based on source content selected by the user, according to one or more exemplary embodiments of the present disclosure.

[0134] For example, FIG. 6a illustrates a portion of a first user interface screen of a notebook application according to one or more exemplary embodiments of the present disclosure. For example, in FIG. 6a, a first portion (6110) and a second portion (6120) of the user interface screen are illustrated. The first portion (6110) corresponds to a note section (e.g., a scratchpad) that may include a plurality of notes that may have been generated through various methods described herein (e.g., notes automatically generated by the notebook application (3100), notes manually entered by the user, notes automatically generated by the notebook application (3100) in response to the selection of a user interface element corresponding to an action to be performed, etc.). For example, the first portion (6110) may indicate how a particular note was written (e.g., as a stored response, as a note written by the user, as a document generated from another note, etc.).

[0135] For example, the second part (6120) corresponds to a query section that may include one or more user interface elements for submitting or providing a query to the notebook application (3100) in relation to source content or a plurality of notes. For example, the second part (6120) includes a plurality of user interface elements (6122) that correspond to a proposed question or action related to source content or a plurality of notes. For example, the notebook application (3100) may be configured to generate a proposed question or action based on information contained in the source content and / or information displayed in the first part (6110). The second part (6120) may further include a text input box in which a user can provide input to query the notebook application (3100) (e.g., via a keyboard, voice input, etc.), a user interface element indicating the number of content items constituting the source content, etc.

[0136] In the example of FIG. 6a, the notebook application (3100) may be configured to dynamically change user interface elements of the second part (6120) based on (in response to) the selection of one or more notes (6112) among the plurality of notes provided in the first part (6110). For example, as shown in FIG. 6a, one or more notes may be selected through user input (e.g., through drag input or by selecting a checkbox), and in response to the selection of one or more notes, actions may include actions to create content based on the selected notes (6114) (e.g., creating a study guide, creating an outline, creating a spreadsheet, creating a presentation, etc.), actions to suggest related ideas, etc. For example, the notebook application (3100) may be configured to create a note corresponding to the content outline of the selected note (6114) in response to receiving a selection of a user interface element (6122a) corresponding to an action of creating an outline and saving the outline to a note. For example, a notebook application (3100) may be configured to implement one or more machine learning models to generate a note corresponding to a selected note (6114) in response to receiving a selection of a user interface element corresponding to an action of creating content in relation to one or more selected notes and saving the content as a note.

[0137] For example, FIG. 6b illustrates a portion of a second user interface screen of a notebook application according to one or more exemplary embodiments of the present disclosure. For example, in FIG. 6b, the first portion (6210) may correspond to the first portion (6110) of FIG. 6a.

[0138] As described in relation to FIG. 6a, the notebook application (3100) may be configured to generate a note based on one or more selected notes by implementing one or more machine-learned models, wherein the selected note may correspond to source content (e.g., source content selected by a user and used as input for generating a note). For example, the generated note may summarize the selected note or present an overview thereof, as described in relation to FIG. 6a. The notebook application (3100) may be configured to generate a generated note (6214) based on the selected note (6114) in response to receiving a selection of a user interface element (e.g., user interface element (6122a)) corresponding to an action of summarizing the selected note (6114) into a generated note (6214), by implementing one or more machine-learned models. In this case, the selected note (6114) may correspond to source content (e.g., source content selected by a user and used as input for generating a note). FIG. 6b illustrates a generated note (6214) stored in a first part (6210) containing one or more other notes (6212). Additionally, in some implementations, after the generated note (6214) is stored in the first part (6210), a plurality of user interface elements (6122) of FIG. 6a may be configured to dynamically return to a previous state.

[0139] In some implementations, the notebook application (3100) may be configured to allow the generated note (6214) to be exported to another application (e.g., a word processing application, a presentation application, a spreadsheet application, a social media application, etc.) through the selection of a user interface element for sending the document to another application. In some implementations, the notebook application (3100) may be configured to allow the generated note (6214) and / or content item (e.g., source content (3400)) to be shared with another user through the selection of a user interface element for sharing the document and / or source content with another user.

[0140] According to an example of the present disclosure, a notebook application (3100) may be configured to generate output (e.g., an overview, a report, a summary, etc.) through one or more machine learning models based on source content provided to the notebook application (e.g., source content provided by a user). The notebook application (3100) may be configured to allow a user to create various projects to complete various tasks. Each project may be configured to function in a manner similar to a folder where the user can store various information in each project. In some implementations, individual scratchpads may correspond to a specific project or be used exclusively for a specific project. In some implementations, the notebook application (3100) may be configured to receive source content specified by a user. The notebook application (3100) may be configured to add, delete, or modify projects based on input received from a user. Each project may be provided with a name generated by the notebook application (3100), a name provided by a user, or a default name based on information stored in the project (e.g., based on source content) (e.g., through one or more machine learning models).

[0141] Examples of the present disclosure also relate to user-facing modes that enable a user to manage, organize, create, etc. content through a notebook application configured to implement one or more machine-learned models for source content selected by the user. For example, FIG. 7 illustrates an example of a notebook or project that may be represented in a specific way so that the user can easily identify the content included in the notebook or project.

[0142] In some implementations, the notebook application (3100) may be configured to automatically generate a graphic image (e.g., emoji, icon, etc.) or graphic animation corresponding to or representing the source content in response to source content being provided to the notebook application (3100) (e.g., using one or more machine learning models, one or more generative machine learning models, semantic retrieval techniques, etc.). In some implementations, the graphic image or graphic animation may be overlaid on the folder provided as a user interface element that causes the folder to open and the folder's contents to be displayed to the user when selected. Additionally or alternatively, in some implementations, the notebook application (3100) may be configured to automatically generate a text description (name) corresponding to or representing the source content in response to source content being provided to the notebook application (3100) (e.g., using one or more machine learning models, one or more generative machine learning models, semantic retrieval techniques, etc.). The text description may be overlaid on the folder provided as a user interface element that causes the folder to open and the folder's contents to be displayed to the user when selected.

[0143] Referring to FIG. 7, the notebook application (3100) may have a user-only section (7100) for storing various projects in specific folders. For example, a first folder (7110) (e.g., a default folder) may be represented by a default image (7112) and may have a common name (7114) (e.g., 'default notebook'). For example, a second folder (7120) may be represented by a graphic image (7122), which is generated through machine learning and may have a text description (7124) (e.g., 'profit') that indicates the content contained in the second folder (7120). For example, the notebook application (3100) may be configured to automatically generate a graphic image (7122) corresponding to or representing the source content, such as an emoji, icon, etc., in response to source content being provided to the notebook application (3100) (e.g., using one or more machine learning models, one or more generative machine learning models, semantic search techniques, etc.). In some implementations, the graphic image (7122) may be overlaid on a second folder (7120) provided as a user interface element, and when the second folder (7120) is selected, the second folder (7120) is opened and its contents are displayed to the user. Additionally or alternatively, in some implementations, the notebook application (3100) may be configured to automatically generate a text description (7124) (name) corresponding to or representing the source content in response to source content being provided to the notebook application (3100) (e.g., using one or more machine learning models, one or more generative machine learning models, semantic search techniques, etc.). The text description (7124) can be overlaid on the second folder (7120) provided as a user interface element, and when the second folder (7120) is selected, the second folder (7120) is opened and its contents are displayed to the user.

[0144] Examples of the present disclosure relate to a computer implementation method for generating a document template and providing a user interface for organizing, managing, and creating content by implementing one or more machine learning models in relation to source content selected by a user and the generated document template. FIG. 8a illustrates a flowchart of an exemplary and non-limiting computer implementation method according to one or more exemplary embodiments of the present disclosure. FIG. 8b illustrates another flowchart of an exemplary and non-limiting computer implementation method according to one or more exemplary embodiments of the present disclosure. FIG. 9 illustrates a block diagram of a document extractor application according to one or more exemplary embodiments of the present disclosure.

[0145] The flowchart of FIG. 8a illustrates a method (8000) for generating a document template that can be used to organize, manage, and create content by implementing one or more machine-learned models in relation to training content selected by a user (e.g., sample documents, training documents, etc.). Although illustrated in a specific sequence or order, the order of the processes may be modified unless otherwise specified. Accordingly, the illustrated embodiments should be understood merely as examples, and the illustrated processes may be performed in a different order, and some processes may be performed in parallel. Additionally, one or more processes may be omitted in various embodiments. Accordingly, not all processes are required in all embodiments. Other process flows are also possible.

[0146] Referring to FIG. 8a, in operation (8100), the method (8000) comprises a computing device receiving input from a user regarding the selection of a sample document (e.g., training content, training document, etc.). As described herein, the computing device may be implemented as a computing device (100), a server computing system (300), or a combination thereof. For example, the input may be provided by the user through an input device (150). For example, the input may be provided by selecting a specific file or document uploaded to the computing device to be used by a document extractor application (136). In some implementations, the content may be uploaded from local memory, another application (e.g., a PDF application), copied text, or a website. The selected file, text, document, etc. may be referred to as training content, training document, sample document, etc. In some implementations, the training content may be a subset of a larger content corpus. For example, input can be provided to or entered into a document extractor application (136) or a document extractor application (336).

[0147] In some implementations, the response to an input selecting source content may be processed at the computing device (100) without involving the server computing system (300). In some implementations, an input selecting training content may be transmitted from the computing device (100) to the server computing system (300), and at least a portion of the response to the input may be processed by the server computing system (300). For example, an input related to the selection of training content may be provided at the computing device (100), and the server computing system (300) may be configured to perform an action in response to receiving an indication of the input.

[0148] In operation (8200), the method (8000) includes the computing device receiving input from a user requesting the creation of a document template in connection with the selection of a sample document (e.g., training content, training document, etc.). For example, the input may be provided by the user through an input device (150). In some implementations, the computing device (e.g., a document extractor application (136)) may be configured to provide a graphical user interface on a display device that allows the user to request the creation of a document template. For example, the input may be provided by selecting a user interface element associated with creating a document template. For example, the input may be provided to or entered into the document extractor application (136) or the document extractor application (336).

[0149] In operation (8300), the method (8000) comprises a computing device implementing one or more machine-learned models in relation to selected training content (sample document, training document, etc.) to generate a document template. In some implementations, the document template generated by the one or more machine-learned models may include multiple sections, multiple headings representing different sections of the document, etc. In some implementations, the document template may be further associated with a specific style, intent, and / or format that can be inferred or scraped from the content of the training content.

[0150] For example, a computing device may obtain information indicating that a user has selected training content. The computing device may process the training content using one or more machine learning models (e.g., one or more large-scale language models) to obtain language output. Then, the computing device may generate summary output using one or more machine learning models (e.g., one or more large-scale language models, one or more generative machine learning models, etc.). In particular, the machine learning large-scale language models may be trained to process various outputs to generate language output. For example, the machine learning large-scale language models may process embeddings generated by a machine learning embedding generative model, parts of training content identified using an embedding generative model, language output generated using the machine learning large-scale language models or some other models, etc.

[0151] In some implementations, one or more machine-learned models may be configured to determine (learn) the intent, style, and / or format of training content, for example, through various natural language processing operations. For example, training content may be broken down into tokens (e.g., words, phrases, individual characters, etc.) and converted into embeddings (e.g., numerical vector representations) capable of capturing semantic information regarding the training content. In some implementations, each training document among multiple training documents may be classified into a specific type of document (e.g., resume, PRD, outline, legal opinion, etc.). Training documents may be classified based on the aggregation of token embeddings to generate a representation of the entire training document (e.g., through embedding averaging, TF-IDF weighting, etc.). One or more machine-learned models may be configured to analyze the vocabulary of the training documents (e.g., based on the frequency of specific words, the presence of specific terms, the use of domain-specific terminology, etc.). One or more machine learning models may be further configured to determine the syntax of a training document that can provide information regarding the specific style and / or intent of the document (e.g., based on sentence structure, sentence length, use of grammatical syntax, etc.). One or more machine learning models may be configured to determine semantic information based on the meaning of the content (e.g., the meaning of specific sentences and paragraphs of the document) and to analyze the training content to identify the specific style and / or intent of the training document. Additionally, one or more machine learning models may be configured to determine the context of each word in relation to the entire training document.

[0152] To determine (learn) the format of the training content, one or more machine learning models may be configured to analyze the sequential structure of the training content to identify recurring patterns (e.g., recognizing headings, subheadings, paragraphs, bullet points, numbered lists, and other common formatting elements). One or more machine learning models may also be configured to learn document structure based on consistent patterns or layouts (e.g., tables, images, captions, etc.) to understand the spatial relationships between different elements. One or more machine learning models may also be configured to identify specific formats (e.g., the use of indentation, font style, font size, etc.) to identify document structure and perform pattern matching. One or more machine learning models may be configured to learn and recognize specific document formats so that when a user uploads multiple training documents, a document template following the document structure of the training documents can be generated. In some implementations, the training content may share common features (e.g., common format, common style, common intent, etc.).

[0153] In some implementations, one or more machine-learned models may be configured to receive from a user information identifying information about training content (e.g., labeled data such as identifying information about document type, document style, document intent, document format, etc.). One or more machine-learned models may be trained and / or improved based on the labeled data as well as feedback provided by the user. Document templates generated by one or more machine-learned models (e.g., one or more large-scale language models, one or more generative machine-learned models, etc.) in operation (8300) may be stored in a computing device and may be output to be displayed to a user on a display device (160).

[0154] The flowchart of FIG. 8b illustrates a method (8400) for generating an output document based on a document template generated according to the flowchart of FIG. 8a by implementing one or more machine-learned models in relation to source content (e.g., source document) selected by a user. Although illustrated in a specific sequence or order, the order of the processes may be modified unless otherwise specified. Accordingly, the illustrated embodiments should be understood merely as examples, and the illustrated processes may be performed in a different order, and some processes may be performed in parallel. Additionally, one or more processes may be omitted in various embodiments. Accordingly, not all processes are required in all embodiments. Other process flows are also possible.

[0155] Referring to FIG. 8b, in operation (8500), the method (8400) comprises the computing device receiving input from a user regarding the selection of a source document (e.g., source document). As described herein, the computing device may be implemented as a computing device (100), a server computing system (300), or a combination thereof. For example, the input may be provided by the user through an input device (150). For example, the input may be provided by selecting a specific file or document uploaded to the computing device to be used by a document extractor application (136). In some implementations, the content may be uploaded from local memory, another application (e.g., a PDF application), copied text, or a website. The selected file, text, document, etc. may be referred to as source content, source document, etc. In some implementations, the source content may be a subset of a larger content corpus. For example, the input may be provided to or input into the document extractor application (136) or the document extractor application (336).

[0156] In some implementations, the response to an input selecting source content may be processed at the computing device (100) without involving the server computing system (300). In some implementations, the input selecting source content may be transmitted from the computing device (100) to the server computing system (300), and at least a portion of the response to the input may be processed by the server computing system (300). For example, the input related to the selection of source content may be provided at the computing device (100), and the server computing system (300) may be configured to perform an action (e.g., the creation of an output document) in response to receiving information representing said input.

[0157] In operation (8600), the method (8400) comprises receiving input from a user requesting the creation of an output document in relation to a specific document template that may be selected via user input and a source document (e.g., source content). For example, the input may be provided by the user via an input device (150). In some implementations, the computing device (e.g., a document extractor application (136)) may be configured to provide a graphical user interface on a display device that allows the user to request the creation of an output document in relation to a specific document template that may be selected via user input. For example, the input may be provided by selecting a desired document template and selecting a user interface element associated with creating an output document. In some implementations, input for creating an output document may be provided, and the document extractor application (136) or the document extractor application (336) may be configured to determine a suitable document template that may be applied to the source document to create an output document. For example, input can be provided to or entered into a document extractor application (136) or a document extractor application (336).

[0158] In operation (8700), the method (8400) comprises a computing device implementing one or more machine learning models (e.g., one or more large language models, one or more generative machine learning models, etc.) in relation to a source document (source content) selected for generating an output document and an identified document template. In some implementations, the output document generated by the one or more machine learning models may include multiple sections corresponding to the identified or selected document template, multiple headings representing different sections of the document, etc. In some implementations, the output document may be additionally associated with a specific style, intent, and / or format based on the style, intent, and / or format of the document template.

[0159] For example, a computing device may obtain information indicating that a user has selected a source document. The computing device may process the source document using one or more machine learning models (e.g., one or more large language models) to obtain language output. Then, the computing device may generate summary output using one or more machine learning models (e.g., one or more large language models, one or more generative machine learning models, etc.). In particular, the machine learning large language models may be trained to process various outputs to generate language output. For example, the machine learning large language models may process embeddings generated by a machine learning embedding generative model, parts of the source document identified using an embedding generative model, language output generated using the machine learning large language models or some other models, etc.

[0160] In some implementations, one or more machine-learned models may be configured to determine (learn) the intent, style, and / or format of source documents, for example, through various natural language processing operations. For example, source documents may be broken down into tokens (e.g., words, phrases, individual characters, etc.) and converted into embeddings (e.g., numerical vector representations) capable of capturing semantic information about the source documents. In some implementations, each source document among multiple source documents may be classified into a specific document type (e.g., resume, PRD, outline, legal opinion, etc.). Source documents may be classified based on the aggregation of token embeddings to generate representations for all training documents (e.g., through embedding averaging, TF-IDF weighting, etc.). One or more machine-learned models may be configured to analyze the vocabulary of source documents (e.g., based on the frequency of specific words, the presence of specific terms, the use of domain-specific terminology, etc.). One or more machine learning models may be further configured to determine the syntax of a source document that can provide information regarding the specific style and / or intent of the source document (e.g., based on sentence structure, sentence length, use of grammatical syntax, etc.). One or more machine learning models may be configured to analyze the source document to determine semantic information based on the meaning of the content (e.g., the meaning of specific sentences and paragraphs of the document) and to identify the specific style and / or intent of the source document. Additionally, one or more machine learning models may be configured to determine the context of each word in relation to the entire source document.

[0161] One or more machine learning models (e.g., one or more large-scale language models, one or more generative machine learning models, etc.) may be configured to apply document templates to multiple source documents to generate an output document. For example, one or more machine learning models may be configured to extract first content (e.g., background information, title information, body information, conclusion information, etc.) from multiple source documents and associate the first content with a first section of the output document (e.g., background section, title section, body section, conclusion section, etc.). Similarly, one or more machine learning models may be configured to extract second content from multiple source documents and associate the second content with a second section of the output document, etc. One or more machine learning models (e.g., one or more large-scale language models, one or more generative machine learning models, etc.) may be configured to associate content from source documents with a specific section of the output document based on determined semantic information, context information, intent information, etc. associated with the content. For example, if one or more machine learning models determine that a specific part of a source document is associated with background information regarding a specific topic or target (e.g., based on confidence levels), one or more machine learning models may be configured to implement part or all of the specific part in the background section of the output document.

[0162] For example, one or more machine learning models may be configured to apply the style, intent, and / or format of a document template to content extracted from multiple source documents. For example, if a document template is associated with persuasive intent and an opinion-presenting style, one or more machine learning models may be configured to generate an output document having these characteristics based on content from multiple source documents, wherein the output document may have a document structure defined by or associated with the document template.

[0163] As another example, one or more machine learning models may be configured to identify document types associated with multiple source documents based on the content of each of multiple source documents. One or more machine learning models may be configured to apply the style, intent, and / or format of document templates associated with the identified document types. For example, if one or more machine learning models determine that the document type associated with a source document is PRD, one or more machine learning models may be configured to apply the style, intent, and / or format of document templates associated with the PRD document type to the content from multiple source documents.

[0164] In some implementations, one or more machine learning models may be configured to receive from a user information identifying information regarding the output document and / or source document (e.g., labeled data such as identifying information about the document type, document style, document intent, document format, etc.). One or more machine learning models for generating the output document may be trained and / or improved based on the labeled data as well as feedback provided by the user. For example, the output document generated in operation (8700) may be stored in a computing device and may be output to be displayed to the user on a display device (160).

[0165] Referring to FIG. 9, a document extractor application (9100) (which may correspond to a document extractor application (136) and / or a document extractor application (336)) may include a conditioning parameter generator (9110), one or more sequence processing models (9120), one or more large-scale language models (9130), and one or more generative machine learning models (9140). The document extractor application (9100) may receive input (9200) from a user as discussed above in relation to the operations (8100, 8200, 8500, 8600) of FIG. 8a and 8b. A conditioning parameter generator (9110) may be configured to generate a conditioning parameter based at least partially on an input, and the conditioning parameter provides a value for one or more conditions associated with the content to be generated, at least partially associated with the input (9200) and the training content (9400) and / or source content (9500) selected by the user.

[0166] For example, training content (9400) and source content (9500) may include any type of document (e.g., a document in digital form), and may include books, product manuals, legal opinions, academic papers, proprietary data files, patent documents, web pages, emails, forum posts, social media posts, videos, images, geographic information, or any other type or form of content that can be stored or accessed in digital form (e.g., content stored in a database, memory device, etc.). In some implementations, training content (9400) and source content (9500) may be stored in the content data store (350) by the user selecting specific documents, images, or other content to be stored in the content data store (350). In some implementations, training content (9400) and source content (9500) may be stored in a computing device (100) and / or a server computing system (300).

[0167] To generate a conditioning parameter, the conditioning parameter generator (9110) may be configured to retrieve values ​​for one or more conditions associated with the input. For example, to generate a conditioning parameter, the conditioning parameter generator (9110) may be configured to extract values ​​for one or more conditions from the input. The input may contain information indicating the user's intent or requirements. In some implementations, the conditioning parameter generator (9110) (or one or more sequence processing models (9120) or one or more large language models (9130)) may be configured to extract information from the input (9200) to identify values ​​for one or more conditions, and the conditioning parameter generator (9110) may be configured to generate a conditioning parameter based on the extracted values. For example, the input itself can identify a color to be used for the heading of the generated document (e.g., 'blue font for heading') or an attribute or feature (e.g., 'circular bullet points') that can be used to generate a document template related to training content (9400) or to generate a conditioning parameter for generating an output document related to source content (9500).

[0168] To generate a conditioning parameter, the conditioning parameter generator (9110) may be configured to infer a value for one or more conditions from the input. The input may contain information indicating a user's intent or requirement. In some implementations, the conditioning parameter generator (9110) (or one or more sequence processing models (9120) or one or more large language models (9130)) may be configured to infer information from the input (9200) to identify a value for one or more conditions, and the conditioning parameter generator (9110) may be configured to generate a conditioning parameter based on the inferred value. For example, the input may include a reference to the length ('short', 'long', etc.) of the output document to be generated based on the source content (9500), and the conditioning parameter generator (9110) (or one or more sequence processing models (9120) or one or more large language models (9130)) may be configured to infer a value based on the input. For example, a document extractor application (9100) can infer a value of about 1 page from an input requesting the generation of a 'standard' resume, while a 'long' resume can be associated with a value of about 2 to 3 pages. For example, the document extractor application (9100) can be configured to verify the inferred value based on information obtained through external content (9300) (e.g., a website describing the length of a resume).

[0169] In some implementations, the conditioning parameter generator (9110) may be configured to infer values ​​for one or more conditions from the input by providing input to one or more sequence processing models (9120), wherein one or more sequence processing models (9120) are configured to respond to a query or output values ​​for one or more conditions based on a query. One or more sequence processing models (9120) may include one or more machine learning models configured to process and analyze sequential data and to process data occurring in a specific order or sequence, including time-series data, natural language text, or any other data having a temporal or sequential structure.

[0170] One or more sequence processing models (9120) receive an input containing text and can tokenize the input by breaking down the sequence of text into small units (tokens) to provide a structured representation of the input sequence. One or more sequence processing models (9120) can represent the tokens as vectors in a continuous vector space by mapping each token to a high-dimensional vector, where the relationship between the tokens (words) is reflected in the geometric relationship between the corresponding vectors. For example, one or more sequence processing models (9120) receive an input extracted from training content (9400) and / or source content (9500) containing the text 'that busy market' and can provide a structured representation of the input sequence by tokenizing the input by dividing the text sequence into small units (tokens) (e.g., 'that', 'busy', and 'market'). In word embeddings, semantically similar words are closer to each other in the vector space. For example, vectors for 'busy' and 'busy' can be close to each other due to their semantic relationship, whereas vectors for 'busy' and 'stagnant' can be far apart from each other compared to the vectors for 'busy' and 'busy'.

[0171] One or more large language models (9130) may be models trained on a large corpus of language training data to enable the one or more large language models (9130) to perform multiple language tasks, or otherwise may include such models. For example, one or more large language models (9130) may be trained to perform summarizing tasks, conversation tasks, simplification tasks, opposing viewpoint tasks, etc. In particular, one or more large language models (9130) may be trained to generate language output by processing various outputs. For example, one or more large language models (9130) may process embeddings generated by a machine learning embedding generation model, source content or parts of training content identified using an embedding generation model (e.g., document chunks), language output generated using one or more large language models (9130) or some other model, etc.

[0172] One or more generative machine learning models (9140) may include a deep neural network or a generative adversarial network (GAN), a variational autoencoder, a stable diffusion machine learning model, a visual transformer, a neural radiation field (NeRF), etc., to generate content (e.g., a resume, an outline, a PRD, etc.) having a value of a condition associated with one or more features. For example, a computing device may include a database (e.g., a machine learning model data repository (370)) configured to store a plurality of generative machine learning models associated with a plurality of different content types or a plurality of different document types (e.g., different genres or topics, different kinds of content including images, videos, and text, different types of content including outlines, reports, spreadsheets, resumes, product requirements documents (PRD), etc.).

[0173] In some implementations, the computing device may be configured to search among one or more generative machine learning models (9140) for a generative machine learning model associated with a specific type of content (document) and / or document template to generate an output document associated with an input. In some implementations, the computing device may be configured to search among one or more generative machine learning models (9140) for a generative machine learning model associated with a specific type of content (document) to generate a document template associated with a specific type of input.

[0174] In some implementations, one or more generative machine learning models (9140) may be trained on a large dataset of content having corresponding information on conditions associated with the content (e.g., a large corpus of language training data). During training, one or more generative machine learning models (9140) may be configured to learn the relationships between elements within the output (e.g., content) and the conditions affecting them. This may involve a computing device adjusting the internal parameters of each generative machine learning model to generate realistic or accurate content (e.g., grammatically correct content, consistent content, etc.) based on the training data. One or more generative machine learning models (9140) may be trained on one or more training datasets containing multiple reference document templates. One or more generative machine learning models (9140) may be trained on one or more training datasets containing multiple reference output documents associated with one or more document templates. One or more training datasets may contain values ​​for one or more conditions.

[0175] In some implementations, one or more generative machine learning models (9140) are configured to generate a document template (9700) in response to receiving a selection for training content (9400). For example, the document template (9700) may be generated based on conditioning parameters (and corresponding values ​​for one or more conditions) to make a decision to generate the content of the document template (9700). In some implementations, one or more generative machine learning models (9140) are configured to generate an output document (9800) in response to receiving a selection for source content (9500) and based on a specific document template (9700) that may be selected by a user or automatically determined based on the source content (9500). For example, the output document (9800) may be generated based on conditioning parameters (and corresponding values ​​for one or more conditions) to make a decision to generate the content of the output document (9800).

[0176] In some implementations, the server computing system (300) may provide (transmit) content or part of the generated content to the computing device (100), or the server computing system (300) may provide access to the generated content to the computing device (100). For example, a document template (9700) and / or output document (9800) may be generated in the server computing system (300) and stored in one or more computing devices (e.g., one or more of the computing device (100), external computing device (200), server computing system (300), external content (500), content data store (350), user data store (360), etc.).

[0177] In some implementations, after a document template (9700) is generated, the user may provide feedback or additional input regarding the generated document template (9700) based on the provided training content (9400) (and / or based on a query provided by the user), and one or more of the actions (8100 to 8300) may be repeated. In some implementations, after an output document (9800) is generated, the user may provide feedback or additional input regarding the generated output document (9800) based on the provided source content (9500) (and / or based on a query provided by the user), and one or more of the actions (8500 to 8700) may be repeated.

[0178] Examples of the present disclosure also relate to user-facing modes that enable a user to manage, organize, create, etc. content through a notebook application and / or a document extractor application, each configured to implement one or more machine-learned models in relation to content selected by the user. For example, FIGS. 10a through 10f illustrate examples of operations that can be implemented for a project in which, according to one or more exemplary embodiments of the present disclosure, a document template is generated through one or more machine-learned models based on training content selected by the user, and an output document is generated through one or more machine-learned models based on source content selected by the user.

[0179] For example, FIG. 10a illustrates a first user interface screen (e.g., a template builder user interface screen) of a document extractor application (which may be integrated as part of a notebook application) according to one or more exemplary embodiments of the present disclosure.

[0180] In FIG. 10a, a first user interface screen (1010) illustrates a user interface (e.g., a template builder user interface screen) that provides information regarding a document extractor application (9100). In particular, the document extractor application (9100) is configured to present the first user interface screen (1010) for display, which includes a visual representation (1012) regarding how a document template can be generated and structured based on training content, and information (1014) regarding features of the document extractor application (9100).

[0181] As illustrated in FIG. 10a, the first user interface screen (1010) further includes a plurality of user interface elements that the user can select (or interact with) to create a document template. For example, a first portion of the first user interface screen (1010) includes a plurality of first user interface elements (1016) configured to be selectable as training content for creating a document template. In FIG. 10a, the training content that the user can select to create a document template includes a PRD file ('Product PRD'), meeting notes ('Brainstorming Meeting'), and notes on strategy ('Notebook LLM Strategy').

[0182] For example, a second portion of the first user interface screen (1010) includes a second user interface element (1018) configured to allow the user to upload one or more training documents for creating a document template. As illustrated in FIG. 10a, a third user interface element (1019) is configured to receive input associated with creating a document template based on training content (e.g., training documents) identified (selected) by the user through the first user interface screen (1010). For example, a document extractor application (9100) may be configured to create a document template according to the examples described herein in response to user input (e.g., selecting the third user interface element (1019)) based on the training content selected by the user.

[0183] FIG. 10b illustrates a second user interface screen (e.g., a document template customization user interface screen) of a document extractor application (which may be integrated as part of a notebook application) according to one or more exemplary embodiments of the present disclosure.

[0184] In FIG. 10b, a second user interface screen (1020) illustrates a user interface (e.g., a template builder user interface screen) associated with enabling a user to customize one or more features associated with a document template generated by the document extractor application (9100). In particular, the document extractor application (9100) is configured to present for display the second user interface screen (1020), which includes various parts and user interface elements that enable a user to modify or customize a document template generated by the document extractor application (9100).

[0185] For example, the second user interface screen (1020) includes a first part (1021) that identifies the name of the generated document template ('PRD template').

[0186] For example, the second user interface screen (1020) includes a second part (1022) associated with a training document selected to generate a document template. For example, the training document (1023) includes a 'Product PRD' training document and a 'Brainstorming Meeting' document. The first user interface element (1024) may be configured to allow the user to add a training document so that the document template can be regenerated based on the training document to which the document template was added.

[0187] For example, the second user interface screen (1020) includes a third part (1025) associated with the style of the generated document template. For example, the third part (1025) of the second user interface screen (1020) includes a plurality of first user interface elements (1026) associated with different possible styles that can be selected to generate (or regenerate) the document template. In FIG. 10b, exemplary styles that can be selected include 'professional style tone', 'everyday language style', 'use of metaphors', 'opinion-based style', and 'MBTI type: INTJ'. The second user interface element (1027) may be configured to allow the user to add new styles so that the document template can be regenerated based on the added styles.

[0188] For example, the second user interface screen (1020) includes a fourth part (1028) associated with the document format of the generated document template. For example, the fourth part (1028) of the second user interface screen (1020) includes a plurality of second user interface elements (1029) associated with different sections of the document structure that can be selected and modified to generate (or regenerate) the document template. In FIG. 10b, exemplary document sections shown in the drawing and selectable include a 'Title' section and a 'Description' section. Other document sections may include, for example, a 'Background' section and a 'Conclusion' section. Each of the plurality of second user interface elements (1029) may be configured to be manipulated by the user so that different sections can be rearranged according to user input, and the document template can be regenerated based on the rearranged document structure (format).

[0189] Although not illustrated in FIG. 10b, the second user interface screen (1020) may also include additional parts associated with the intent of the generated document template. For example, the additional parts of the second user interface screen (1020) may include multiple user interface elements associated with different possible intents that can be selected to generate (or regenerate) the document template. Exemplary intents that can be selected include a 'persuasion' intent, an 'information' intent, an 'entertainment' intent, an 'inspiration' intent, etc.

[0190] FIG. 10c visually illustrates a method in which a document extractor application (which may be integrated as part of a notebook application) can learn the format, style, and / or intent of a training document according to one or more exemplary embodiments of the present disclosure. As illustrated in FIG. 10c, a training document (1032) ('Product PRD') includes a plurality of sections (1034) of a PRD document, including an introduction section, a core user experience (CUJ) section, a target audience section, a problem statement section, and a proposal section. The document extractor application (9100) may be configured to learn the document type of the training document (1032) through a selection of user interface elements (1036). As illustrated in FIG. 10c, the learned document section (1038) for the PRD document includes the same section as the training document (1032). The document extractor application (9100) may be configured to generate a PRD in response to a user uploading a source document, and the generated PRD may have the format shown in FIG. 10c or a similar format based on learned document type information obtained from the training document (1032).

[0191] For example, FIG. 10d illustrates a fourth user interface screen (e.g., an output document generation user interface screen) of a document extractor application (which may be integrated as part of a notebook application) according to one or more exemplary embodiments of the present disclosure.

[0192] In FIG. 10d, a fourth user interface screen (1040) illustrates a user interface (e.g., an output document creation user interface screen) that enables a user to create an output document based on one or more source documents that the user can select, and a document template that can likewise be selected or provided to the user based on, for example, the content of the source documents. For example, in FIG. 10d, the user has selected a plurality of source documents (1042) (which may correspond to notes added to the scratchpad or note section of the notebook application (3100)).

[0193] In some implementations, the user may identify or select a specific document template and provide input that causes the document extractor application (9100) to generate an output document based on the selected source document and the selected document template. In some implementations, the document extractor application (9100) may be configured to analyze the content of the selected source document and determine or suggest one or more document templates suitable or applicable to the source document.

[0194] For example, in FIG. 10d, the fourth user interface screen (1040) includes a first user interface element (1044) configured to generate a PRD based on a document template previously created and associated with a PRD document when selected.

[0195] For example, FIG. 10e illustrates a fifth user interface screen (e.g., output document user interface screen) of a document extractor application (which may be integrated as part of a notebook application) according to one or more exemplary embodiments of the present disclosure.

[0196] In FIG. 10e, the fifth user interface screen (1050) illustrates a user interface (e.g., output document user interface screen) comprising an output document generated by the document extractor application (9100) according to a selected source document and document template (e.g., PRD document template). For example, the output document may correspond to a note (1052) for a notebook application (3100). As illustrated in FIG. 10e, the output document may have a document structure or format that matches the format of the PRD document and may include similar sections such as an introduction section (1054) and a CUJ section (1056). The document extractor application (9100) may be configured to generate the content of each section based on the content of the source document through one or more machine-learned models, as described according to the examples provided herein. For example, the output document may be used in a manner such as being stored in a computing device, transmitted to another computing device, saved as a specific document file type in another document application (e.g., via the first user interface element (1058)), or shared with other users (e.g., via the second user interface element (1059)).

[0197] FIG. 11a illustrates a block diagram of an exemplary computing system for organizing, managing, and creating content by implementing one or more machine-learned models in relation to source content and / or training content selected by a user, according to one or more exemplary embodiments of the present disclosure. The system (1100) includes a user computing device (1102), a server computing system (1130), and a training computing system (1150) that are coupled to communicate via a network (1180).

[0198] FIG. 11b illustrates a block diagram of an exemplary computing device for organizing, managing, and creating content by implementing one or more machine-learned models in relation to source content and / or training content selected by a user, according to one or more exemplary embodiments of the present disclosure.

[0199] FIG. 11c illustrates a block diagram of an exemplary computing device for organizing, managing, and creating content by implementing one or more machine-learned models in relation to source content and / or training content selected by a user, according to one or more exemplary embodiments of the present disclosure.

[0200] The user computing device (1102) (which may correspond to the computing device (100)) may be any type of computing device, such as a personal computing device (e.g., a laptop or desktop), a mobile computing device (e.g., a smartphone or tablet), a game console or controller, a wearable computing device, an embedded computing device, or any other type of computing device.

[0201] The user computing device (1102) includes one or more processors (1112) and memory (1114). The one or more processors (1112) may be any suitable processing device (e.g., processor core, microprocessor, ASIC, FPGA, controller, microcontroller, etc.) and may be a single processor or a plurality of processors operably connected. The memory (1114) may include one or more non-transient computer-readable storage media such as RAM, ROM, EEPROM, EPROM, flash memory device, magnetic disk, etc. and combinations thereof. The memory (1114) may store data (1116) and instructions (1118) executed by the processor (1112) to enable the user computing device (1102) to perform operations.

[0202] In some implementations, the user computing device (1102) may store or contain one or more machine-learned models (1120) (e.g., large-scale language models, sequence processing models, generative machine-learned models, etc.). For example, one or more machine-learned models (1120) may be or otherwise contain various machine-learned models, such as neural networks (e.g., deep neural networks) or other types of machine-learned models including non-linear models and / or linear models. Exemplary neural networks may include feedforward neural networks, recurrent neural networks (RNNs) including long short-term memory (LSTM)-based recurrent neural networks, convolutional neural networks (CNNs), diffusion models, generative adversarial networks, or other forms of neural networks. Exemplary neural networks may be deep neural networks. Some exemplary machine-learned models may utilize attention mechanisms such as self-attention. For example, some exemplary machine-learned models may include multi-head self-attention models (e.g., Transformer models). An exemplary machine-learned model has been described herein with reference to FIGS. 1a through 10e.

[0203] In some implementations, one or more machine-learned models (1120) may be received from a server computing system (1130) via a network (1180), stored in memory (1114), and then used by one or more processors (1112) or otherwise implemented. In some implementations, a user computing device (1102) may implement multiple parallel instances of a single machine-learned model (e.g., to perform a task in parallel across multiple instances of a machine-learned model). In some implementations, the task is a generative task, and one or more machine-learned models may be implemented to output content (e.g., document template, output document, etc.) considering various inputs (e.g., query, training document, source document, conditioning parameter, etc.). More specifically, machine-learned models disclosed herein (e.g., large-scale language models, sequence processing models, generative machine-learned models, etc.) may be implemented to perform various tasks related to input queries.

[0204] According to examples of the present disclosure, a computing system may implement one or more sequence processing models (3120, 9120) described herein to respond to a query or output a value for one or more conditions based on a query. One or more sequence processing models (3120, 9120) may include one or more machine-learned models configured to process and analyze sequential data and to process data occurring in a specific order or sequence, including time-series data, natural language text, or any other data having a temporal or sequential structure.

[0205] According to examples of the present disclosure, a computing system may implement one or more large-scale language models (3130, 9130) to determine a plurality of variables based on a query. For example, the large-scale language model may include a BERT large-scale language model. The large-scale language model may be trained to understand and process natural language, for example. The large-scale language model may be configured to identify keywords, intents, and contexts within the input by extracting information from the input (e.g., query, training document, source document, etc.) to determine a plurality of variables for generating content. The variables may include latent variables representing the underlying structure of the language.

[0206] According to examples of the present disclosure, a computing system may implement one or more generative machine learning models (3140, 9140) to generate various content having values ​​for one or more conditions (e.g., content for generating an overview, summary, response to a query, document template, output document generated based on a document template, etc.). One or more generative machine learning models (3140, 9140) may include a deep neural network or a generative adversarial network (GAN) to generate content having one or more features having values ​​for one or more conditions associated with one or more features. For example, one or more generative machine learning models (3140, 9140) may include a variational autoencoder, a stable diffusion machine learning model, a visual transformer, a neural radiation field (NeRF), etc. to generate content.

[0207] Additionally or alternatively, one or more machine-learned models (1140) may be included in a server computing system (1130) that communicates with a user computing device (1102) according to a client-server relationship, or otherwise stored in the server computing system (1130) and implemented by it. For example, one or more machine-learned models (1140) may be implemented by the server computing system (1130) as part of a web service (e.g., navigation service, word processing service, education service, etc.). Thus, one or more machine-learned models (1120) may be stored in and implemented on the user computing device (1102) and / or one or more machine-learned models (1140) may be stored in and implemented on the server computing system (1130).

[0208] The user computing device (1102) may also include one or more user input components (1122) that receive user input. For example, the user input component (1122) may be a touch-sensitive component (e.g., a touch-sensitive display screen or touchpad) that responds to the touch of a user input object (e.g., a finger or a stylus). The touch-sensitive component may serve to implement a virtual keyboard. Other examples of user input components include a microphone, a standard keyboard, or other devices and methods that enable the user to provide user input.

[0209] A server computing system (1130) (which may correspond to a server computing system (300)) includes one or more processors (1132) and memory (1134). One or more processors (1132) may be any suitable processing device (e.g., processor core, microprocessor, ASIC, FPGA, controller, microcontroller, etc.) and may be a single processor or a plurality of processors operably connected to each other. Memory (1134) may include one or more non-transient computer-readable storage media such as RAM, ROM, EEPROM, EPROM, flash memory device, magnetic disk, etc. and combinations thereof. Memory (1134) may store data (1136) and instructions (1138) executed by the processor (1132) to enable the server computing system (1130) to perform operations.

[0210] In some implementations, the server computing system (1130) includes one or more server computing devices or is otherwise implemented by them. Where the server computing system (1130) includes multiple server computing devices, these server computing devices may operate according to a sequential computing architecture, a parallel computing architecture, or some combination thereof.

[0211] As described above, the server computing system (1130) may store one or more machine-learned models (1140) or otherwise contain them. For example, one or more machine-learned models (1140) may be various machine-learned models or otherwise contain them. An exemplary machine-learned model includes a neural network or other multilayer nonlinear model. An exemplary neural network may include a feedforward neural network, a recurrent neural network (RNN) including a long short-term memory (LSTM)-based recurrent neural network, a convolutional neural network (CNN), a diffusion model, a generative adversarial network, or other forms of neural networks. An exemplary neural network may be a deep neural network. Some exemplary machine-learned models may utilize attention mechanisms such as self-attention. For example, some exemplary machine-learned models may include a multi-head self-attention model (e.g., a Transformer model). An exemplary machine-learned model has been described herein with reference to FIGS. 1a through 10e.

[0212] A user computing device (1102) and / or a server computing system (1130) can train one or more machine-learned models (1120 and / or 1140) through interaction with a training computing system (1150) that is coupled to communicate via a network (1180). The training computing system (1150) may be separate from the server computing system (1130) or may be part of the server computing system (1130).

[0213] The training computing system (1150) includes one or more processors (1152) and memory (1154). One or more processors (1152) may include any suitable processing device (e.g., processor core, microprocessor, ASIC, FPGA, controller, microcontroller, etc.) and may be one or more processors operably connected. The memory (1154) may include one or more non-transient computer-readable storage media such as RAM, ROM, EEPROM, EPROM, flash memory device, magnetic disk, etc. and combinations thereof. The memory (1154) may store data (1156) and instructions (1158) executed by the processor (1152) to enable the training computing system (1150) to perform operations. In some implementations, the training computing system (1150) includes one or more server computing devices or is otherwise implemented by them.

[0214] The training computing system (1150) may include a model trainer (1160) that trains one or more machine-learned models (1120 and / or 1140) stored in a user computing device (1102) and / or a server computing system (1130) using various training or learning techniques, such as, for example, backpropagation of errors. For example, a loss function may be backpropagated through the model to update one or more parameters of the model (e.g., based on the gradient of the loss function). Various loss functions may be used, such as mean squared error, likelihood loss, cross-entropy loss, hinge loss, and / or various other loss functions. Gradient descent techniques may be used to iteratively update parameters over a number of training iterations.

[0215] In some implementations, performing error backpropagation may include performing truncation backpropagation over time. The model trainer (1160) may perform a number of generalization techniques (e.g., weight decay, dropout, etc.) to improve the generalization ability of the model being trained.

[0216] In particular, the model trainer (1160) may train one or more machine-learned models (1120 and / or 1140) based on a set of training data (1162). The training data (1162) may include various datasets that may be stored, for example, remotely or in a training computing system (1150). For example, in some implementations, an exemplary dataset used for training includes a large language training data corpus that enables one or more large language models to perform multiple language tasks. For example, one or more large language models may be trained to perform summarization tasks, conversation tasks, simplification tasks, opposing viewpoint tasks, etc. In particular, one or more large language models may be trained to generate language output by processing various outputs. However, other datasets (e.g., image datasets) may also be used (e.g., images obtained from an external website). In some implementations, the dataset may be limited to a specific genre or topic, a specific type of content including images, videos, and text, or a specific style or type of content (e.g., outlines, reports, presentations, spreadsheets, resumes, PRDs, etc.). In some implementations, the dataset may include various topics.

[0217] In some implementations, if the user consents, training examples may be provided by the user computing device (1102). Thus, in such implementations, one or more machine-learned models (1120) provided to the user computing device (1102) may be trained by the training computing system (1150) on user-specific data received from the user computing device (1102). In some cases, this process may be referred to as personalizing the model.

[0218] A model trainer (1160) includes computer logic utilized to provide desired functions. The model trainer (1160) may be implemented as hardware, firmware, and / or software that controls a general-purpose processor. For example, in some implementations, the model trainer (1160) includes a program file that is stored on a storage device, loaded into memory, and executed by one or more processors. In other implementations, the model trainer (1160) includes one or more sets of computer-executable instructions stored on a computer-readable storage medium of a type such as RAM, a hard disk, or optical or magnetic media.

[0219] The network (1180) may be any type of communication network, such as a local area network (e.g., intranet), a wide area network (e.g., the Internet), or some combination thereof, and may include any number of wired or wireless links. Generally, communication through the network (1180) may be performed via any type of wired and / or wireless connection using a wide variety of communication protocols (e.g., TCP / IP, HTTP, SMTP, FTP), encoding or format (e.g., HTML, XML), and / or protection methods (e.g., VPN, secure HTTP, SSL).

[0220] The machine learning models described in this specification can be used in various tasks, applications, and / or use cases.

[0221] In some implementations, the input to the machine learning model(s) of the present disclosure may be text or natural language data. The machine learning model(s) may process the text or natural language data to generate an output. For example, the machine learning model(s) may process the natural language data to generate a language encoding output. For another example, the machine learning model(s) may process the text or natural language data to generate a latent text embedding output. For another example, the machine learning model(s) may process the text or natural language data to generate a translation output. For another example, the machine learning model(s) may process the text or natural language data to generate a classification output. For another example, the machine learning model(s) may process the text or natural language data to generate a text segmentation output. For another example, the machine learning model(s) may process the text or natural language data to generate a semantic intent output. As another example, machine learning model(s) can process text or natural language data to generate upscaled text or natural language output (e.g., text or natural language data of higher quality than the input text or natural language). As another example, machine learning model(s) can process text or natural language data to generate predictive output.

[0222] In some implementations, the input to the machine learning model(s) of the present disclosure may be speech data. The machine learning model(s) may process the speech data to generate an output. For example, the machine learning model(s) may process the speech data to generate a speech recognition output. As another example, the machine learning model(s) may process the speech data to generate a speech translation output. As another example, the machine learning model(s) may process the speech data to generate a latent embedding output. As another example, the machine learning model(s) may process the speech data to generate an encoded speech output (e.g., an encoded and / or compressed representation of the speech data, etc.). As another example, the machine learning model(s) may process the speech data to generate an upscaled speech output (e.g., speech data of higher quality than the input speech data, etc.). As another example, machine learning model(s) can process speech data to generate a text representation output (e.g., a text representation of the input speech data). As another example, machine learning model(s) can process speech data to generate a prediction output.

[0223] In some implementations, the input to the machine learning model(s) of the present disclosure may be sensor data. The machine learning model(s) may process the sensor data to generate an output. As one example, the machine learning model(s) may process the sensor data to generate a recognition output. As another example, the machine learning model(s) may process the sensor data to generate a prediction output. As another example, the machine learning model(s) may process the sensor data to generate a classification output. As another example, the machine learning model(s) may process the sensor data to generate a segmentation output. As another example, the machine learning model(s) may process the sensor data to generate a visualization output. As another example, the machine learning model(s) may process the sensor data to generate a diagnosis output. As another example, the machine learning model(s) may process the sensor data to generate a detection output.

[0224] FIG. 11a illustrates an exemplary computing system that may be used to implement an embodiment of the present disclosure. Other computing systems may also be used. For example, in some implementations, a user computing device (1102) may include a model trainer (1160) and training data (1162). In such implementations, one or more machine-learned models (1120) may be trained locally on the user computing device (1102) and used. In some of these implementations, the user computing device (1102) may implement the model trainer (1160) to personalize one or more machine-learned models (1120) based on user-specific data.

[0225] FIG. 11b illustrates a block diagram of an exemplary computing device for organizing, managing, and creating content by implementing one or more machine-learned models in relation to training content and / or source content selected by a user, according to one or more exemplary embodiments of the present disclosure. The computing device (1200) may be a user computing device or a server computing device.

[0226] The computing device (1200) includes a plurality of applications (e.g., applications 1 through N). Each application includes its own machine learning library and machine-learned model(s). For example, each application may include a machine-learned model. Exemplary applications include the notebook application described herein, the document extractor application described herein, the text messaging application, the email application, the dictation application, the virtual keyboard application, the browser application, the social media application, the map application, the navigation application, etc.

[0227] As illustrated in FIG. 11b, each application may communicate with a number of other components of the computing device, such as, for example, one or more sensors, a context manager, a device state component, and / or additional components. In some implementations, each application may communicate with each device component using an API (e.g., a public API). In some implementations, the API used by each application is specific to that application.

[0228] FIG. 11c illustrates a block diagram of an exemplary computing device for organizing, managing, and creating content by implementing one or more machine-learned models in relation to training content and / or source content selected by a user, according to one or more exemplary embodiments of the present disclosure. The computing device (1300) may be a user computing device or a server computing device.

[0229] The computing device (1300) includes a number of applications (e.g., applications 1 through N). Each application communicates with a central intelligence layer. Exemplary applications include the notebook application described herein, the document extractor application described herein, the text messaging application, the email application, the dictation application, the virtual keyboard application, the browser application, the map application, the social media application, the navigation application, the social media application, etc. In some implementations, each application may communicate with the central intelligence layer (and the model stored therein) using an API (e.g., an API common to all applications).

[0230] The central intelligence layer includes multiple machine-learned models. For example, as illustrated in FIG. 11c, individual machine-learned models may be provided for each application and managed by the central intelligence layer. In other implementations, two or more applications may share a single machine-learned model. For example, in some implementations, the central intelligence layer may provide a single model for all applications. In some implementations, the central intelligence layer is included within or otherwise implemented by the operating system of the computing device (1300).

[0231] The central intelligence layer can communicate with the central device data layer. The central device data layer may be a centralized repository of data for the computing device (1300). As illustrated in FIG. 11c, the central device data layer may communicate with a number of other components of the computing device, such as, for example, one or more sensors, a context manager, a device state component, and / or additional components. In some implementations, the central device data layer may communicate with each device component using an API (e.g., a private API).

[0232] Insofar as so-called general terms including 'module' and 'unit' are used herein, such terms may refer to, but are not limited to, software or hardware components or devices such as Field Programmable Gate Arrays (FPGAs) or Application-Specific Integrated Circuits (ASICs) that perform specific tasks. A module or unit may be configured to reside on an addressable storage medium and to be executed on one or more processors. Accordingly, a module or unit may include components such as, for example, software components, object-oriented software components, class components and task components, processes, functions, attributes, procedures, subroutines, program code segments, drivers, firmware, microcode, circuits, data, databases, data structures, tables, arrays, and variables. The functions provided by components and modules / units may be combined into fewer components and modules / units or further separated into additional components and modules.

[0233] Aspects of the exemplary embodiments described above may be recorded on a non-transient computer-readable medium containing program instructions for implementing various operations implemented by a computer. The medium may also include data files, data structures, etc., separately from or in combination with the program instructions. Examples of non-transient computer-readable media include magnetic media such as hard disks, floppy disks, and magnetic tapes; optical media such as CD-ROM disks, Blu-ray disks, and DVDs; magneto-optical media such as optical discs; and other hardware devices specifically configured to store and execute program instructions, such as semiconductor memory, read-only memory (ROM), random access memory (RAM), flash memory, and USB memory. Examples of program instructions include both machine code, such as that generated by a compiler, and files containing high-level code that can be executed by a computer using an interpreter. Program instructions may be executed by one or more processors. To perform the operations of the aforementioned embodiments, the described hardware devices may be configured to function as one or more software modules, and vice versa. Additionally, non-transient computer-readable storage media may be distributed across computer systems connected via a network, and computer-readable code or program instructions may be stored and executed in a distributed manner. Furthermore, non-transient computer-readable storage media may be implemented in at least one Application-Specific Integrated Circuit (ASIC) or Field Programmable Gate Array (FPGA).

[0234] Each block depicted in the flowchart may represent a unit, module, segment, or section of code containing one or more executable instructions for implementing a specified logical function. Additionally, it should be noted that in some alternative implementations, the functions mentioned in the blocks may occur out of order. For example, two blocks depicted consecutively may actually be executed substantially in parallel, or blocks may often be executed in reverse order depending on the related functions.

[0235] Although the present disclosure has been described in relation to various exemplary embodiments, each example is provided for illustrative purposes rather than to limit the present disclosure. Those skilled in the art can easily devise changes, variations, and equivalents to these embodiments upon understanding the foregoing. Accordingly, the present disclosure does not exclude such modifications, variations, and / or additions to the subject matter disclosed that may be readily apparent to those skilled in the art. For example, features illustrated or described as part of one embodiment may be used in conjunction with other embodiments to derive other embodiments. Accordingly, the present disclosure is intended to include such changes, variations, and equivalents.

Claims

Claim 1 A computing device for generating an output document, comprising: one or more memories configured to store instructions; and one or more processors configured to execute said instructions for performing operations, wherein the operations include: receiving a plurality of source documents; receiving an input associated with a request to generate said output document based on said source documents and a specific document template; and generating said output document having said specific document template through one or more machine-learned models. Claim 2 A computing device according to claim 1, wherein the operations further include an operation of providing a graphical user interface configured to receive a selection of the plurality of source documents for display on a display device. Claim 3 A computing device according to claim 1, wherein the operations further include an operation of determining an intention associated with the output document based on the content associated with each of the plurality of source documents by the one or more machine-learned models. Claim 4 A computing device according to claim 1, wherein the operations further include an operation of determining a style associated with the output document based on the content of each of the plurality of source documents by the one or more machine-learned models. Claim 5 A computing device according to claim 1, wherein the operations further include an operation of determining a format associated with the output document based on the content of each of the plurality of source documents by the one or more machine-learned models. Claim 6 A computing device according to claim 5, wherein the operation of determining the format associated with the output document based on the content of each of the plurality of source documents by the one or more machine-learned models includes the operation of identifying a plurality of headings for individual sections of the output document. Claim 7 A computing device according to claim 1, wherein the operations include: an operation of providing a graphical user interface including a plurality of user interface elements for display on a display device, wherein each of the plurality of user interface elements corresponds to a different style that can be applied by the one or more machine-learned models to generate the output document having the specific document template; and further including an operation of receiving a selection for one or more of the plurality of user interface elements, wherein the operation of generating the output document having the specific document template through the one or more machine-learned models includes an operation of applying one or more styles corresponding to the selection for one or more of the plurality of user interface elements. Claim 8 A computing device according to claim 1, wherein the operations include: an operation of providing a graphical user interface comprising a plurality of user interface elements for display on a display device, wherein each of the plurality of user interface elements corresponds to a different heading that can be applied by the one or more machine-learned models to generate a section of the output document having the specific document template; and further comprising an operation of receiving a selection for one or more of the plurality of user interface elements, wherein, through the one or more machine-learned models, the operation of generating the output document having the specific document template includes an operation of applying one or more headings corresponding to the selection for one or more of the plurality of user interface elements to a section of the output document. Claim 9 The computing device according to claim 1, wherein the computing device further comprises one or more databases configured to store a plurality of generative machine learning models each associated with a plurality of different document types, and the operations further comprise an operation of searching among the plurality of generative machine learning models for at least one generative machine learning model associated with a document type represented by the common content of the plurality of source documents. Claim 10 A computing device according to claim 1, wherein the operations further include: receiving a plurality of training documents associated with a document type—each of the plurality of training documents includes one or more common features associated with the specific document template—; and training the one or more machine-learned models to learn at least one of the intent, style, or format of the specific document template for the document type. Claim 11 A computer implementation method comprising: receiving a plurality of source documents by a computing system comprising one or more processors and one or more machine-learned models; receiving an input associated with a request to generate an output document based on the plurality of source documents and a specific document template by the computing system; and generating the output document having the specific document template through the one or more machine-learned models. Claim 12 A computer-implemented method according to claim 11, further comprising the step of determining at least one of a style, intent, and format associated with an output document based on the content of each of the plurality of source documents by the one or more machine-learned models. Claim 13 A computer-implemented method according to claim 11, comprising the step of providing a graphical user interface including a plurality of user interface elements for display on a display device of the computing system, wherein each of the plurality of user interface elements corresponds to a style, intent, or format that can be applied by the one or more machine-learned models to generate the output document having the specific document template; and further comprising the step of receiving a selection for one or more of the plurality of user interface elements, wherein the step of generating the output document having the specific document template through the one or more machine-learned models comprises the step of applying the style, intent, or format corresponding to the selection for one or more of the plurality of user interface elements. Claim 14 A computer-implemented method according to claim 11, further comprising: a step of storing a plurality of generative machine learning models associated with a plurality of different document types, respectively, in one or more databases; and a step of searching among the plurality of generative machine learning models for at least one generative machine learning model associated with a document type represented by the common content of the plurality of source documents. Claim 15 A computer-implemented method according to claim 11, further comprising the step of receiving a plurality of training documents associated with a document type—each of which the plurality of training documents includes one or more common features associated with the specific document template—; and the step of training the one or more machine-learned models to learn at least one of the intent, style, or format of the specific document template for the document type. Claim 16 A computing device for generating a document template, comprising: one or more memories configured to store instructions; and one or more processors configured to execute said instructions for performing operations, wherein the operations include: receiving a selection of a plurality of sample documents; receiving an input for generating said document template based on the selection of said sample documents; and in response to receiving said input, implementing one or more machine-learned models for generating said document template based on at least one of a style, format, or intent associated with said sample documents. Claim 17 A computing device according to claim 16, wherein the operations further include: an operation of receiving a plurality of source documents; an operation of receiving additional input associated with a request to generate an output document based on the plurality of source documents and the document template; and an operation of generating the output document according to the document template through the one or more machine-learned models. Claim 18 A computing device according to claim 17, wherein the operations further include an operation of determining at least one of the style, the format, and the intent associated with the plurality of sample documents based on the content and structure of each of the plurality of sample documents by the one or more machine-learned models. Claim 19 In claim 18, the above operations further include: an operation of providing a graphical user interface including a plurality of user interface elements for display on a display device, wherein at least one of the plurality of user interface elements is selectable to change at least one of the style, the format, and the intent associated with the plurality of sample documents determined by the one or more machine learning models; and an operation of receiving a selection for one or more of the plurality of user interface elements to change at least one of the style, the format, and the intent associated with the plurality of sample documents determined by the one or more machine learning models, wherein the operation of generating the output document according to the document template through the one or more machine learning models includes an operation of applying at least one of the style, the format, and the intent changed according to the selection for one or more of the plurality of user interface elements, a computing device. Claim 20 In claim 16, the computing device further comprises one or more databases configured to store a plurality of generative machine learning models associated with a plurality of different document types, and the operations further comprise an operation of searching among the plurality of generative machine learning models for at least one generative machine learning model associated with a document type represented by the common content of the plurality of sample documents.

Images