Information processing systems, information processing methods, and programs

Abstract

This technology provides the ability to obtain the information to be included in the output. [Solution] The information processing system acquires one or more items, acquires target information, generates input information for a machine learning model based on one or more items and target information, acquires the information generated by inputting the input information into the machine learning model, and generates output information including information corresponding to one or more items based on the generated information.

Description

Technical Field

[0001] The present disclosure relates to an information processing system, an information processing method, and a program.

Background Art

[0002] Machine learning models such as large language models (LLMs) are known. A large language model executes a task according to input information called a prompt and outputs the data generated as a result. In order to obtain good output results from a large language model, a technique of including reference information in the input information to the large language model is known.

Prior Art Documents

Non-Patent Documents

[0003]

Non-Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] An object of the present disclosure is to provide a technique for acquiring information to be included in output information.

Means for Solving the Problems

[0005] An information processing device according to one aspect of the present disclosure comprises at least one memory and at least one processor, wherein the at least one processor acquires one or more items, acquires target information, generates input information for a machine learning model based on the one or more items and the target information, acquires information generated by inputting the input information into the machine learning model, and generates output information including information corresponding to one or more items based on the generated information. [Brief explanation of the drawing]

[0006] [Figure 1] This is a block diagram showing an example of the overall configuration of the information processing system according to the first embodiment. [Figure 2] This block diagram shows an example of the functional configuration of a document generation device according to the first embodiment. [Figure 3] This figure shows an example of a document template. [Figure 4] This figure shows an example of the target information. [Figure 5] This figure shows an example of a prompt template. [Figure 6] This is a flowchart illustrating an example of the document generation process. [Figure 7] This is a block diagram showing an example of the overall configuration of the information processing system according to the second embodiment. [Figure 8] This block diagram shows an example of the functional configuration of a document generation device according to the second embodiment. [Figure 9] A block diagram showing an example of a computer hardware configuration. [Modes for carrying out the invention]

[0007] Hereinafter, embodiments of this disclosure will be described with reference to the accompanying drawings. In this specification and the drawings, components having substantially the same functional configuration are denoted by the same reference numerals, and redundant descriptions will be omitted.

[0008] [First Embodiment] A first embodiment of this disclosure is an information processing system that performs a predetermined task based on a machine learning model. The machine learning model may, for example, be a large-scale language model. The machine learning model may be a generative model, a foundational model, or a neural network capable of generating various types of data such as audio, images, and videos. The machine learning model may also support multimodal processing.

[0009] The information processing system according to this embodiment generates document data, which is an example of output information, based on the information to be processed (hereinafter also referred to as "target information"). The document data is electronic data in which a predetermined document is electronically recorded. The information processing system performs a generation task to generate information to be included in the document data based on a machine learning model. The document may include, for example, a quotation, an invoice, a proposal, a report, meeting minutes, an email, a letter, etc. Note that the output information is not limited to document data. The output information may also be any electronic data including image data (including still images or videos) or audio data, as another example.

[0010] The information processing system according to this embodiment may generate document data by embedding information generated based on target information into a document template having one or more items. The document template may have placeholders for embedding information corresponding to one or more items. The information processing system may generate information corresponding to one or more items contained in the document based on a machine learning model. The information processing system may generate document data by embedding the generated information into placeholders.

[0011] This embodiment provides a technology capable of generating information to be included in output information. In this embodiment, input information to a machine learning model is generated based on one or more items and target information, the generated information is obtained by inputting the input information into the machine learning model, and output information including information corresponding to one or more items is generated based on the generated information.

[0012] On one hand, according to the present embodiment, information corresponding to one or more items included in the output information can be generated based on the target information. On the other hand, according to the present embodiment, since information to be included in the output information is generated based on the machine learning model, various forms of output information can be easily generated.

[0013] <Overall Configuration of Information Processing System> The overall configuration of the information processing system according to the present embodiment will be described with reference to FIG. 1. FIG. 1 is a block diagram showing an example of the overall configuration of the information processing system according to the first embodiment.

[0014] As shown in FIG. 1, the information processing system 1000 includes a document generation device 10, a generation device 20, and a terminal device 30. The document generation device 10, the generation device 20, and the terminal device 30 may be connected to each other so as to be capable of data communication via a communication network such as a LAN (Local Area Network) or the Internet. <0000​​​​​​​​​​​​​​The document generation device 10 may generate input information for the machine learning model based on the item information and target information included in the generation request. The document generation device 10 may obtain the output information generated by inputting the input information for the machine learning model into the machine learning model from the generation device 20. The document generation device 10 may generate document data based on the output information obtained from the generation device 20 and transmit it to the terminal device 30 as the generation result for the generation request.

[0019] The generation device 20 is an example of an information processing device such as a personal computer, a workstation, or a server that executes a predetermined task using a machine learning model. The generation device 20 may include a machine learning model M. The machine learning model M is a machine learning model for executing a predetermined task. The machine learning model M may be, for example, a large language model, a base model, a generation model, or a neural network.

[0020] The machine learning model M may be realized by one machine learning model. The machine learning model M may be realized by the cooperation of a plurality of machine learning models. The machine learning model M may be composed of a plurality of machine learning models according to the task to be executed.

[0021] The generation device 20 may be realized by a plurality of devices or systems including different machine learning models M. The generation device 20 may be realized by one device or system including a plurality of machine learning models M. The generation device 20 may be an external device or system. The generation device 20 may execute a predetermined task based on an external machine learning model M. Note that "external" means not included in the information processing system 1000.

[0022] Terminal device 30 is an example of an information processing terminal such as a personal computer, smartphone, or tablet terminal operated by a user of the information processing system 1000. Terminal device 30 may request document generation from document generation device 10. Terminal device 30 may accept the specification of item information and target information in response to user operations. Terminal device 30 may transmit a generation request including item information and target information to document generation device 10. Terminal device 30 may receive the generation result from document generation device 10 and present it to the user.

[0023] For example, the terminal device 30 may display the generated results on its display device. Presenting information to the user may include the processor performing at least a portion of the processing necessary to display the information on the display device. The display device may be provided in the same device as the processor, or in a different device from the processor. There may be multiple display devices.

[0024] For example, the terminal device 30 may output synthesized audio from its speaker. Presenting information to the user may include the processor performing at least a portion of the processing necessary for it to output audio to the speaker. The speaker may be located in the same device as the processor or in a different device. There may be multiple speakers.

[0025] Note that the overall configuration of the information processing system 1000 shown in Figure 1 is just one example, and various system configurations are possible depending on the application and purpose. The information processing system 1000 may consist of one or more information processing devices. The devices included in the information processing system 1000 may be a system composed of multiple devices. Each function included in the information processing system 1000 may be implemented by any device that constitutes the system. Each component included in the information processing system 1000 may be included in any device that constitutes the system.

[0026] The machine learning model M may be built into the document generation device 10. Alternatively, the machine learning model M may be distributed and held in an external information processing system consisting of multiple devices. In this case, the information processing system 1000 does not need to include the generation device 20.

[0027] Multiple copies of one or more of the document generation devices 10, generation devices 20, and terminal devices 30 may be included in the information processing system 1000. The document generation device 10 or generation device 20 may be implemented using multiple computers, or as a cloud computing service. The document generation devices 10 and 20 may also be implemented using standalone computers. The classification of devices such as the document generation device 10, generation device 20, and terminal device 30 shown in Figure 1 is just one example.

[0028] As an example, the information processing system 1000 may consist of a server device and one or more terminal devices 30. The server device may include the functions of a document generation device 10 and a generation device 20. The server device may be implemented as a system including multiple information processing devices. The server device may be implemented as a cloud computing service.

[0029] For example, the information processing system 1000 may consist of a single information processing device. The information processing device may include the functions of a document generation device 10, a generation device 20, and a terminal device 30.

[0030] <Functional Configuration of Document Generation Device> The functional configuration of the document generation device 10 will be explained with reference to Figure 2. Figure 2 is a block diagram showing an example of the functional configuration of the document generation device according to the first embodiment.

[0031] As shown in Figure 2, the document generation device 10 comprises a request receiving unit 110, an item acquisition unit 120, an information acquisition unit 130, a generation unit 140, an embedding unit 150, a document storage unit 160, and an output unit 170. The document generation device 10 functions as the request receiving unit 110, item acquisition unit 120, information acquisition unit 130, generation unit 140, embedding unit 150, document storage unit 160, and output unit 170 when a pre-installed document generation program is executed.

[0032] The request receiving unit 110 receives requests for the generation of document data. The request receiving unit 110 may also receive generation requests from the terminal device 30. Generation requests may also be transmitted from the terminal device 30 in response to operations on the screen displayed on the display device of the terminal device 30. The request receiving unit 110 may also receive generation requests input to the document generation device 10. Generation requests may also be input to the document generation device 10 in response to operations on the screen displayed on the display device of the document generation device 10.

[0033] A generation request may include item information and target information. Item information is information about items included in a document. Item information may include information about one or more items. Target information is information used to generate document data. A generation request may include multiple target information.

[0034] Item information may, for example, be a document template that serves as a model for a document. A document template may include areas for embedding information corresponding to one or more items contained in the document. A document template may, for example, include placeholders corresponding to one or more items. A placeholder is an example of an area for embedding information. A document template may include the item names of one or more items, descriptive information for each item, etc. The descriptive information may be text data describing an explanation or annotation about the item. A document template may also include a list of placeholders.

[0035] A placeholder only needs to contain information that can be identified as corresponding to an item. Placeholders may be written in a prescribed format. Placeholders may also contain information that can identify an item, with a prescribed symbol or graphic attached. The information that can identify an item may be the item name. The information that can identify an item may also include descriptive information about the item. The prescribed symbol may be, for example, a parenthetical symbol. That is, a placeholder may be a string enclosed in a prescribed parenthetical symbol. The parenthetical symbol may be any symbol, including, for example, square brackets, round brackets, quotation marks, and black brackets. Placeholders may also contain an initial value for the information corresponding to the item.

[0036] A document template may include a program that dynamically generates placeholders. For example, a document template containing a variable number of items may include a program that generates placeholders corresponding to the number of pieces of information to be embedded. A document template may include a program that processes the information to be embedded in the placeholders. A document template may include a program that validates the information to be embedded in the placeholders. The program that validates the information may highlight or display an error message for placeholders that fail validation.

[0037] A document template may include text data, structured data, image data, or audio data. Image data may be a still image or a video, for example. Text data may be text data obtained by speech recognition of audio data or video. Text data may also be text obtained by character recognition of image data. Audio data may be speech data obtained by speech synthesis of text data. Structured data may be a table or graph, for example.

[0038] A document template may, for example, be an electronic file created using a text editor or other document creation tool such as office software. A document template may also be an electronic file stored in the document creation tool's specified output format, or an electronic file obtained by converting such files to another output format.

[0039] Document templates may be written in various formats. Examples of document formats include natural language, Markdown, HTML (Hyper Text Markup Language), or XML (Extensible Markup Language).

[0040] Document templates may be created in various file formats. These file formats may include, for example, text, image data, video data, audio data, tabular formats, presentation formats, or PDF (Portable Document Format) formats.

[0041] The document template may be selected from several pre-created document templates. The pre-created document templates may include document data generated in the past. The document template may be newly created in response to user operations. The document template may be selected in response to user operations. The document template may be selected or generated in response to operations on the screen displayed on the display device of the terminal device 30. The screen for selecting a document template may, for example, display a list of document templates, a description of each document template, an overall image, a thumbnail image, the names of items included in each document template, placeholders, descriptions, etc.

[0042] Item information may also be, as another example, electronic data describing a list of items included in the document. The list of items may also be a list of placeholders included in a pre-created document template. The list of items may also be a list of item names included in a pre-created document template. The list of items may include descriptive information for each item, or initial values ​​for the information corresponding to each item. The list of items may also be extracted in advance from the document template.

[0043] The target information may include information corresponding to one or more items contained in the document. The target information may include information from which information corresponding to one or more items contained in the document can be derived. Information from which information corresponding to an item can be derived from multiple pieces of information to determine information corresponding to one item. Information from which information corresponding to an item can be derived from a single piece of information to determine information corresponding to multiple items.

[0044] The target information may be electronic data recording a pre-created document. The pre-created document may be a document related to the document to be generated. The pre-created document may be the document to be generated, or it may be a different document from the document to be generated. The target information may also be document data generated by the information processing system 1000 or an external information processing device or information processing system. The target information may also be electronic data obtained by scanning a document created on paper.

[0045] The information may be written in various formats. Examples of document formats include natural language, Markdown, HTML (Hyper Text Markup Language), or XML (Extensible Markup Language).

[0046] The target information may be created in various file formats. Examples of file formats include text, image data, video data, audio data, tabular formats, presentation formats, or PDF (Portable Document Format). The target information may also include text recognized or extracted from data created in image data, video data, audio data, tabular formats, presentation formats, or PDF formats. Recognition may include image recognition, speech recognition, or optical character recognition.

[0047] The target information may be selected from a set of pre-created target information. The target information may also be newly created in response to user operations. The target information may also be selected or generated in response to operations on the screen displayed on the display device of the terminal device 30. The screen for selecting target information may, for example, display a list of target information, a description of each target information, an overall image, a thumbnail image, and information corresponding to one or more items included in each target information.

[0048] The item acquisition unit 120 acquires one or more items to be included in the document data. The one or more items may also be identification information for one or more items. The identification information for one or more items may include at least one of the following: an item name, item description information, or one or more placeholders corresponding to each of the one or more items.

[0049] The item acquisition unit 120 may acquire one or more items based on the generation request. The item acquisition unit 120 may acquire one or more items based on the item information included in the generation request. The item acquisition unit 120 may extract one or more items from the item information.

[0050] The item acquisition unit 120 may acquire a list of placeholders included in a document template, which is an example of item information. The item acquisition unit 120 may extract one or more placeholders by detecting areas written in a predetermined format from the document template. The item acquisition unit 120 may also detect a predetermined symbol from the document template and extract one or more placeholders to which the detected symbol is attached. As an example, the item acquisition unit 120 may detect square brackets from the document template shown in Figure 3 and extract the string within the detected square brackets as a single item.

[0051] One or more items acquired by the item acquisition unit 120 may be modified by the user. The item acquisition unit 120 may present the acquired one or more items to the user and accept modifications to one or more items by the user. One or more items may be specified by the user. The item acquisition unit 120 may accept input of one or more items by the user. The item acquisition unit 120 may present the acquired one or more items to the user and accept additions of one or more items by the user.

[0052] The information acquisition unit 130 acquires one or more target information. The information acquisition unit 130 may acquire one or more target information based on a generation request. The information acquisition unit 130 may acquire one or more target information included in a generation request. The information acquisition unit 130 may acquire one or more target information from an external data source. The information acquisition unit 130 may acquire one or more target information by searching an external data source based on a generation request.

[0053] The generation unit 140 generates output information from the machine learning model M based on one or more items acquired by the item acquisition unit 120 and one or more target information acquired by the information acquisition unit 130. As an example, the generation unit 140 may generate input information for the machine learning model M based on one or more items and one or more target information and transmit it to the generation device 20. Hereinafter, the input information for the machine learning model M will also be referred to as "model input information." Similarly, the output information from the machine learning model M will also be referred to as "model output information."

[0054] The generation unit 140 may generate model input information by embedding one or more items and one or more target information into a predetermined template. Hereinafter, the template for generating model input information will be referred to as a "prompt template". The generation unit 140 may also generate model input information by processing the prompt template, one or more items, and one or more target information based on predetermined rules. The generation unit 140 may also generate model input information by inputting the prompt template, one or more items, and one or more target information into a machine learning model different from the machine learning model M. The generation unit 140 may receive model output information generated by the generation device 20 inputting the model input information into the machine learning model M from the generation device 20.

[0055] The generation unit 140 may generate model output information by inputting model input information into one machine learning model M selected from multiple machine learning models M. The machine learning model M used to generate the model output information may be specified by a generation request, for example. The machine learning model M specified in the generation request may be selected by the user.

[0056] A prompt template may include one or more placeholders that embed one or more items and one or more pieces of target information. A prompt template may also include instruction information regarding the process to be executed by the machine learning model M, and constraint information regarding the model output information to be output by the machine learning model M. The instruction information and constraint information may be predetermined fixed statements.

[0057] The prompt template may include a first instruction that instructs the system to extract information corresponding to one or more items from the target information. The prompt template may also include a second instruction that instructs the system to generate information corresponding to one or more items. The second instruction may instruct the system to generate information corresponding to one or more items based on the target information, or it may instruct the system to generate information corresponding to one or more items without being based on the target information.

[0058] The prompt template may include a third instruction that specifies what to do if information corresponding to one or more items cannot be extracted from the target information. The third instruction may be an instruction to output empty information, an initial value, or information newly generated by the machine learning model M if information corresponding to one or more items cannot be extracted from the target information. The empty information may include any information indicating that information could not be extracted. The information newly generated by the machine learning model M may be information not included in the target information. The information newly generated by the machine learning model M may be information that can be generated based on the target information. The third instruction may also include information specifying an initial value.

[0059] A prompt template may include constraint information regarding the format of information corresponding to one or more items. This constraint information may include information specifying the format of information corresponding to one or more items. The formatting information may be specified for each type of information. The type of information may, for example, be a data type. The data type may include, for example, dates, times, amounts, numbers, etc.

[0060] The prompt template may include information specifying the format of the model output information. Examples of model output information formats include natural language text, Markdown notation, HTML (Hyper Text Markup Language), XML (Extensible Markup Language), JSON (JavaScript Object Notation) format, CSV (Comma Separated Values) format, Mermaid format, etc.

[0061] The model input information may include, for each of the one or more items, a first instruction information that instructs the extraction of information corresponding to that item from the target information, or a second instruction information that instructs the generation of information corresponding to that item. Whether to include the first instruction information or the second instruction information for each of the one or more items may be indicated in the generation request or item information. Whether to include the first instruction information or the second instruction information for each of the one or more items may be selected by the user.

[0062] Model input information may include text data, image data, or audio data. Text data may, for example, be natural language sentences called prompts. Image data may, for example, be still images or videos. Text data may also be text data obtained by speech recognition of audio data or videos. Text data may also be text obtained by character recognition of image data. Image data may, for example, include images of the user. Audio data may, for example, include the voice of the user speaking. Audio data may also be speech data obtained by speech synthesis of text data.

[0063] Model input information may include information indicating the correspondence between items included in the target information and placeholders included in the document template. For example, the information indicating the correspondence between items and placeholders may include at least one of the item name or item description information and information indicating the correspondence with the placeholder.

[0064] Model input information may include information about the structure of the document template. Information about the structure of the document template may also include information indicating the structure of placeholders included in the document template. Information indicating the structure of placeholders may include, for example, information about at least one of the item names, item order, or item hierarchy.

[0065] Model input information may include supplementary information about the document template. This supplementary information may include the language used in the document template and the number of characters in the placeholders included in the document template (at least one of the minimum or maximum).

[0066] Model input information may include a document template. Model input information may include image data of the document template. The document template may be embedded in placeholders for item information included in the prompt template. The image data of the document template may be attached to the model input information. In this case, the model input information may include instruction information that instructs to retrieve one or more items from the attached image data.

[0067] Model input information may include image data of the target information. Model input information may include audio data of the target information. Model input information may include text data obtained by recognizing the image data or audio data. Image data or audio data of the target information may be embedded in a placeholder for the target information included in the prompt template. Image data or audio data of the target information may be attached to the model input information. In this case, the model input information may include instruction information that instructs the model to extract information corresponding to one or more items from the attached image data or audio data.

[0068] The model input information may include all or at least part of the document template, target information, and instruction information that instructs the extraction of information corresponding to placeholders included in the document template from the target information. In this case, the model input information may further include instruction information that instructs the generation of output information in which the extracted information is embedded in the placeholders included in the document template.

[0069] The generation unit 140 can generate model input information by embedding the necessary information into the placeholders of the prompt template. The generation unit 140 may also generate model input information without using a prompt template. The generation unit 140 can easily generate model input information by using a prompt template. The prompt template may be pre-optimized to obtain good model output information. By generating prompts using an optimized template, good model output information can be stably obtained from the machine learning model M.

[0070] The model input information generated by the generation unit 140 may be modified by the user. The generation unit 140 may present the generated model input information to the user and accept modifications to the model input information by the user.

[0071] Model output information (information generated by the machine learning model M) includes information corresponding to one or more items. Model output information may also include information extracted from target information by the machine learning model M.

[0072] Model output information is output in the format specified in the model input information. The format of the model output information does not need to be one that allows for the identification of the correspondence with one or more items. Model output information may be generated for each of one or more items. In this case, model input information is generated for each of one or more items. The generation unit 140 may transmit the model input information for each item to the generation device 20 sequentially or in parallel, and receive the model input information for each item from the generation device 20 sequentially or in parallel.

[0073] The information corresponding to one or more items may be generated in any data format. Examples of such data formats include text data, structured data, image data, or sound data.

[0074] The embedding unit 150 generates document data, which is an example of the generation result. The embedding unit 150 may generate document data based on the model output information acquired by the generation unit 140. The embedding unit 150 may generate document data by embedding the model output information into a document template. The embedding unit 150 may extract information corresponding to one or more items from the model output information and embed the extracted information into placeholders in the document template.

[0075] The embedding unit 150 may generate information corresponding to one or more items based on the model output information, and use the generated information to generate document data. For example, the machine learning model M may generate identification information for information corresponding to one or more items as model output information, and the embedding unit 150 may acquire information corresponding to one or more items based on that identification information. The identification information for information corresponding to one or more items may include, as an example, information indicating an address or location. Alternatively, for example, the machine learning model M may generate a predetermined number as model output information, and the embedding unit 150 may generate information corresponding to one or more items based on that number.

[0076] The embedded unit 150 may process the information extracted from the model output information. The embedded unit 150 may generate one piece of information based on multiple pieces of information extracted from the model output information. The embedded unit 150 may generate multiple pieces of information based on one piece of information extracted from the model output information. The information processing method may include, as an example, format conversion, predetermined calculations, data joining, data splitting, etc.

[0077] The embedding unit 150 may embed information that is not based on model output information into the document template. For example, the embedding unit 150 may embed information based on the current time into the document template. For example, the information based on the current time may be the current day, or it may include dates or times that are earlier or later than the current day according to a predetermined rule.

[0078] The embedded unit 150 may verify the information extracted from the model output information. The embedded unit 150 may determine whether the multiple pieces of information extracted from the model output information conform to a predetermined rule when multiple pieces of information contained in the document conform to that rule. For example, if a total amount is stated in the document, the embedded unit 150 may calculate the sum of multiple amounts included in the total amount and determine whether it matches the total amount contained in the model output information.

[0079] The embedded unit 150 may correct the information extracted from the model output information. For example, if the information contained in the document conforms to a predetermined rule, and the information extracted from the model output information does not conform to that rule, the embedded unit 150 may correct at least one piece of information to conform to that rule. For example, if the document contains a total amount, and the sum of multiple amounts included in the total amount does not match the total amount contained in the model output information, the embedded unit 150 may correct the information corresponding to the total amount to the sum of multiple amounts.

[0080] The embedded unit 150 may correct the information extracted from the model output information in response to user operations. The embedded unit 150 may present the user with the model output information, the information extracted from the model output information, or document data in which the model output information is embedded, in order to have the user correct the information extracted from the model output information.

[0081] The embedded unit 150 may display items from which information has been extracted from the model output information and items from which information has not been extracted in different display formats in the model output information or document data presented to the user. The embedded unit 150 may also display the items themselves in different display formats. The embedded unit 150 may also display the information embedded in the items in different display formats. As an example, the embedded unit 150 may highlight or display an error for items from which information has not been extracted. Highlighting may include, as an example, changing at least one of the text color or background color, changing the text size, displaying an icon or message, etc.

[0082] Error displays may include icons or messages indicating the nature of the error. These icons or messages may be superimposed on items from which information was not extracted. They may also be displayed in response to user actions. For example, error displays may be shown as tooltips when the pointer is near a highlighted item.

[0083] The embedded unit 150 may present the user with the verification results of the information extracted from the model output information. The embedded unit 150 may display information that failed verification and information that succeeded verification in different display formats in the model output information or document data presented to the user. For example, the embedded unit 150 may highlight or display an error for information that failed verification.

[0084] The embedding unit 150 may embed information corresponding to items in a document template that includes items of a variable number. The embedding unit 150 may notify the user if the number of pieces of information included in the model output information is greater than the number of placeholders included in the document template. For example, the embedding unit 150 may highlight or display an error for items from which more pieces of information have been extracted than the number of placeholders.

[0085] The embedded unit 150 may present the user with the confidence level of the information extracted from the model output information. The confidence level may be included in the model output information. The confidence level may be generated by the machine learning model M. In order to cause the machine learning model M to generate the confidence level, the model input information may include instruction information that instructs the model to generate the confidence level.

[0086] The embedding unit 150 may accept corrections to the information by the user. The embedding unit 150 may verify the corrected information. If the embedding unit 150 fails to verify the corrected information, it may not accept the correction. If the embedding unit 150 fails to verify the corrected information, it may display an error regarding the corrected information. The embedding unit 150 may embed the corrected information into the document template. The embedding unit 150 may update the generated document data with document data in which the corrected information has been embedded.

[0087] The document storage unit 160 stores one or more document data. The document storage unit 160 may store document data generated by the embedding unit 150. The document storage unit 160 may store document data in chronological order of creation date and time or update date and time. The document storage unit 160 may store information used to generate the document data. The information used to generate the document data may include, as an example, at least one of the following: information indicating a machine learning model, item information, target information, model input information, prompt template, model output information, document template, information extracted from the model output information, and information embedded in the document template.

[0088] The document data stored in the document storage unit 160 may be updated by the embedding unit 150. The document data stored in the document storage unit 160 may be updated with document data to which information has been corrected by the user. When document data is updated, the document storage unit 160 may store both the document data before the update and the document data after the update. The document storage unit 160 may store both the values ​​of the corrected information before the update and the values ​​after the update.

[0089] The document storage unit 160 may store document data and information used to generate the document data in association with each other. As an example of how to associate document data and information used to generate the document data, one piece of information and the other piece of information may be stored as a set, information that allows the other piece of information to be retrieved from the first piece of information may be stored, or identification information for one piece of information and the other piece of information may be stored as a set.

[0090] The document storage unit 160 may store document data in a predetermined file format. The file format may be specified in the model input information. The file format may also be specified by the user.

[0091] The output unit 170 outputs the generation result for the generation request received by the request reception unit 110. The output unit 170 may generate the generation result based on the document data stored in the document storage unit 160. The output unit 170 may generate the generation result based on the information used to generate the document data stored in the document storage unit 160. The output unit 170 may transmit the generation result to the terminal device 30. The output unit 170 may display the generation result on the display device of the document generation device 10.

[0092] The output unit 170 may output document data in response to an acquisition request from the terminal device 30. The output unit 170 may automatically output document data when document data is generated by the embedded unit 150. Whether or not to automatically output document data may be indicated in the document data generation request. The output unit 170 may read the document data requested by the terminal device 30 from the document storage unit 160 and transmit it to the terminal device 30. The output unit 170 may accept a specification of document data by the user, read the specified document data from the document storage unit 160 and display it on the display device of the document generation device 10.

[0093] The output unit 170 may output connection information to the document data. This connection information may include, for example, a hostname, IP (Internet Protocol) address, directory name, file name, etc. It may also be a URL (Uniform Resource Locator) or URI (Uniform Resource Identifier). The output unit 170 may cause the terminal device 30 to download the document data in response to an operation based on the connection information in the terminal device 30. An operation based on the connection information may, for example, be an operation to select or execute link information indicating the URL of the document data.

[0094] Note that the functional configuration of the document generation device 10 shown in Figure 2 is just one example, and various configurations are possible depending on the application and purpose. There may be multiple users operating the terminal device 30. For example, the user who inputs item information and target information and the user who refers to the document data may be the same user or different users. The user who inputs item information and the user who inputs target information may be the same user or different users. The terminal device that performs the operation to input item information and target information and the terminal device that presents the document data to the user may be the same terminal device or different terminal devices. The terminal device that performs the operation to input item information and the terminal device that performs the operation to input target information may be the same terminal device or different terminal devices.

[0095] <Examples of document templates> Figure 3 shows an example of a document template. Figure 3 shows a document template for an invoice, which is an example of a document.

[0096] As shown in Figure 3, the document template 300 includes one or more placeholders. In the example shown in Figure 3, the placeholders are strings enclosed in square brackets (e.g., [ISSUE_DATE], [INVOICE_ID], [CLIENT_COMPANY_NAME], etc.). The document template 300 may also include item names corresponding to the placeholders (e.g., issue date, order number, etc.).

[0097] Placeholders may, for example, be strings that describe the item name in a specified format (e.g., [Issue Date], [Order Number], etc.). Placeholders may also include, for example, descriptive information for the item (e.g., [ISSUE_DATE:Issue Date], [INVOICE_ID:Order Number], etc.).

[0098] Placeholders may contain information that is not based on the target information. For example, placeholders may contain information based on the current time. For instance, the payment due date ([DUE_DATE]) may contain the last day of the following month. Also, for example, special notes ([NOTE]) may contain a predetermined fixed string.

[0099] Placeholders may contain information based on information embedded in other placeholders. For example, a placeholder may contain the result of calculations performed on information embedded in multiple placeholders. For instance, the tax-exclusive price ([ITEM_TOTAL_PRICE_WITHOUT_TAX]) in the total column may contain the sum of the subtotal (tax-exclusive) columns ([ITEM_PRICE_1] to [ITEM_PRICE_5]). Alternatively, the total price ([ITEM_TOTAL_PRICE]) in the total column may contain the sum of the tax-exclusive price ([ITEM_TOTAL_PRICE_WITHOUT_TAX]) and the consumption tax ([TOTAL_TAX]).

[0100] <Specific examples of target information> Figure 4 shows an example of the target information. Figure 4 shows a quotation, which is an example of the target information.

[0101] As shown in Figure 4, the target information 310 may include information corresponding to the items of the document template 300 shown in Figure 3. For example, the target information 310 may include the subject, item (corresponding to "deliverables"), unit price, number of people (corresponding to "quantity"), amount (corresponding to "subtotal (excluding tax)"), remarks, total (excluding tax, consumption tax, total amount), etc., which are included in the document template 300.

[0102] The correspondence between items in the target information and items in the document template may be explicitly indicated in the document template or model input information. For example, the document template may include information listing the names of items that can be used as each item. Alternatively, the correspondence between items in the target information and items in the document template may be interpreted by a machine learning model M. As a reference for the machine learning model M to interpret the correspondence, the model input information may include instructional information containing rules for interpreting the correspondence between items.

[0103] <Example of a prompt template> Figure 5 shows an example of a prompt template. Figure 5 shows a prompt template for generating an invoice, which is an example of a document, based on a quotation, which is an example of the target information.

[0104] As shown in Figure 5, the prompt template 320 may include instruction information 321, placeholder 322, constraint information 323, placeholder 324, and instruction information 325. Instruction information 321, 325, and constraint information 323 may be predetermined fixed statements. Placeholder 322 is a placeholder for embedding target information. Placeholder 324 is a placeholder for embedding item information.

[0105] Instruction information 321 may include information that instructs the extraction of information necessary to create an invoice from a quotation (specifically, "Please extract the information necessary to create an invoice from the following quotation"). Therefore, instruction information 321 is an example of the first instruction information.

[0106] The placeholder 322 may include a placeholder {QUOTATION} for embedding target information. As an example, the source code of a quotation described in HTML format may be embedded in the placeholder {QUOTATION}. As another example, the image data of a quotation created in PDF format or the like may be embedded in the placeholder {QUOTATION}. A plurality of target information may be embedded in the placeholder {QUOTATION}. As an example, information to be embedded in an invoice can be extracted from a plurality of quotations.

[0107] The constraint information 323 may include information specifying the format of the model output information (specifically, "Please output the extracted information as JSON data"). The constraint information 323 may include information specifying the format for each data type (specifically, "Regarding the date, use the notation 'YYYY / MM / DD', and regarding the amount, include the unit as necessary").

[0108] The placeholder 324 may include a placeholder {INVOICE_TEMPLATE} for embedding one or more items. As an example, a list of placeholders included in an invoice template may be embedded in the placeholder {INVOICE_TEMPLATE}. Explanation information, constraint information, etc. of the placeholders included in the invoice template may be embedded in the placeholder {INVOICE_TEMPLATE}. As an example, when the document template 300 shown in FIG. 3 is specified, strings such as {INVOICE_ID, ISSUE_DATE, DUE_DATE, INVOICE_TITLE ···} may be embedded in the placeholder {INVOICE_TEMPLATE}.

[0109] As another example, the source code of an invoice template described in HTML format may be embedded in the placeholder {INVOICE_TEMPLATE}. As another example, the image data of an invoice created in PDF format or the like may be embedded in the placeholder {INVOICE_TEMPLATE}.

[0110] Instruction information 325 may include information that instructs the system to output empty information if it is not possible to extract the necessary information to create an invoice from the quotation (specifically, "If a property is not found, it is not necessary to include that property in the JSON data"). Thus, instruction information 325 is an example of a third type of instruction information.

[0111] If the machine learning model M cannot obtain information corresponding to items contained in the document from the target information, it may output incorrect or unreliable information. By including a third set of instructional information in the model input information, the possibility of incorrect or unreliable information being included in the model output information can be suppressed.

[0112] Figure 5 shows a prompt template for generating an invoice based on a quotation, but the prompt template can be modified as needed depending on the combination of target information and document. As another example, a prompt template for generating a recruitment email, which is an example of a document, based on company information, which is an example of target information, and a candidate's resume, may be written as follows.

[0113] ========== The following is our company information.

[0114] {COMPANY_OVERVIEW} The following is a resume of a candidate we are interested in hiring.

[0115] {RESUME} Please review our company information and the resume of this individual, and write a compelling recruitment email for them.

[0116] Additionally, please extract the following information necessary for sending recruitment emails (candidate name, email address, and content to include in the email) from the resume. Please output the extracted information as JSON data.

[0117] [HR_CANDIDATE_NAME] [HR_CANDIDATE_EMAIL] [HR_SCOUT_MESSAGE] ==========

[0118] {COMPANY_OVERVIEW} and {RESUME} are examples of placeholders for embedding target information. "Write an attractive recruitment email for this person" is an example of a second set of instructions. "Extract the following information (candidate name, email address, and content to include in the email) from the resume necessary for sending the recruitment email" is an example of a first set of instructions. You may also use the model output information generated using this prompt template to generate the output information necessary for sending the recruitment email. In that case, you may use a recruitment email template that includes [HR_CANDIDATE_NAME], [HR_CANDIDATE_EMAIL], and [HR_SCOUT_MESSAGE] as placeholders to generate the output information.

[0119] <Document generation process flow> The document generation process performed by the information processing system 1000 will be explained with reference to Figure 6. Figure 6 is a flowchart showing an example of the document generation process.

[0120] In step S1, the terminal device 30 sends a generation request to the document generation device 10 in response to user operation. The generation request includes the document template and target information specified by the user.

[0121] The document generation device 10 receives a generation request from the terminal device 30. The request receiving unit 110 of the document generation device 10 receives the generation request received from the terminal device 30. The request receiving unit 110 sends the generation request to the item acquisition unit 120 and the information acquisition unit 130.

[0122] In step S2, the item acquisition unit 120 of the document generation device 10 receives a generation request from the request reception unit 110. The item acquisition unit 120 acquires a document template from the generation request. The item acquisition unit 120 acquires one or more items from the document template. The item acquisition unit 120 sends information indicating one or more items to the generation unit 140. The item acquisition unit 120 also sends the document template to the embedding unit 150.

[0123] In step S3, the information acquisition unit 130 of the document generation device 10 receives a generation request from the request reception unit 110. The information acquisition unit 130 acquires the target information from the generation request. The information acquisition unit 130 sends the target information to the generation unit 140.

[0124] In step S4, the generation unit 140 of the document generation device 10 receives information indicating one or more items from the item acquisition unit 120. The generation unit 140 also receives target information from the information acquisition unit 130.

[0125] The generation unit 140 generates model input information based on one or more items and target information. The generation unit 140 may also generate model input information by embedding one or more items and target information into a predetermined prompt template. The generation unit 140 transmits the model input information to the generation device 20.

[0126] In step S5, the generation device 20 receives model input information from the document generation device 10. The generation device 20 inputs the model input information received from the document generation device 10 into the machine learning model M. The machine learning model M performs a predetermined task based on the model input information and outputs the resulting model output information. The generation device 20 transmits the model output information output from the machine learning model M to the document generation device 10.

[0127] The document generation device 10 receives model output information from the generation device 20. The generation unit 140 of the document generation device 10 acquires the model output information received from the generation device 20. The generation unit 140 sends the model output information to the embedding unit 150.

[0128] In step S6, the embedding unit 150 of the document generation device 10 receives model output information from the generation unit 140. The embedding unit 150 also receives a document template from the item acquisition unit 120.

[0129] The embedding unit 150 generates document data based on the model output information. The embedding unit 150 may also generate document data by embedding the model output information into a document template. The embedding unit 150 may extract information corresponding to one or more items from the model output information and embed the extracted information into placeholders in the document template. The embedding unit 150 may generate information corresponding to one or more items based on the model output information and embed the generated information into placeholders in the document template. The embedding unit 150 may generate information not based on the model output information and embed the generated information into placeholders in the document template.

[0130] The embedded unit 150 may process the information extracted from the model output information. The embedded unit 150 may verify the information extracted from the model output information. The embedded unit 150 may correct the information extracted from the model output information. The embedded unit 150 may present the information extracted from the model output information to the user. The embedded unit 150 may accept corrections to the information by the user. The embedded unit 150 stores the document data and the information used to generate the document data in the document storage unit 160.

[0131] In step S7, the embedding unit 150 of the document generation device 10 presents the generation result in response to the generation request to the user. The embedding unit 150 may present the user with model output information received from the generation unit 140. The embedding unit 150 may present the user with information extracted from the model output information. The embedding unit 150 may present the user with document data in which the model output information is embedded. The embedding unit 150 may also present the generation result to the user by transmitting the generation result, which includes at least one of the model output information, information extracted from the model output information, or document data in which the model output information is embedded, to the terminal device 30.

[0132] The embedding unit 150 may accept corrections to the information presented to the user in response to user operations. The embedding unit 150 may accept corrections to model output information. The embedding unit 150 may accept corrections to information extracted from model output information. The embedding unit 150 may accept corrections to document data in which model output information is embedded. The embedding unit 150 may verify the corrected information.

[0133] In step S8, the document generation device 10 determines whether or not to regenerate the document data. For example, if the document generation device 10 accepts a correction to the information presented to the user in step S7, it may decide to regenerate the document data. If the document data is to be regenerated (YES), the document generation device 10 returns to step S6. On the other hand, if the document data is not to be regenerated (NO), the document generation device 10 proceeds to step S9.

[0134] As another example, the document generation device 10 may decide to regenerate the document data if the user performs an operation related to regeneration. An operation related to regeneration may be, for example, an operation in which the user re-selects target information or a document template. If the target information or a document template is re-selected (YES), the document generation device 10 may return to step S1.

[0135] In step S9, the terminal device 30 transmits a request to the document generation device 10 to acquire document data in response to user operation. The acquisition request includes information indicating the document data specified by the user.

[0136] The document generation device 10 receives a request to acquire document data from the terminal device 30. The output unit 170 of the document generation device 10 reads the document data indicated in the acquisition request from the document storage unit 160. The output unit 170 transmits the document data read from the document storage unit 160 to the terminal device 30.

[0137] The terminal device 30 receives document data from the document generation device 10. The terminal device 30 stores the received document data in its storage device. The terminal device 30 may also display the document data stored in its storage device on a display device.

[0138] [Second Embodiment] In the first embodiment, a configuration was described in which model input information for a machine learning model M is generated based on target information specified by the user. The information processing system 1000 may apply a technique called Retrieval Augmented Generation (RAG). Retrieval Augmented Generation is a technique that includes reference information obtained by searching a predetermined data source in the prompt in order to obtain good output results from a large-scale language model.

[0139] In search extension generation, reference information is searched from a predetermined data source based on instruction information that directs the execution of a task, and a prompt containing the search results for both the instruction information and the reference information is input to a large-scale language model. The large-scale language model can execute the task while considering reference information that is not included in the training data, and is therefore able to generate appropriate data for the instruction information.

[0140] The following describes the information processing system 1000 according to the second embodiment, focusing on the differences from the first embodiment. Unless otherwise specified, the information processing system 1000 according to the second embodiment may be configured in the same way as the first embodiment.

[0141] <Overall configuration of the information processing system> The overall configuration of the information processing system according to this embodiment will be described with reference to Figure 7. Figure 7 is a block diagram showing an example of the overall configuration of the information processing system according to the second embodiment.

[0142] As shown in Figure 7, the information processing system 1000 includes a document generation device 10, a generation device 20, a terminal device 30, and a search device 40. The information processing system 1000 according to the second embodiment differs from the first embodiment in that it further includes a search device 40.

[0143] The search device 40 is an example of an information processing device such as a personal computer, workstation, or server that searches for target information. The search device 40 may search for target information in response to a search request from the document generation device 10. The search device 40 may receive a search request from the document generation device 10. The search device 40 may transmit the search results for the search request to the document generation device 10.

[0144] The search device 40 may include a data source D that stores various types of information that are candidates for target information. For example, the data source D may be a storage device or database containing information that is a candidate for target information. The data source D may also be provided by an external storage device, information processing device, or information processing system.

[0145] The search device 40 may be implemented using multiple devices or systems having different data sources D. The search device 40 may be implemented using one device or system having multiple data sources D. The search device 40 may be an external device or system. The search device 40 may retrieve target information from an external data source D.

[0146] The data source D may be built into the document generation device 10 or the generation device 20. The data source D may also be distributed and held in an external information processing system consisting of multiple devices. In this case, the information processing system 1000 does not need to include the search device 40.

[0147] The document generation device 10 according to this embodiment may obtain search results for target information from the search device 40. The document generation device 10 may transmit a search request to the search device 40 based on a generation request received from the terminal device 30. The generation request may include search conditions for the target information. The search conditions may be specified by the user.

[0148] A search request is information or a signal that requests the retrieval of target information. A search request may, for example, include at least one of a query or search criteria. A query may be generated based on at least one of the item information or target information included in a generation request. Search criteria may include search criteria included in a generation request. Search criteria may include predetermined search criteria.

[0149] Multiple search devices 40 may be included in the information processing system 1000. The search device 40 may be implemented by multiple computers, or as a cloud computing service. The search device 40 may also be implemented by a standalone computer in conjunction with at least one of the document generation device 10 or generation device 20.

[0150] <Functional Configuration of Document Generation Device> The functional configuration of the document generation device 10 will be explained with reference to Figure 8. Figure 8 is a block diagram showing an example of the functional configuration of the document generation device according to the second embodiment.

[0151] As shown in Figure 8, the document generation device 10 comprises a request receiving unit 110, an item acquisition unit 120, an information acquisition unit 130, a search unit 135, a generation unit 140, an embedding unit 150, a document storage unit 160, and an output unit 170. The document generation device 10 according to the second embodiment differs from the first embodiment in that it further comprises a search unit 135.

[0152] The search unit 135 searches for one or more target information. For example, the search unit 135 may transmit a search request to the search device 40. The search unit 135 may also receive the search results transmitted by the search device 40 in response to the search request. The search results may include one or more target information that satisfies the search conditions indicated in the search request.

[0153] The search unit 135 may set the search conditions included in the search request based on the target information included in the generation request. The search unit 135 may also set the search conditions specified by the user in the search request.

[0154] The search criteria may include, for example, information indicating the data source to be searched, the type of data source, the time range of the creation date, attribute information of the reference information to be searched, or at least one of the number of results to include in the search results.

[0155] Information indicating the data source may be any information that can identify the data source. This information may include, for example, identification information that identifies the data source, or information that indicates the location of the data source (e.g., hostname, IP (Internet Protocol) address, connection string, URL (Uniform Resource Locator), etc.).

[0156] The search unit 135 may search for target information from multiple data sources indicated in the search conditions. The search unit 135 may include the number of target information items indicated in the search conditions in the search results. The search unit 135 may include a predetermined number of target information items in the search results. The search unit 135 may set the number of target information items to be included in the search results according to the amount of data that can be input into the machine learning model M.

[0157] The search unit 135 may present to the user information regarding the search results for the target information. The search unit 135 may present the search results for the target information in a way that allows the user to select which ones to select. The information regarding the search results may include, as an example, data source, link information, at least some of the information, summary, thumbnail image, creator, creation date, or at least one of the terms of use.

[0158] The search unit 135 may transmit the search results for the target information to the terminal device 30 and display the search results for the target information on the display device of the terminal device 30. The search unit 135 may also display the search results for the target information on the display device of the document generation device 10.

[0159] The search unit 135 may present information related to the search results (hereinafter also referred to as supplementary information for the target information) along with the search results for the target information. The supplementary information for the target information may include, as an example, at least one of the following: data source, link information, partial information, summary, thumbnail image, creator, creation date, or terms of use. The supplementary information for the target information may present any combination of two or more of these. The search unit 135 may generate at least a portion of the supplementary information for the target information based on a machine learning model M or another machine learning model.

[0160] The data source is information indicating the data source D in which the target information was stored. The link information is information indicating the location of the target information (e.g., URL). Some information may include, for example, the title or heading of the target information, a predetermined number of characters or lines from the beginning, or at least one of the keywords extracted from the target information. The summary or thumbnail image may be obtained from the data source D, for example, or generated based on the machine learning model M or another machine learning model.

[0161] The search unit 135 may accept the selection of target information. The search unit 135 may receive a request to select target information from the terminal device 30. The selection request may be transmitted from the terminal device 30 in response to an operation on the screen displayed on the display device of the terminal device 30. The search unit 135 may accept a selection request input to the document generation device 10. The selection request may be input to the document generation device 10 in response to an operation on the screen displayed on the display device of the document generation device 10.

[0162] A selection request is information or a signal that requests the selection of one or more target information to be used in generating document data. A selection request may also be information indicating one or more target information selected by the user from among the search results of the target information presented by the search unit 135.

[0163] The search unit 135 may acquire the target information indicated in the selection request. The search unit 135 may acquire information that can identify the target information. The information that can identify the target information may, for example, be identification information that identifies the target information, or information indicating the location of the file in which the target information is described.

[0164] A selection request may include one or more target information specified by the user. The target information specified by the user may not be included in the search results for the target information. The target information specified by the user may, for example, be information indicating the location of the target information (e.g., a URL).

[0165] <Application Fields> The information processing system 1000 according to each of the above embodiments can be applied to document generation in the following fields, for example. Note that all of the following examples can also be applied to search-enhanced generation.

[0166] The first example is a configuration that creates an invoice based on a quotation. In this case, the target information may be an issued quotation. Alternatively, the item information may be an invoice template.

[0167] The second example is a structure for creating a customer proposal slide for a new product. In this case, the target information may include product information and the customer's business plan information. Alternatively, the item information may include a proposal template.

[0168] The third example is a configuration for creating monthly reports for customers. In this case, the target information could be the minutes of meetings with customers. Alternatively, the item information could be a template for the monthly report.

[0169] The fourth example is a configuration for conducting a competitive analysis. In this case, the target information may be the search results of competitive information published on the web. Alternatively, the item information may be a template for the analysis report.

[0170] The fifth example is a configuration for creating a patent search or patent summary. In this case, the target information may be specified as a patent gazette or published gazette. Alternatively, the item information may be specified as a template for a search report or summary.

[0171] The sixth example is a configuration for creating a scouting email or candidate summary. In this case, candidate information may be specified as the target information. Alternatively, a template for the email or summary may be specified as the item information.

[0172] The seventh example is a configuration for performing an anti-social forces check. In this case, the target information may be the registration records or web information of the trading partner. Alternatively, the item information may be a list of check items.

[0173] The eighth example is a configuration for summarizing or analyzing user feedback or surveys. In this case, user feedback or surveys may be specified as the target information. Alternatively, a template for the summary or analysis results may be specified as the item information.

[0174] The ninth example is a structure that summarizes meeting minutes or meeting audio. In this case, the target information may be specified as meeting minutes or a recording of the meeting. Alternatively, a summary template may be specified as item information.

[0175] The tenth example is a configuration for creating a summary of IR (Investor Relations) information for a specific industry (for example, the automotive industry). In this case, the target information may be the IR information of each company in the industry. Alternatively, a summary template may be specified as the item information. The target information may also be the search results for IR information. The search criteria may be set based on the template or set by the user.

[0176] This section provides a more detailed explanation of the configuration for creating patent summaries in a configuration that applies search extension generation. The summary template may include the following items and placeholders.

[0177] 1. Title, "TITLE" 2. Patent number, "PATENT NO" 3. Patent holder, "ASSIGNEE" 4. Summary of the Invention 5. The challenge, "PROBLEM" 6.Representative claim, “MAIN CAIM” 7. Representative diagram, "MAIN FUGURE" 8. Conventional technology, "PRIOR ART" 9. Overview of Conventional Technology, "SUMMARY OF PRIOR ART" 10.DIFFERENCE FROM PRIOR ART 11. Overview of the selection process, "PROSECUTION HISTORY" The search criteria for the target information may include a condition for searching for prior art patent publications in the patent database. The search criteria for prior art patent publications may also include various numbers (for example, application number, publication number, patent number, etc.) extracted from the patent publications specified by the user according to predetermined rules.

[0178] The search criteria for the target information may include conditions for searching for examination progress information from the patent database. Examination progress information may include, for example, notices of reasons for rejection, written opinions, and amendments to the application. The search criteria for examination progress information may also include various numbers included in the patent gazette specified by the user (for example, application number, publication number, patent number, etc.).

[0179] The model input information may include constraint information specifying the number of characters (for example, 150 characters or less) for the summary information to be included in the summary template (e.g., summary of the invention, summary of prior art, summary of the examination process, etc.). The model input information may also include constraint information specifying the language to be used. If a language is specified in the model input information, patent publications written in other languages ​​will be translated into the specified language and output.

[0180] The model input information may include target information retrieved from a patent database. This target information may include, for example, at least one of either a patent publication or examination progress information.

[0181] Model input information may include instruction information that instructs the model to extract information corresponding to one or more items from the target information. For example, instruction information may be natural language sentences such as, "Please extract the information necessary to create a patent summary from the following information."

[0182] Model input information may include placeholders for item names included in the summary template. Model input information may also include descriptive information or constraint information for the items corresponding to the placeholders.

[0183] <Summary> As is clear from the above description, an information processing system 1000 according to one embodiment of the present disclosure acquires one or more items, acquires target information, generates input information for a machine learning model based on one or more items and target information, acquires information generated by inputting the input information into the machine learning model, and generates output information including information corresponding to one or more items based on the generated information.

[0184] The information processing system 1000 may obtain one or more items based on an output information template. The information processing system 1000 may generate output information by embedding the generated information into a template. The information processing system 1000 may select a template from multiple templates.

[0185] The input information may include a first instruction that instructs the system to extract information corresponding to one or more items from the target information. The input information may also include a second instruction that instructs the system to process cases where information corresponding to one or more items cannot be extracted from the target information. The second instruction may include information that instructs the system to output at least one of the following as information corresponding to items from which information cannot be extracted: empty information, an initial value, or information newly generated by a machine learning model.

[0186] The information processing system 1000 may display items from which information has been extracted from the target information and items from which information has not been extracted in different display formats. The input information may include information about the format of the information generated by the machine learning model. The format may use one or more items.

[0187] The information processing system 1000 may search for second target information and generate input information for a machine learning model based on one or more items, target information, and the second target information. The search conditions for the second target information may be set based on the aforementioned target information.

[0188] The information processing system 1000 may verify the information generated by the machine learning model. The information processing system 1000 may present the verification results to the user.

[0189] One or more items may also be identification information for one or more items. Identification information for one or more items may include at least one of the following: the names of one or more items, descriptive information for one or more items, or one or more placeholders corresponding to each of the one or more items.

[0190] The information processing system 1000 may generate information by inputting input information into a machine learning model. The generated information may include information extracted from the target information.

[0191] As a result, according to one embodiment of this disclosure, information corresponding to items included in the output information can be generated. In one aspect, according to this embodiment, since information corresponding to one or more items can be generated based on the target information, information corresponding to items included in the output information can be generated. In another aspect, according to this embodiment, since information to be included in the output information is generated based on a machine learning model, various formats of output information can be easily generated.

[0192] For example, if you want to change only the output format of a document while maintaining the information contained within it, you can do so simply by replacing the document template with the new output format. For example, if you want to generate a large number of documents with the same output format, you can generate a large number of documents through batch processing by sending generation requests with different target information sequentially or in parallel. For example, since users can specify the information to be generated by the machine learning model, the machine learning model can appropriately determine what information to extract from the target information and extract the appropriate information from it. For example, in a series of processes including inputting target information, displaying model output information, displaying the document, and saving the document, it is possible to check and correct the information at each step, allowing for flexible document generation.

[0193] [Hardware configuration of information processing equipment] In the embodiments described above, some or all of the devices (document generation device 10, generation device 20, and terminal device 30) may be composed of hardware, or they may be composed of information processing by software (programs) executed by a CPU (Central Processing Unit), GPU (Graphics Processing Unit), etc. If the information processing is composed of software, the software that realizes at least some of the functions of each device in the embodiments described above may be stored on a non-temporary storage medium (non-temporary computer-readable medium) such as a CD-ROM (Compact Disc-Read Only Memory) or USB (Universal Serial Bus) memory, and the information processing of the software may be executed by loading it into a computer. Alternatively, the software may be downloaded via a communication network. Furthermore, all or part of the processing of the software may be implemented in a circuit such as an ASIC (Application Specific Integrated Circuit) or FPGA (Field Programmable Gate Array), so that the information processing by the software is executed by hardware.

[0194] The storage medium for the software may be a removable medium such as an optical disc, or a fixed storage medium such as a hard disk or memory. Furthermore, the storage medium may be located inside the computer (main memory, auxiliary storage, etc.) or outside the computer.

[0195] Figure 9 is a block diagram showing an example of the hardware configuration of each device (document generation device 10, generation device 20, and terminal device 30) in the embodiment described above. Each device may be implemented as a computer 7, for example, comprising a processor 71, main memory 72 (memory), auxiliary memory 73 (memory), network interface 74, and device interface 75, which are connected via a bus 76.

[0196] The computer 7 in Figure 9 has one of each component, but it may have multiple identical components. Also, although Figure 9 shows one computer 7, the software may be installed on multiple computers, and each of these computers may execute the same or different parts of the software's processing. In this case, it may be a distributed computing configuration in which each computer communicates via a network interface 74 or the like to execute processing. In other words, each device in the above-described embodiment (document generation device 10, generation device 20, and terminal device 30) may be configured as a system that realizes its function by one or more computers executing instructions stored in one or more storage devices. Alternatively, it may be configured so that information transmitted from the terminal is processed by one or more computers located on the cloud, and the processing results are transmitted to the terminal.

[0197] The various calculations performed by each device (document generation device 10, generation device 20, and terminal device 30) in the embodiments described above may be performed in parallel using one or more processors, or using multiple computers via a network. Alternatively, the various calculations may be distributed to multiple processing cores within a processor and performed in parallel. Furthermore, some or all of the processing and means of this disclosure may be implemented by at least one of a processor and a storage device located on a cloud that can communicate with computer 7 via a network. Thus, each device in the embodiments described above may be implemented in the form of parallel computing using one or more computers.

[0198] The processor 71 may be an electronic circuit (processing circuit, processing circuitry, CPU, GPU, FPGA, ASIC, etc.) that performs either control or calculations of a computer. The processor 71 may also be a general-purpose processor, a dedicated processing circuit designed to perform specific calculations, or a semiconductor device that includes both a general-purpose processor and a dedicated processing circuit. Furthermore, the processor 71 may include optical circuits or quantum computing-based calculation functions.

[0199] The processor 71 may perform calculations based on data and software input from various devices within the computer 7, and may output calculation results and control signals to these devices. The processor 71 may also control the various components of the computer 7 by executing the computer 7's OS (Operating System) or applications.

[0200] Each of the devices in the above-described embodiment (document generation device 10, generation device 20, and terminal device 30) may be implemented by one or more processors 71. Here, processor 71 may refer to one or more electronic circuits arranged on one chip, or one or more electronic circuits arranged on two or more chips or two or more devices. When multiple electronic circuits are used, each electronic circuit may communicate by wire or wireless.

[0201] The main memory 72 may store instructions executed by the processor 71 and various data, and the information stored in the main memory 72 may be read by the processor 71. The auxiliary memory 73 is a memory device other than the main memory 72. These memory devices refer to any electronic component capable of storing electronic information, and may be semiconductor memory. The semiconductor memory may be either volatile memory or non-volatile memory. In the embodiments described above, the memory devices for storing various data in each device (document generation device 10, generation device 20, and terminal device 30) may be implemented by the main memory 72 or the auxiliary memory 73, or by the built-in memory of the processor 71. For example, each storage unit in the embodiments described above may be implemented by the main memory 72 or the auxiliary memory 73.

[0202] In the embodiments described above, if each device (document generation device 10, generation device 20, and terminal device 30) consists of at least one storage device (memory) and at least one processor connected to (coupled with) this at least one storage device, then at least one processor may be connected to one storage device. Also, at least one storage device may be connected to one processor. Furthermore, the configuration may include at least one processor among a plurality of processors being connected to at least one storage device among a plurality of storage devices. This configuration may also be realized by storage devices and processors included in a plurality of computers. Furthermore, the configuration may include a storage device integrated with a processor (for example, a cache memory including an L1 cache and an L2 cache).

[0203] The network interface 74 is an interface for connecting to the communication network 8 wirelessly or via a wired connection. The network interface 74 can be any appropriate interface, such as one conforming to existing communication standards. Information may be exchanged between the computer 7 and an external device 9A connected via the communication network 8 through the network interface 74. The communication network 8 may be a WAN (Wide Area Network), LAN (Local Area Network), PAN (Personal Area Network), or a combination thereof, as long as information is exchanged between the computer 7 and the external device 9A. An example of a WAN is the Internet, an example of a LAN is IEEE 802.11 or Ethernet (registered trademark), and an example of a PAN is Bluetooth (registered trademark) or NFC (Near Field Communication).

[0204] The device interface 75 is an interface such as USB that connects directly to the external device 9B.

[0205] External device 9A is a device connected to computer 7 via a network. External device 9B is a device directly connected to computer 7.

[0206] External device 9A or external device 9B may, for example, be an input device. The input device may be a camera, microphone, motion capture device, various sensors, keyboard, mouse, touch panel, etc., and provides the acquired information to the computer 7. Alternatively, it may be a device equipped with an input unit, memory, and processor, such as a personal computer, tablet terminal, or smartphone.

[0207] Furthermore, external device 9A or external device 9B may, for example, be an output device. The output device may be a display device such as an LCD (Liquid Crystal Display) or an organic EL (Electro Luminescence) panel, or a speaker that outputs sound, etc. It may also be a device equipped with an output unit, memory, and a processor, such as a personal computer, tablet terminal, or smartphone.

[0208] Furthermore, external devices 9A and 9B may be storage devices (memory). For example, external device 9A may be network storage, and external device 9B may be storage such as an HDD.

[0209] Furthermore, the external device 9A or external device 9B may be a device that has some of the functions of the components of each device (document generation device 10, generation device 20, and terminal device 30) in the embodiment described above. In other words, the computer 7 may transmit some or all of the processing results to the external device 9A or external device 9B, or may receive some or all of the processing results from the external device 9A or external device 9B.

[0210] In this specification (including the claims), when the expression "at least one of a, b, and c" or "at least one of a, b, or c" (including similar expressions) is used, it includes any of a, b, c, ab, ac, bc, or abc. Furthermore, any element may have multiple instances, such as aa, abb, aabbcc, etc. In addition, it is also possible to add other elements other than the enumerated elements (a, b, and c), such as abcd which has d.

[0211] In this specification (including the claims), when expressions such as "using data as input / based on data / according to / in accordance with data" (including similar expressions) are used, unless otherwise specified, this includes using the data itself or using data that has been processed in some way (e.g., data with added noise, normalized data, features extracted from the data, intermediate representations of the data, etc.). Furthermore, when it is stated that some result is obtained "using data as input / based on data / according to / in accordance with data" (including similar expressions), unless otherwise specified, this includes cases where the result is obtained based solely on the data in question or where the result is influenced by other data, factors, conditions, and / or states other than the data in question. Furthermore, when it is stated that "data is output" (including similar expressions), unless otherwise specified, this includes cases where the data itself is used as output or where data that has been processed in some way (e.g., data with added noise, normalized data, features extracted from the data, intermediate representations of various types of data, etc.) is used as output.

[0212] In this specification (including the claims), the terms “connected” and “coupled” are intended to be non-restrictive terms that include any direct connection / coupling, indirect connection / coupling, electrical connection / coupling, communicative connection / coupling, operational connection / coupling, physical connection / coupling, etc. The terms should be interpreted as appropriate in the context in which they are used, but any form of connection / coupling that is not intentionally or naturally excluded should be interpreted non-restrictively as being included in the terms.

[0213] In this specification (including the claims), when the expression "A configured to B" is used, it may include that the physical structure of element A has a configuration capable of performing operation B, and that the permanent or temporary setting / configuration of element A is configured to actually perform operation B. For example, if element A is a general-purpose processor, it is sufficient that the processor has a hardware configuration capable of performing operation B, and that it is configured to actually perform operation B by the setting of a permanent or temporary program (instruction). Furthermore, if element A is a dedicated processor, dedicated arithmetic circuit, etc., it is sufficient that the circuit structure of the processor is implemented to actually perform operation B, regardless of whether control instructions and data are actually attached.

[0214] Wherever terms meaning "comprising" or "possessing" (e.g., "comprising / including," "having," etc.) are used herein, they are intended to be open-ended terms, including cases where the subject matter of such terms is not the object of the term. Where the object of such terms meaning "comprising" or "possessing" is an expression that does not specify a quantity or suggests a singular number (an expression with the article "a" or "an"), such expression should be interpreted as not being limited to a specific number.

[0215] In this specification (including the claims), even if expressions such as "one or more" or "at least one" are used in some places, and expressions that do not specify a quantity or suggest a singularity (expressions using the articles a or an) are used in other places, the latter expressions are not intended to mean "one." In general, expressions that do not specify a quantity or suggest a singularity (expressions using the articles a or an) should not necessarily be interpreted as not being limited to a specific number.

[0216] In this specification, if a particular configuration of an embodiment is described as having a specific advantage or result, it should be understood, unless otherwise stated, that the same advantage or result can also be obtained from one or more other embodiments having that configuration. However, it should be understood that the presence or absence of such advantage or result generally depends on various factors, conditions, and / or states, and that the configuration does not necessarily guarantee that the advantage or result can be obtained. The advantage or result can only be obtained from the configuration described in the embodiment when various factors, conditions, and / or states are met, and the advantage or result cannot necessarily be obtained in the invention claimed to define that configuration or a similar configuration.

[0217] In this specification (including the claims), when multiple hardware components perform a predetermined process, each component may cooperate to perform the predetermined process, or some components may perform all of the predetermined process. Alternatively, some components may perform part of the predetermined process, while other components perform the remainder. In this specification (including the claims), when expressions such as "one or more hardware components perform a first process, and the one or more hardware components perform a second process" (including similar expressions) are used, the hardware component performing the first process and the hardware component performing the second process may be the same or different. In other words, it is sufficient that the hardware component performing the first process and the hardware component performing the second process are included in the one or more hardware components. Hardware may include electronic circuits, devices containing electronic circuits, etc.

[0218] In this specification (including the claims), when multiple memory devices store data, each of the multiple memory devices may store only a portion of the data or the entire data. Furthermore, a configuration in which some of the multiple memory devices store data is also included.

[0219] In this specification (including the claims), terms such as “first,” “second,” etc., are used merely as a way of distinguishing between two or more elements and are not necessarily intended to impose technical meanings such as temporal, spatial, order, or quantity on the subject. Therefore, for example, references to a first element and a second element do not necessarily mean that only two elements can be employed therein, that the first element must precede the second element, or that the first element must exist for the second element to exist.

[0220] While embodiments of this disclosure have been described in detail above, this disclosure is not limited to the individual embodiments described above. Various additions, modifications, substitutions, and partial deletions are possible, provided that they do not depart from the conceptual idea and spirit of the present invention derived from the claims and their equivalents. For example, where numerical values ​​or mathematical formulas are used in the description of the embodiments described above, these are provided for illustrative purposes only and do not limit the scope of this disclosure. Similarly, the sequence of operations shown in the embodiments is also illustrative and does not limit the scope of this disclosure.

[0221] Furthermore, the following forms are possible for disclosure technology.

[0222] (Note 1) At least one memory, Equipped with at least one processor, The aforementioned at least one processor is Obtain one or more items, Obtain the target information, Based on the one or more items mentioned above and the target information, input information for the machine learning model is generated. The information generated by inputting the aforementioned input information into the machine learning model is obtained. Based on the generated information, output information is generated that includes information corresponding to one or more of the above items. Information processing system.

[0223] (Note 2) The aforementioned at least one processor is Based on the template of the output information, one or more items are obtained. The information processing system described in Appendix 1.

[0224] (Note 3) The aforementioned at least one processor is The output information is generated by embedding the generated information into the template. The information processing system described in Appendix 2.

[0225] (Note 4) The input information includes first instruction information that instructs the extraction of information corresponding to one or more items from the target information. An information processing system as described in any of the appendices 1 to 3.

[0226] (Note 5) The input information includes a second instruction information that instructs the processing to be performed when it is not possible to extract information corresponding to one or more items from the target information. The information processing system described in Appendix 4.

[0227] (Note 6) The second instruction information includes, as information corresponding to items from which the information cannot be extracted, information instructing the output of at least one of the following: empty information, an initial value, or information newly generated by the machine learning model. The information processing system described in Appendix 5.

[0228] (Note 7) The aforementioned at least one processor is Of the one or more items mentioned above, items from which information was extracted from the target information and items from which information was not extracted are displayed in different display formats. An information processing system as described in any of the appendices 4 to 6.

[0229] (Note 8) The aforementioned input information includes information regarding the format of the information generated by the machine learning model, The aforementioned format utilizes one or more of the above items. An information processing system as described in any of the appendices 1 to 7.

[0230] (Note 9) The aforementioned at least one processor is Search for the second target information, Based on the one or more items mentioned above, the target information mentioned above, and the second target information, input information for the machine learning model is generated. An information processing system as described in any of the appendices 1 to 8.

[0231] (Note 10) The search conditions for the second target information are set based on the target information. The information processing system described in Appendix 9.

[0232] (Note 11) The aforementioned at least one processor is The information generated by the machine learning model is verified, The results of the aforementioned verification are presented to the user. An information processing system as described in any of the appendices 1 to 10.

[0233] (Note 12) The aforementioned at least one processor is Select the aforementioned template from multiple templates. The information processing system described in Appendix 2 or 3.

[0234] (Note 13) The one or more items mentioned above are identification information for the one or more items mentioned above. An information processing system as described in any one of the items 1 through 12 of the appendix.

[0235] (Note 14) The identification information for the one or more items includes at least the name of the one or more items, descriptive information for the one or more items, or one or more placeholders corresponding to each of the one or more items. The information processing system described in Appendix 13.

[0236] (Note 15) The aforementioned at least one processor is The aforementioned information is generated by inputting the aforementioned input information into the machine learning model. An information processing system as described in any one of the items 1 through 14 of the appendix.

[0237] (Note 16) The generated information includes information extracted from the target information, An information processing system as described in any one of the items 1 through 15 of the appendix.

[0238] (Note 17) At least one processor, Obtain one or more items, Obtain the target information, Based on the one or more items mentioned above and the target information, input information for the machine learning model is generated. The information generated by inputting the aforementioned input information into the machine learning model is obtained. Based on the generated information, output information is generated that includes information corresponding to one or more of the above items. Information processing methods.

[0239] (Note 18) At least one processor, Obtain one or more items, Obtain the target information, Based on the one or more items mentioned above and the target information, input information for the machine learning model is generated. The information generated by inputting the aforementioned input information into the machine learning model is obtained. Based on the generated information, output information is generated that includes information corresponding to one or more of the above items. A program to execute a process. [Explanation of Symbols]

[0240] M: Machine learning model D: Data source 10: Document generation device 20:Generation device 30: Terminal device 40: Search device 110: Request Reception Department 120: Item acquisition part 130: Information acquisition department 135: Search section 140: Generation part 150: Recessed part 160: Document Storage Unit 170: Output section 1000: Information Processing Systems

Claims

1. At least one memory, It comprises at least one processor, The aforementioned at least one processor is Obtain one or more items, Obtain the target information, Based on the one or more items mentioned above and the target information, input information for the machine learning model is generated. The information generated by inputting the aforementioned input information into the machine learning model is obtained. Based on the generated information, output information is generated that includes information corresponding to one or more of the above items. Information processing system.

2. The aforementioned at least one processor is Based on the template of the output information, one or more of the above items are obtained. The information processing system according to claim 1.

3. The aforementioned at least one processor is The output information is generated by embedding the generated information into the template. The information processing system according to claim 2.

4. The input information includes first instruction information that instructs the extraction of information corresponding to one or more items from the target information. The information processing system according to claim 1.

5. The input information includes a second instruction information that instructs the processing to be performed when it is not possible to extract information corresponding to one or more items from the target information. The information processing system according to claim 4.

6. The second instruction information includes, as information corresponding to items from which the information cannot be extracted, information instructing the output of at least one of the following: empty information, an initial value, or information newly generated by the machine learning model. The information processing system according to claim 5.

7. The aforementioned at least one processor is Of the one or more items mentioned above, items from which information was extracted from the target information and items from which information was not extracted are displayed in different display formats. The information processing system according to claim 4.

8. The aforementioned input information includes information regarding the format of the information generated by the machine learning model, The aforementioned format utilizes one or more of the above items. The information processing system according to claim 1.

9. The aforementioned at least one processor is Search for the second target information, Based on the one or more items mentioned above, the target information, and the second target information, input information for the machine learning model is generated. The information processing system according to claim 1.

10. The search conditions for the second target information are set based on the target information. The information processing system according to claim 9.

11. The aforementioned at least one processor is The information generated by the machine learning model is verified, The results of the aforementioned verification are presented to the user. The information processing system according to claim 1.

12. The aforementioned at least one processor is Select the aforementioned template from multiple templates. The information processing system according to claim 2.

13. The one or more items mentioned above are identification information for the one or more items mentioned above. The information processing system according to any one of claims 1 to 12.

14. The identification information for the one or more items includes at least the name of the one or more items, descriptive information for the one or more items, or one or more placeholders corresponding to each of the one or more items. The information processing system according to claim 13.

15. The aforementioned at least one processor is The aforementioned information is generated by inputting the aforementioned input information into the machine learning model. The information processing system according to any one of claims 1 to 12.

16. The generated information includes information extracted from the target information, The information processing system according to any one of claims 1 to 12.

17. At least one processor, Obtain one or more items, Obtain the target information, Based on the one or more items mentioned above and the target information, input information for the machine learning model is generated. The information generated by inputting the aforementioned input information into the machine learning model is obtained. Based on the generated information, output information is generated that includes information corresponding to one or more of the above items. Information processing methods.

18. At least one processor, Obtain one or more items, Obtain the target information, Based on the one or more items mentioned above and the target information, input information for the machine learning model is generated. The information generated by inputting the aforementioned input information into the machine learning model is obtained. Based on the generated information, output information is generated that includes information corresponding to one or more of the above items. A program to execute a process.

Images