# US 2026/0288777 A1 · System

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2026/0288777 A1
- Document ID: US20260288777A1
- Status: Published application
- Kind: A1
- Application number: 19/567,493
- Filed: 2026-03-16
- Published: 2026-09-24
- Earliest claimed: 2025-03-19
- Assignees: SoftBank Group Corp.
- Applicants: SoftBank Group Corp.
- Inventors: Hiroyuki MIZUNO
- Web page: https://patentgrep.com/patent/US20260288777A1

## CPC

- G06F 16/3326: {using relevance feedback from the user, e.g. relevance feedback on documents, documents sets, document terms or passages}
- G06F 16/3344: {using natural language analysis}
- G06F 16/337: {Profile generation, learning or modification}
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

A system includes a processor that is configured to analyze contract information input by a user and generate a prompt for instructing generation of an appropriate contract document based on an analysis result of the contract information, output a generated contract document in a document file format and display the generated contract document via a user interface so that the user can review and edit the generated contract document, and analyze an emotion of the user by using an emotion analysis algorithm to analyze input information of the user and adjust wording of the contract document based on the emotion of the user.

## Claims

### Claim 1

A system comprising:
circuitry configured to:
receive, via a communication interface coupled to a packet-switched network, structured input data from a terminal device, and extract, by applying a natural language processing model to the structured input data, a purpose identifier, a party identifier, a period parameter, and a type identifier;
construct a prompt sentence encoding the purpose identifier, the party identifier, the period parameter, and the type identifier, and input the prompt sentence and the structured input data to a generative neural network model to obtain text output data;
post-process the text output data to assign heading information, clause information, and format information to segments of the text output data, and generate electronic document data conforming to a predetermined document file format based on the post-processed text output data;
store the electronic document data in a storage device, and register the purpose identifier, the party identifier, the period parameter, the type identifier, and metadata associated with the electronic document data as search information in an index database;
transmit the electronic document data to the terminal device via the communication interface so that the electronic document data is displayable and editable at the terminal device;
execute emotion analysis processing on user input information received from the terminal device to calculate emotion information, and adjust at least one of a tone parameter, an expression strength parameter, or a follow-up condition of the prompt sentence based on the emotion information; and
acquire, from the terminal device, a correction instruction, re-input a corrected prompt sentence incorporating the correction instruction and at least a portion of the electronic document data into the generative neural network model to obtain updated text output data, and update the electronic document data based on the updated text output data.

### Claim 2 (depends on 1)

The system according to claim 1, wherein the circuitry is configured to extract the purpose identifier, the party identifier, the period parameter, and the type identifier by tokenizing the structured input data, applying a named entity recognition model to identify entity spans, and mapping the identified entity spans to structured attribute fields.

### Claim 3 (depends on 2)

The system according to claim 2, wherein the circuitry is configured to post-process the text output data by applying a structural parsing algorithm to detect clause boundaries and heading markers, assigning each detected segment a clause identifier and a heading level, and constructing a document tree representation that maps each segment to a position in the predetermined document file format.

### Claim 4 (depends on 2)

The system according to claim 2, wherein the structured input data comprises contract information specifying terms of a legal agreement, the purpose identifier comprises a contract purpose field, the party identifier comprises a contracting party field, and the type identifier specifies a document type selected from a plurality of predetermined document categories.

### Claim 5 (depends on 4)

The system according to claim 4, wherein the circuitry is configured to generate the electronic document data in a document file format that is directly editable at the terminal device, and to output the electronic document data for display via a user interface that allows the user to review and modify individual clauses of the electronic document data.

### Claim 6 (depends on 1)

The system according to claim 1, wherein the circuitry is configured to execute the emotion analysis processing by applying an emotion analysis algorithm to at least one of text input, voice input, or interaction timing data received from the terminal device to compute the emotion information as a multi-dimensional emotional state vector.

### Claim 7 (depends on 6)

The system according to claim 6, wherein the circuitry is configured to apply an emotion identification model to the at least one of text input, voice input, or interaction timing data to output a probability distribution over a plurality of emotional state categories, and to select the tone parameter, the expression strength parameter, or the follow-up condition based on a mapping between the probability distribution and predefined adjustment rules.

### Claim 8 (depends on 6)

The system according to claim 6, wherein the circuitry is configured to adjust the follow-up condition by modifying a notification frequency parameter and a notification timing parameter of an approval follow-up workflow based on the emotional state vector, such that a higher urgency emotional state increases the notification frequency parameter.

### Claim 9 (depends on 1)

The system according to claim 1, wherein the circuitry is configured to acquire, from the terminal device, a search condition relating to prior document information, search the index database using the search condition to retrieve related document records, and transmit the retrieved related document records to the terminal device to augment construction of the prompt sentence.

### Claim 10 (depends on 9)

The system according to claim 9, wherein the circuitry is configured to incorporate retrieved related document records into the prompt sentence by generating a reference clause set derived from the related document records, and encoding the reference clause set as additional context in the prompt sentence input to the generative neural network model.

### Claim 11 (depends on 1)

The system according to claim 1, wherein the circuitry is configured to update the index database after each update to the electronic document data by re-extracting the purpose identifier, the party identifier, the period parameter, and the type identifier from the updated electronic document data, and updating the search information stored in the index database with the re-extracted values.

### Claim 12 (depends on 1)

The system according to claim 1, wherein the circuitry is configured to generate a plurality of text output candidates by executing the generative neural network model with varied sampling parameters, score each candidate based on a structural completeness metric derived from the heading information and clause information, and select the highest-scoring candidate as the text output data.

### Claim 13 (depends on 1)

The system according to claim 1, wherein the circuitry is configured to apply a retrieval-augmented generation architecture by querying a vector database using an embedding vector derived from the structured input data to retrieve reference documents, and incorporating the retrieved reference documents into the prompt sentence.

### Claim 14 (depends on 1)

The system according to claim 1, wherein the circuitry is configured to calculate the emotion information by applying a regression model to the user input information to compute at least a valence score and an arousal score, and to select the tone parameter by mapping the valence score and the arousal score to a tone selection table stored in the storage device.

### Claim 15 (depends on 14)

The system according to claim 14, wherein the circuitry is configured to generate an approval workflow follow-up notification sequence based on the emotion information, the approval workflow follow-up notification sequence comprising a plurality of notification events each associated with a timing parameter and a notification content template selected based on the tone parameter.

### Claim 16 (depends on 1)

The system according to claim 1, wherein the circuitry is configured to apply a clause completeness check to the post-processed text output data by comparing extracted clause identifiers against a required clause set derived from the type identifier, and to regenerate missing clauses by constructing a targeted prompt sentence for input to the generative neural network model.

### Claim 17 (depends on 1)

The system according to claim 1, wherein the circuitry is configured to monitor a response latency of the generative neural network model, and to switch between a full-parameter model and a distilled model based on whether the response latency exceeds a predefined latency threshold.

### Claim 18

A system comprising:
circuitry configured to:
receive, via a communication interface coupled to a packet-switched network, structured input data from a terminal device;
apply a natural language processing model to the structured input data to extract a purpose identifier, a party identifier, a period parameter, and a type identifier;
construct a prompt sentence encoding the extracted identifiers and parameters, and input the prompt sentence to a generative neural network model to obtain text output data;
post-process the text output data by applying a structural parsing algorithm to assign heading information, clause information, and format information, and generate electronic document data conforming to a predetermined document file format;
store the electronic document data in a storage device and register the extracted identifiers and metadata in an index database;
execute emotion analysis processing on user input information to compute emotion information, and adjust at least one of a tone parameter or an expression strength parameter of the prompt sentence based on the emotion information; and
transmit the electronic document data and the adjusted prompt sentence parameters to the terminal device via the communication interface.

### Claim 19 (depends on 18)

The system according to claim 18, wherein the circuitry is configured to apply an emotion identification model to the user input information to generate an emotional state vector, and to adjust a follow-up condition of an approval workflow based on the emotional state vector.

### Claim 20

A method comprising:
receiving, via a communication interface coupled to a packet-switched network, structured input data from a terminal device, and extracting, by applying a natural language processing model to the structured input data, a purpose identifier, a party identifier, a period parameter, and a type identifier;
constructing a prompt sentence encoding the purpose identifier, the party identifier, the period parameter, and the type identifier, and inputting the prompt sentence and the structured input data to a generative neural network model to obtain text output data;
post-processing the text output data to assign heading information, clause information, and format information to segments of the text output data, and generating electronic document data conforming to a predetermined document file format based on the post-processed text output data;
storing the electronic document data in a storage device, and registering the purpose identifier, the party identifier, the period parameter, the type identifier, and metadata associated with the electronic document data as search information in an index database;
transmitting the electronic document data to the terminal device via the communication interface so that the electronic document data is displayable and editable at the terminal device;
executing emotion analysis processing on user input information received from the terminal device to calculate emotion information, and adjusting at least one of a tone parameter, an expression strength parameter, or a follow-up condition of the prompt sentence based on the emotion information; and
acquiring, from the terminal device, a correction instruction, re-inputting a corrected prompt sentence incorporating the correction instruction and at least a portion of the electronic document data into the generative neural network model to obtain updated text output data, and updating the electronic document data based on the updated text output data.
