IP Library › Granted Patent US 12705272
Granted Patent B1
US 12705272 · App. 18/232,447 · Granted Aug 11, 2026

Entire document summarization

Inventors: Ramachandra Kota (San Francisco, CA); Souleiman Hasan (Dublin, IE); Yan He (San Francisco, CA); Seán Mahon (Dublin, IE); Alexey Zakhvatov (San Francisco, CA); Yan Pui Lam (San Francisco, CA); Navin Albert (San Francisco, CA)
Assignee: Docusign, Inc.
G06F16/345G06V10/82G06V30/274G06V30/413G06V30/414
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12705272
App. No.
18/232,447
Granted
Aug 11, 2026
Kind
B1
Abstract

A method, a system, and a computer program product for summarizing an electronic document. A structural representation of the document is generated and/or received. The structural representation specifies one or more elements of the document and one or more relationships between one or more elements of the document. A machine learning model is identified for the document. The model is applied, based on the structural representation, to one or more portions of the document to generate a hierarchical representation for the portions. A request to execute at least one processing task related to the document is received. At least one generated hierarchical representation of at least one portion of the document is sent to a generative artificial intelligence model to execute the processing task by generating a response to the request. The response is and presented on a graphical user interface of a user computing device.

Claims (64)

1 . A computer implemented method, comprising:

generating, using at least one processor, a structural representation of an electronic document, the structural representation specifying one or more elements of the electronic document and one or more relationships between the one or more elements of the electronic document;

selecting, using the at least one processor, based on the structural representation, a machine learning model from a plurality of machine learning models for the electronic document, and applying, based on the structural representation of the electronic document, the selected machine learning model to one or more portions of the electronic document to generate a hierarchical representation for the one or more portions of the electronic document;

receiving, using the at least one processor, a request to execute at least one processing task related to the electronic document;

sending, using the at least one processor, at least one generated hierarchical representation of at least one portion in the one or more portions of the electronic document to a generative artificial intelligence (AI) model to execute the at least one processing task by generating a response to the request; and

receiving, using the at least one processor, the response and presenting the response on a graphical user interface of at least one user computing device.

2 . The method of claim 1 , further comprising

receiving, using the at least one processor, at least one feedback from the at least one user computing device.

3 . The method of claim 2 , further comprising

performing, using the at least one processor, based on the received at least one feedback, at least one of the following:

updating the generated structural representation of the electronic document to generate an updated structural representation of the electronic document;

identifying at least another machine learning model for the electronic document and applying the another machine learning model to one or more portions of the electronic document to generate at least another hierarchical representation for the one or more portions of the electronic document;

updating the machine learning model selected for the electronic document to generate an updated machine learning model applying the updated machine learning model to one or more portions of the electronic document to generate an updated hierarchical representation for the one or more portions of the electronic document;

generating an updated hierarchical representation for at least one portion in the one or more portions of the electronic document; and

any combination thereof; and

receiving, using the at least one processor, in response to the performing, an updated response from the generative AI model and presenting the updated response on the graphical user interface of at least one user computing device.

4 . The method of claim 1 , wherein the generating the structural representation of the electronic document includes hierarchically arranging of the one or more elements based on the one or more relationships between one or more elements of the electronic document.

5 . The method of claim 4 , wherein the generating the structural representation of the electronic document includes generating one or more groups of the one or more elements within hierarchically arranged one or more elements.

6 . The method of claim 5 , wherein at least one of hierarchical arrangement of the one or more elements and the one or more groups of the one or more elements are determined based on at least one of the following: a position of each element in the one or more elements in the electronic document, a type of each element in the one or more elements in the electronic document, one or more functions of each element in the one or more elements in the electronic document, and any combination thereof.

7 . The method of claim 1 , wherein the one or more elements include at least one of the following: a text, an audio, a video, an image, a table, and any combination thereof.

8 . The method of claim 1 , wherein the at least one processing task includes at least one of the following: a summarization of the electronic document, a summarization of at least one element in the one or more elements, a contextual extraction from the electronic document, an explanation of the electronic document, an explanation of at least one element in the one or more elements, an explanation of at least one portion in the one or more portions, a semantic search of the electronic document, a generation of an outline of the electronic document, and any combinations thereof.

9 . The method of claim 1 , wherein the selecting includes

generating the hierarchical representation for the one or more portions of the electronic document based on at least one of the following: an importance parameter associated with at least one element in the one or more elements, the at least one processing task, a content of at least one element in the one or more elements, and any combinations thereof.

10 . The method of claim 9 , wherein the importance parameter is defined by the at least one processing task.

11 . The method of claim 10 , wherein the importance parameter is defined by at least one of the following: a type of the electronic document, a position of each element in the one or more elements in the electronic document, a type of each element in the one or more elements in the electronic document, one or more functions of each element in the one or more elements in the electronic document, and any combination thereof.

12 . A system, comprising:

at least one processor; and

at least one non-transitory storage media storing instructions, that when executed by the at least one processor, cause the at least one processor to perform operations including

selecting, based on a structural representation of an electronic document, a machine learning model from a plurality of machine learning models for the electronic document, and applying, based on the structural representation of the electronic document, the selected machine learning model to one or more portions of the electronic document to generate a hierarchical representation for the one or more portions of the electronic document, where the structural representation of the electronic document specifies one or more elements of the electronic document and one or more relationships between the one or more elements of the electronic document;

sending, in response to receiving a request to execute at least one processing task related to the electronic document, at least one generated hierarchical representation of at least one portion in the one or more portions of the electronic document to a generative artificial intelligence (AI) model to execute the at least one processing task by generating a response to the request; and

presenting the response on a graphical user interface of at least one user computing device.

13 . The system of claim 12 , wherein the operations further comprise

receiving at least one feedback from the at least one user computing device;

performing based on the received at least one feedback, at least one of the following:

updating the structural representation of the electronic document to generate an update structural representation of the electronic document;

identifying at least another machine learning model for the electronic document and applying the another machine learning model to one or more portions of the electronic document to generate at least another hierarchical representation for the one or more portions of the electronic document;

updating the machine learning model selected for the electronic document to generate an updated machine learning model applying the updated machine learning model to one or more portions of the electronic document to generate an updated hierarchical representation for the one or more portions of the electronic document;

generating an updated hierarchical representation for at least one portion in the one or more portions of the electronic document; and

any combination thereof; and

receiving in response to the performing, an updated response from the generative AI model and presenting the updated response on the graphical user interface of at least one user computing device.

14 . The system of claim 12 , wherein the generating the structural representation of the electronic document includes

hierarchically arranging of the one or more elements based on the one or more relationships between one or more elements of the electronic document; and

generating one or more groups of the one or more elements within hierarchically arranged one or more elements;

wherein at least one of hierarchical arrangement of the one or more elements and the one or more groups of the one or more elements are determined based on at least one of the following: a position of each element in the one or more elements in the electronic document, a type of each element in the one or more elements in the electronic document, one or more functions of each element in the one or more elements in the electronic document, and any combination thereof.

15 . The system of claim 12 , wherein the at least one processing task includes at least one of the following: a summarization of the electronic document, a summarization of at least one element in the one or more elements, a contextual extraction from the electronic document, an explanation of the electronic document, an explanation of at least one element in the one or more elements, an explanation of at least one portion in the one or more portions, a semantic search of the electronic document, a generation of an outline of the electronic document, and any combinations thereof.

16 . The system of claim 12 , wherein the selecting includes

generating the hierarchical representation for the one or more portions of the electronic document based on at least one of the following: an importance parameter associated with at least one element in the one or more elements, the at least one processing task, a content of at least one element in the one or more elements, and any combinations thereof.

17 . The system of claim 16 , wherein the importance parameter is defined by at least one of the following: the at least one processing task, a type of the electronic document, a position of each element in the one or more elements in the electronic document, a type of each element in the one or more elements in the electronic document, one or more functions of each element in the one or more elements in the electronic document, and any combination thereof.

18 . A computer program product comprising a non-transitory machine-readable medium storing instructions that, when executed by at least one programmable processor, cause the at least one programmable processor to perform operations comprising:

receiving a request to execute at least one processing task related to an electronic document;

sending a hierarchical representation of at least one portion in one or more portions of the electronic document to a generative artificial intelligence (AI) model to execute the at least one processing task by generating a response to the request, the hierarchical representation being generated by applying a machine learning model selected from a plurality of machine learning models to one or more portions of the electronic document, the selected machine learning model being selected based on a structural representation of the electronic document, the structural representation specifying one or more elements of the electronic document and one or more relationships between the one or more elements of the electronic document;

presenting the response on a graphical user interface of at least one user computing device;

receiving at least one feedback from the at least one user computing device; and

presenting, in response to the receiving the at least one feedback, an updated response generated by the generative AI model on the graphical user interface of the at least one user computing device.

19 . The computer program product of claim 18 , wherein the operations further comprise

performing, based on the received at least one feedback, and sending to the generating AI model at least one of the following:

updating the structural representation of the electronic document to generate an update structural representation of the electronic document;

identifying at least another machine learning model for the electronic document and applying the another machine learning model to one or more portions of the electronic document to generate at least another hierarchical representation for the one or more portions of the electronic document;

updating the machine learning model selected for the electronic document to generate an updated machine learning model applying the updated machine learning model to one or more portions of the electronic document to generate an updated hierarchical representation for the one or more portions of the electronic document;

generating an updated hierarchical representation for at least one portion in the one or more portions of the electronic document; and

any combination thereof.

20 . The computer program product of claim 19 , wherein the sending includes

generating the hierarchical representation for the one or more portions of the electronic document based on at least one of the following: an importance parameter associated with at least one element in the one or more elements, the at least one processing task, a content of at least one element in the one or more elements, and any combinations thereof;

wherein the importance parameter is defined by at least one of the following: the at least one processing task, a type of the electronic document, a position of each element in the one or more elements in the electronic document, a type of each element in the one or more elements in the electronic document, one or more functions of each element in the one or more elements in the electronic document, and any combination thereof.