Machine Learning Document Creation with Lexical-Semantic Matching

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional software applications for creating electronic documents, such as invoices, are time-consuming, repetitive, and inefficient, particularly for businesses that frequently create similar documents, due to the dynamic nature of the data and the unstructured format of available information.

Innovation Solution

Utilizing machine learning techniques to extract relevant data from source documents and dynamically match it to entities in a data store, employing a combination of lexical and semantic searches to populate document templates automatically, with user feedback for iterative improvement.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If conventional software applications are used to create electronic documents manually, then users can create documents with complete control over each field, but the process becomes time-consuming and inefficient

Engineering Contradiction:
Improvedocument creation speedVSAvoidtime spent on manual data entry
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The system performs preliminary actions by pre-populating document fields with relevant data extracted from source documents before the user finalizes the document. The automated data extraction and field population occurs in the background, preparing the document for quick review and approval, thus eliminating manual data entry while maintaining user control.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system enables self-service by automatically extracting data from source documents, matching it to appropriate fields, and populating the electronic document without requiring manual user input. The intelligent field matching and data population occur autonomously, allowing users to simply review and approve the generated document.

Inventive Principle:
Principle #25Self-service

2Productivity

If automated data extraction is implemented, then document creation time is reduced, but the accuracy of data matching may deteriorate

Engineering Contradiction:
Improveautomation levelVSAvoiddata matching accuracy
Core Design Contradiction:
ProductivityVSManufacturing precision

Solution Approach 1:

The system implements feedback mechanisms where extraction and matching results are validated against expected patterns and user corrections are fed back into the system to refine future extractions. The confidence scoring system provides feedback on match quality, allowing the system to adjust its extraction and matching strategies to improve accuracy while maintaining automation.

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The system employs dynamic field matching that adapts to different document types, structures, and contexts. The matching algorithm dynamically adjusts its behavior based on the source document format and the target document template, enabling accurate data extraction and population across varied scenarios while maintaining high automation levels.

Inventive Principle:
Principle #15Dynamics

3Adaptability or versatility

If a simple extraction system is used, then implementation is easier, but it cannot handle the dynamic nature of unstructured data formats

Engineering Contradiction:
Improvehandling of unstructured dataVSAvoidsystem complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The system achieves universality by implementing a multi-functional extraction engine that can handle various unstructured data formats (emails, PDFs, spreadsheets, images) using the same core technology stack. The intelligent field matching mechanism works across different document types and structures, providing adaptability without requiring separate systems for each format.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The system introduces an intermediary layer of intelligent field matching that mediates between the diverse unstructured source documents and the standardized target document templates. This intermediary matching layer translates various data formats and structures into a common framework, enabling the system to handle dynamic unstructured data without increasing overall system complexity.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP4610876A1Automated electronic document creation through machine learning
Publication Date: 2025.09.03 INTUIT INC
  • EP4610876A1 patent drawingFigure 1
  • EP4610876A1 patent drawingFigure 2
  • EP4610876A1 patent drawingFigure 3

AI summary

Aspects of the present disclosure relate to automated electronic document creation. Embodiments include providing, as inputs to a machine learning model, contents of a source electronic document and a prompt. Embodiments include receiving a particular data item from the machine learning model in response to the inputs. Embodiments include performing a lexical search for the particular data item among a plurality of stored data items. Embodiments include determining a confidence score based on the lexical search and performing a semantic search for the particular data item among the plurality of stored data items based on the confidence score not exceeding a threshold. Embodiments include identifying a stored data item of the plurality of data items that matches the particular data item based on the semantic search. Embodiments include automatically generating an electronic document by populating an electronic document template using the stored data item.