Process Mining With Generative AI for Unstructured Action Extraction

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional technologies are limited to extracting actions from structured data and cannot effectively process unstructured data, such as emails and chats, to identify and extract activities and tasks.

Innovation Solution

An extraction engine utilizing generative artificial intelligence models processes unstructured data to automatically extract actions and convert them into activities or tasks, integrating with process mining and discovery automation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional process mining and task mining techniques are used to extract actions from structured data, then extraction accuracy is improved, but the ability to process unstructured data remains limited to zero

Engineering Contradiction:

Improveextraction accuracyVSAvoiddata type coverage

Core Design Contradiction:

Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The extraction engine is designed to handle multiple data types universally - it can process both structured data (traditionally handled by process mining) and unstructured data (emails, chats, comments) through the same system using generative AI models, making the system multi-functional and adaptable to various data formats

Inventive Principle:

Principle #6Universality (Multi-functionality)

Solution Approach 2:

The system changes the fundamental parameter of data processing by introducing generative AI models that can interpret and extract actions from unstructured data, transforming the extraction engine from one that only handles structured data to one that handles both structured and unstructured data through parameter transformation

Inventive Principle:

Principle #35Parameter changes

2Loss of information

If machine learning is used to identify intent in unstructured data, then some level of understanding is achieved, but substantial action extraction is not possible

Engineering Contradiction:

Improveintent identificationVSAvoidaction extraction capability

Core Design Contradiction:

Loss of informationVSProductivity

Solution Approach 1:

The patent replaces traditional machine learning intent identification mechanisms with generative AI models that can directly extract actions from unstructured data, substituting the mechanical ML process with a more capable generative approach that produces actionable outputs

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The generative AI model acts as an intermediary between the unstructured data and the extraction engine, translating the unstructured data into structured action representations that can be processed and converted into activities or tasks

Inventive Principle:

Principle #24Intermediary (Mediator)

3Productivity

If conventional extraction techniques are used, then structured data processing is efficient, but unstructured data processing is completely ineffective

Engineering Contradiction:

Improveprocessing efficiencyVSAvoiddata format flexibility

Core Design Contradiction:

ProductivityVSAdaptability or versatility

Solution Approach 1:

The system performs preliminary processing by using generative AI models to convert unstructured data into structured action representations before the final extraction and conversion to activities or tasks, preparing the data in advance for efficient processing

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS20250321989A1Process mining and discovery automation for extracting activities and tasks out of unstructured data

Publication Date: 2025.10.16 UIPATH INC

AI summary

A method is provided. The method is executed by an extraction engine implemented as a computer program within a computing environment. The extraction engine executes action and task mining on unstructured data. The method includes receiving a communication including unstructured data defining an action and automatically processing the communication by utilizing at least one generative artificial intelligence (AI) model to extract details of the action being performed in the unstructured data. The method includes automatically converting the action into an activity or a task of a process associated with the communication.