Automated Process Discovery via Keystroke Signature Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for automated process discovery in robotic process automation are limited by their reliance on interviews, which provide incorrect and unusable information, and log mining, which is supported only for a handful of applications and lacks detailed user-level data.
Innovation Solution
The development of techniques to collect and analyze low-level data such as click and keystroke data from multiple users, generating a 'signature' or fingerprint representing a process, and using this signature to identify instances of the process in large volumes of data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If interviews are used for process discovery, then user insights can be obtained, but the information provided is incorrect and unusable
Solution Approach 1:
The patent replaces the manual interview process with an automated monitoring application that captures keystroke data and computer activity. This substitution eliminates the reliability issues of human-provided information by using objective, machine-collected data to discover processes.
Solution Approach 2:
The system enables processes to be discovered automatically through monitoring user interactions with the computer. The monitoring application self-collects data without requiring user intervention or interviews, allowing the system to discover processes on its own through objective behavioral data.
2Adaptability or versatility
If log mining is used for process discovery, then process data can be extracted, but it is supported only for a handful of applications and lacks detailed user-level data
Solution Approach 1:
The monitoring application is designed to work across multiple applications and operating systems without requiring application-specific log mining. It captures low-level keystroke and interaction data universally, enabling process discovery across diverse software environments while maintaining detailed user-level information.
Solution Approach 2:
The patent breaks down process discovery into fundamental keystroke-level events and basic computer interactions. By segmenting high-level application-specific logs into atomic interaction units, the system achieves both broad application coverage and detailed user-level data capture that traditional log mining cannot provide.
3Measurement precision
If traditional process discovery methods are used, then processes can be identified, but the computational burden is high and accuracy is limited
Solution Approach 1:
The patent changes the fundamental parameters of process discovery by working at the keystroke level rather than analyzing complete process logs or interview data. This parameter change enables more accurate process identification through fine-grained event sequencing while reducing computational burden by focusing on essential interaction patterns.
Solution Approach 2:
The monitoring application continuously captures and stores low-level interaction data in the background before process discovery is needed. This preliminary data collection eliminates the need for computationally intensive real-time analysis, allowing accurate process identification when required while minimizing computational resource usage during the discovery phase.
Data Source
AI summary
A method, comprising: receiving, by a computing device, a user indication to start a teaching mode in which the user can teach an instance of a process; configuring the computing device to start the teaching mode in response to receiving the indication; capturing, when in the teaching mode, first information corresponding to a first stream of events captured by the computing device during performance of a first plurality of actions by the user when interacting with the computing device to perform the process; generating, using the first information, a first visualization of at least some of the first plurality of actions; and displaying the first visualization on a graphical user interface of the computing device.