Process Identification Using Event Trace Visualization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing process mining techniques struggle to efficiently identify and visualize complex processes from large datasets, such as insurance claims data, due to the volume and complexity of event data.
Innovation Solution
A computer-implemented method and system that generates aggregated event traces, encodes unique events into graphical representations, compresses data, clusters similar events, labels clusters, and displays process maps to improve process identification and visualization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If traditional process mining captures and stores event content in textual form from large datasets, then complete process information is preserved, but data complexity and analysis difficulty increase significantly
Solution Approach 1:
The patent segments the complex insurance claims data into distinct event traces, where each event trace represents a sequence of related events. This segmentation divides the large, complex dataset into manageable units that can be analyzed independently, reducing overall data complexity while preserving complete process information within each segment.
Solution Approach 2:
The patent introduces event traces as an intermediary representation between raw event data and process analysis. These event traces serve as a mediating structure that organizes raw events into meaningful sequences, making the data more amenable to analysis without losing process information.
2Measurement precision
If all event data is analyzed in detail, then accurate process identification is achieved, but processing time and computational resources increase
Solution Approach 1:
The patent extracts key characteristics and patterns from event traces to create condensed representations. By taking out only the essential features needed for process identification rather than analyzing every detail of all events, the system achieves accurate process identification while significantly reducing processing time.
Solution Approach 2:
The patent applies partial analysis by focusing computational resources on the most informative aspects of event traces. Rather than uniformly analyzing all event data in detail, the system selectively analyzes portions of the data that provide the most value for process identification, achieving good accuracy with reduced processing time.
3Loss of information
If event data is visualized in traditional textual format, then complete information is displayed, but user comprehension and pattern recognition become difficult
Solution Approach 1:
The patent employs visual encoding where different event types or states are represented by different visual characteristics (such as color coding, icons, or graphical symbols). This allows complete event information to be displayed while simultaneously improving user comprehension through intuitive visual patterns that are easier to interpret than traditional textual formats.
Solution Approach 2:
The patent transforms one-dimensional textual event data into multi-dimensional visual representations. By adding visual dimensions (such as spatial arrangement, graphical symbols, or visual hierarchies), the system preserves all event information while making it much easier for users to comprehend and recognize patterns.
Data Source
AI summary
A method for improving identifying of a process using compact visualizations with memorable characters includes generating aggregated event traces by preprocessing formatted data corresponding to insurance claims data, encoding each of the aggregated event traces, wherein each unique event in the aggregated event traces is represented using a respective graphical representation of an atomic word, generating a subset of aggregated event traces by compressing the aggregated event traces, clustering each of the subset of aggregated event traces into a respective cluster, labeling each of the clusters by analyzing the subset of corresponding aggregated event traces, and displaying a map of the one or more of the aggregated event traces in a linear order, wherein the aggregated event traces include one or more complex data types.