Business Process Classification Using Structural Feature Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing business process analysis methods require manual intervention by analysts with domain knowledge, and the computational complexity increases significantly when comparing all pairs of business processes, leading to inefficiencies in analysis.
Innovation Solution
A business process classification device that automatically classifies processes using feature amounts representing structural features, employing clustering techniques like x-means to reduce computational load and facilitate efficient analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If structural comparison between two business processes is repeated to derive similar processes, then similarity analysis can be performed, but calculation amount increases significantly
Solution Approach 1:
The patent segments the business process comparison task by extracting specific feature amounts (activity names, execution times, worker information) from each business process and organizing them into structured data formats. This segmentation allows for efficient comparison by focusing on key characteristics rather than performing exhaustive structural comparisons of entire process flows.
Solution Approach 2:
The patent introduces an intermediary mechanism by using feature amount extraction and data structure standardization as intermediate steps between raw business process data and similarity comparison. This intermediary layer transforms complex process data into comparable feature vectors, reducing the computational burden of direct structural comparison.
2Productivity
If narrowing down business processes using information such as execution time and performed worker is performed, then operation for checking business processes is reduced, but analyst needs domain knowledge and preliminary investigation
Solution Approach 1:
The patent implements self-service by enabling the system to automatically extract feature amounts from business process data and perform clustering analysis without requiring analyst intervention for feature selection or preliminary investigation. The system autonomously identifies patterns and groups similar business processes, eliminating the need for domain knowledge in the narrowing down process.
Solution Approach 2:
The patent transforms business process data by changing parameters into standardized feature amounts (execution time, activity names, worker information) that can be automatically processed. This parameter transformation enables automated clustering and narrowing down operations, making the process accessible to analysts without specialized domain knowledge.
3Quantity of substance
If enormous number of business processes are generated by process mining, then comprehensive business process analysis is achieved, but manual checking operation increases significantly
Solution Approach 1:
The patent extracts key feature amounts from each business process (activity names, execution times, worker information) and separates them from the complete process data. This extraction enables automated clustering algorithms to process large numbers of business processes efficiently, identifying patterns and grouping similar processes without requiring manual review of each individual process.
Solution Approach 2:
The patent creates simplified copies of business processes in the form of structured feature amount data that captures essential characteristics without containing complete process details. These feature amount copies can be rapidly processed and compared, enabling automated analysis of enormous numbers of business processes while preserving the information needed for meaningful similarity assessment.
Data Source
AI summary
A business process classification device supports improvement in efficiency of analysis of a business process, since the business process classification device includes: a feature amount acquisition unit that acquires from each of a plurality of types of business processes a plurality of feature amounts related to a structure of each of the plurality of types of business processes; and a clustering execution unit that performs clustering based on the plurality of feature amounts for the plurality of types of business processes.