Providing Analysis to Improve Workflows
1. Analysis Centered on Workflows
We provide analysis that can help improve workflows by examining the internal processing of AI models. More specifically, we analyze the tendency of deep learning models to exhibit greater similarities in their internal processing when handling inputs related to commonly used workflows. By identifying the differences between a user’s input and widely used workflows, we provide visual and quantitative insights into which workflows serve as useful references and how different workflows compare with one another.
2. Feature-Centered Workflow Analysis
We conduct workflow analysis centered on features, enabling the identification of commonalities across diverse workflows. Rather than limiting analysis to individual workflows, we provide insights into the interoperability of various concepts through a feature-based approach.
3. Improving Empathy Through Workflow Comparison
We also provide comparative analysis across workflows. By comparing workflows generated from different user inputs, we identify their similarities and differences, helping improve empathy among users and supporting the search for more effective workflows.