Interactive fixes for graph data

What is it about?

Graphs are a powerful way to organize and analyze data in various fields, such as healthcare, finance, and social networks. These graphs help us visualize relationships between different entities, like patients and their treatments in a hospital. However, errors and inconsistencies in graph data can lead to incorrect conclusions or decisions. Fixing these errors is often challenging because it requires specialized knowledge and manual effort. In this study, we propose a new interactive approach that focuses on involving real users (such as domain experts) in identifying and fixing errors in graph data. Unlike automated solutions, our method allows multiple users to collaborate and make corrections based on their expertise. To manage this process efficiently, we developed a framework that detects errors, assigns repair tasks to users, and ensures that their corrections don’t introduce new problems. Our experiments show that this user-centric approach improves the quality of graph repairs by 30% compared to existing methods, even when users are not experts. This makes our method more practical and effective for cleaning complex graph data in real-world applications.

Featured Image

Photo by JJ Ying on Unsplash

Photo by JJ Ying on Unsplash

Why is it important?

Existing solutions for repairing errors in graph data are mostly automated and assume that all necessary knowledge is pre-defined, which is rarely the case in real-world scenarios. What makes our work unique is its user-centric approach. We involve real users with different levels of expertise to collaboratively identify and fix inconsistencies in property graphs. Our framework provides an interactive, step-by-step process that assigns repair tasks to users and minimizes conflicts while ensuring that data corrections are accurate and consistent. This human-in-the-loop method bridges the gap between domain knowledge and technical graph repair solutions, something that has not been extensively explored before. This research comes at a time when collaborative, user-driven solutions are highly relevant for improving data quality in large-scale graph databases. By empowering multiple users to participate in repairing graph data, our approach can significantly improve data quality and reduce the burden on data engineers. This leads to more accurate decision-making in critical areas like healthcare and finance. Additionally, the interactive nature of the framework makes it accessible to a broader range of users, potentially increasing adoption and engagement in both academic and industry settings.