What is it about?

When you try to combine data from different sources, like spreadsheets or websites, you often find that the same information is written in different ways. For example, one table might list full names, and another might list only initials or email addresses. Because of these formatting differences, it's hard to automatically match and join these tables. Our research introduces a method called GXJoin, which helps to find simple and general rules for turning one format into another. These rules can explain how two columns relate, even if their text looks different at first. Unlike earlier tools that often create complicated or overly specific rules, GXJoin focuses on finding patterns that work in many situations and are easier for people to understand.

Featured Image

Why is it important?

This work addresses a major challenge in data integration: how to reliably join tables that describe the same things but use different text formats. Current tools often rely on narrow or overly complex rules that don’t work well when new or messy data is introduced. What makes our approach, GXJoin, unique is its focus on general and explainable transformations that are not only more adaptable but also easier for people to understand and trust.

Perspectives

Our goal was to make the process of joining tables easier, not just for machines but for people trying to understand what's going on behind the scenes. I personally found it exciting to explore ways to make transformation rules both more general and more explainable.

Soroush Omidvartehrani
University of Alberta

Read the Original

This page is a summary of: GXJoin: Generalized Cell Transformations for Explainable Joinability, January 2024, Springer Science + Business Media,
DOI: 10.1007/978-3-031-70626-4_9.
You can read the full text:

Read

Contributors

The following have contributed to this page