This document describes HoliMatch, an approach for integrating query results from multiple sources by performing holistic schema matching. It introduces challenges in matching schemas when explicit schemas are not available. The approach matches fields across sources by enriching field content and constraints observed from multiple sources. It formulates the matching problem as domain schema discovery and presents an iterative algorithm that learns the schema from an initial matching and then refines the matching based on the learned schema. Experimental results on four domains with 10 sources each demonstrate that the holistic approach outperforms other matching methods.

























