SINT-Flow: Schema Integration using Large Language Model Workflows
2026-07-27 • Computation and Language
Computation and Language
AI summaryⓘ
The authors designed SINT-Flow, a system that takes multiple database tables and combines them into one clear overall schema, even when the tables mix different types of data. Unlike earlier systems, SINT-Flow can split these mixed tables into separate groups based on entity types automatically. To test their approach, the authors created a benchmark with 10 tasks involving 93 tables, some with mixed entities. Using advanced language models, their system showed good accuracy in identifying entity types, attributes, and how tables map to each other. They also confirmed that certain design choices improved the system's performance.
Schema IntegrationEntity-Type DetectionAttribute DetectionSchema MappingDenormalized TablesLLM-based OperatorsSelf-Consistency StrategySchema MatchingBenchmarkingAutomated Workflows
Authors
Keti Korini, Christian Bizer
Abstract
The goal of schema integration is, given a set of input schemata or tables, to derive a global, unified schema that is able to represent the concepts, attributes, and relationships of all input tables in a coherent fashion. This paper presents SINT-Flow, a schema integration framework composed of five LLM-based operators that can be combined into workflows to perform fully automated, end-to-end schema integration. In contrast to existing approaches, SINT-Flow can process denormalized source tables that contain attributes describing multiple entity types. During the schema integration process, these tables are decomposed into separate entity-specific relations. To evaluate SINT-Flow, we introduce SINT-Bench, a schema integration benchmark comprising 10 schema integration tasks consisting of altogether 93 relational tables, including tables that describe multiple types of entities. We evaluate SINT-Flow using GPT-5.2 as well as the open-weight model Qwen-3.6-27B as alternative backbone models. Using these models, SINT-Flow achieves F1 scores of at least 96% for entity-type detection, 85% for attribute detection, and 83% for schema mapping. Furthermore, we perform an ablation study to prove the utility of the applied self-consistency strategy as well as the inclusion of a review loop into the schema matching operator.