From Graph Retrieval to Schema Realization: Counterfactual Validation for Text-to-SPARQL over Heterogeneous Knowledge Graphs
Researchers have developed SchemaForge, a new framework designed to improve text-to-SPARQL query generation over collections of heterogeneous knowledge graphs. This system addresses the challenge of dealing with multiple graphs that may have different schemas, partial alignments, and incomplete metadata. SchemaForge uses a question-conditioned schema-slice alignment mechanism to identify plausible graphs and then selects a local schema slice to constrain query generation and verification, leading to improved execution accuracy. AI
IMPACT Improves accuracy for querying diverse knowledge graphs, potentially aiding AI systems that rely on structured data.