Abstract
A computer-implemented system and method to generate trusted analytics result from a natural-language business query. Business-intent elements represented by the query are mapped to semantic objects in a semantic data graph representing business entities, approved relationships, measures, dimensions, filters, and aggregation-grain metadata. Before generation of an executable query, one or more candidate semantic query plans are generated over the semantic data graph. A validation engine validates relationship-path correctness, measure compatibility, aggregation-grain safety, ambiguity, policy constraints, and fanout or double-counting risk. Validated plans may be ranked using confidence scores, and clarification may be requested when materially different interpretations remain. A selected validated semantic query plan is translated into an executable query and executed against one or more data sources. The system returns analytics result together with an explanation identifying the semantic interpretation, measure, dimensions, filters, or relationship path used to generate the analytics result.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Jaiswal, Srishti and Hedau, Anshul, "TRUSTED GEN AI ANALYTICS USING SEMANTIC GRAPH GROUNDING AND GRAIN-SAFE QUERY VALIDATION", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/12400