Scaling Expert Feedback with Reflective Edit Propagation in Compositional Knowledge Bases
Jiajing Guo (Bosch Research North America), Xueming Li (Robert Bosch GmbH), Jorge H Piazentin Ono (Bosch Research North America), Wenbin He (Bosch Research North America), Liu Ren (Bosch Research North America)
Architectural Patterns & Composition
Summary
A reflective agent (RAID) that infers the semantic intent behind a single expert edit to a knowledge base and propagates corrections across the entire KB automatically.
Description
Curation of domain-specific knowledge bases (KBs) is critical for organizations and enterprises, but suffers from a scalability bottleneck: while LLMs can synthesize initial content, only human experts can validate technical accuracy, and manual entry-by-entry correction is labor-intensive. We use an identifier dictionary as a use case and present Reflective Agent for Identifier Dictionary (RAID), a novel system that transforms individual expert edits into systematic knowledge updates. Unlike traditional "correct-and-save" paradigms, RAID utilizes a reflective agent to infer the underlying semantic intent behind a single expert edit and propagates that correction across the entire KB through a three-step architecture: Intent Inference, Reflection-based Planning, and User Controlled Execution. We evaluated the reflection and propagation performance on a public dataset and conducted a user study with subject matter experts with proprietary data. The evaluation shows RAID's technical feasibility in capturing expert intent and its potential to scale specialized expertise across industrial knowledge bases.
