GRAPH-GUIDED RETRIEVAL-AUGMENTED GENERATION FOR ENTERPRISE ELECTRONIC PRODUCT DOCUMENTATION

Date

2026-8-31

Author

Özen, Şeymanur

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

38
views

0
downloads

Cite This

Technical documentation for enterprise electronic products is a primary source of engineering and product knowledge. It spans several formats, such as datasheets and brochures, each presenting information differently: brochures in narrative text, datasheets in tables and product-specific numeric specifications. Because the same product is described at different levels of detail across these document types, locating the required information is challenging, and semantic similarity alone is often insufficient for engineering queries. When a query concerns a specific product, conventional retrieval-augmented generation (RAG) systems do not account for product-level relationships or attributes, and may retrieve chunks belonging to different products together, producing misleading answers. This thesis proposes a graph-guided RAG architecture for question answering over electronic product documentation. Product relationships and attributes are modelled in a Neo4j knowledge graph, and queries are filtered through this graph before vector search to restrict retrieval to the matched product scope. A rule-based intent router directs each query to the appropriate processing flow, and a tiered chunk-gating mechanism enforces this scope during retrieval. The proposed method is compared against a hybrid RAG baseline on a benchmark dataset prepared as part of this work, using standard retrieval evaluation metrics (F1, Hit@K, Recall@K) alongside LLM-based judge evaluations. The results show that the graph-guided approach reduces cross-product contamination by 78%, with statistically significant improvements in retrieval accuracy and answer quality. These findings indicate that incorporating product-level relationships and attributes into RAG improves question answering over technical documentation.

Subject Keywords

Retrieval-Augmented Generation, Knowledge Graph, GraphRAG, Domain-specific Question Answering, Cross-Product Contamination

URI

https://hdl.handle.net/11511/120972

Collections

Graduate School of Informatics, Thesis