Function aware retrieval improves answers from complex technical documents
From Chunks to Functional Evidence: Function-Aware Retrieval for EDA Documentation QA
Information RetrievalArtificial Intelligence
Summary
Finding the right information in complicated technical manuals can be hard because the information is scattered and connected in different ways. The paper shows a better way to organize and search for useful information by grouping related parts based on their function, rather than just looking at isolated pieces. The authors tested their method on electronic design manuals and public technical question benchmarks, showing much better answer quality compared to previous methods. This approach helps computers find more relevant evidence to answer detailed technical questions.
What this means in practice
- •For electronic design engineers: Retrieve better, functionally organized explanations from EDA manuals to support complex design troubleshooting and documentation queries.
- •For technical support teams: Provide more accurate and relevant answers to technical product questions by matching user queries with function-based knowledge segments.
Authors
Xiaotian Qiu, Kairui Liu, Shi Chenyi, Jinyuan Deng, Qi Sun, Cheng Zhuo
Abstract
Retrieval-Augmented Generation (RAG) is widely used to ground answers in documents. For complex technical documentation, however, the primary bottleneck is often not model reasoning but a mismatch between a query and the way knowledge is organized for retrieval. This mismatch is pronounced in Electronic Design Automation (EDA) documentation, where the information needed for an answer is scattered across heterogeneous yet tightly coupled artifacts. We therefore redesign the basic retrieval unit of RAG. Instead of operating on isolated chunks or binary relations, we collect typed artifacts into EDA functional units. Each unit is recorded as a hyperedge with links to its source chunks. We then train an encoder to align queries with functional units and combine unit retrieval with direct chunk retrieval. After mapping the selected units back to their sources, a unified reranker chooses the evidence given to the generator. On the newly constructed EDADocEval-QA dataset, our method improves ROUGE-L by 37.1% over Chunk RAG and 55.6% over the strongest graph baseline. On the public ORD-MMBench benchmark, it improves ROUGE-L by 30.0% over the strongest baseline. These results support function-aware evidence organization in the evaluated EDA documentation settings.