SymRAG: Efficient Neuro-Symbolic Retrieval Through Adaptive Query Routing

Safayat Bin Hakim, Muhammad Adil, Alvaro Velasquez, Houbing Herbert Song
Proceedings of The 19th International Conference on Neurosymbolic Learning and Reasoning, PMLR 284:540-564, 2025.

Abstract

Current Retrieval-Augmented Generation systems use uniform processing, causing inefficiency as simple queries consume resources similar to complex multi-hop tasks. We present SymRAG, a framework that introduces adaptive query routing via real-time complexity and load assessment to select symbolic, neural, or hybrid pathways. SymRAG’s neuro-symbolic approach adjusts computational pathways based on both query characteristics and system load, enabling efficient resource allocation across diverse query types. By combining linguistic and structural query properties with system load metrics, SymRAG allocates resources proportional to reasoning requirements. Evaluated on 2,000 queries across HotpotQA (multi-hop reasoning) and DROP (discrete reasoning) using Llama-3.2-3B and Mistral-7B models, SymRAG achieves competitive accuracy (97.6–100.0% exact match) with efficient resource utilization (3.6–6.2% CPU utilization, 0.985–3.165s processing). Disabling adaptive routing increases processing time by 169–1151%, showing its significance for complex models. These results suggest adaptive computation strategies are more sustainable and scalable for hybrid AI systems that use dynamic routing and neuro-symbolic frameworks.

Cite this Paper

BibTeX
@InProceedings{pmlr-v284-hakim25a, title = {SymRAG: Efficient Neuro-Symbolic Retrieval Through Adaptive Query Routing}, author = {Hakim, Safayat Bin and Adil, Muhammad and Velasquez, Alvaro and Song, Houbing Herbert}, booktitle = {Proceedings of The 19th International Conference on Neurosymbolic Learning and Reasoning}, pages = {540--564}, year = {2025}, editor = {H. Gilpin, Leilani and Giunchiglia, Eleonora and Hitzler, Pascal and van Krieken, Emile}, volume = {284}, series = {Proceedings of Machine Learning Research}, month = {08--10 Sep}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v284/main/assets/hakim25a/hakim25a.pdf}, url = {https://proceedings.mlr.press/v284/hakim25a.html}, abstract = {Current Retrieval-Augmented Generation systems use uniform processing, causing inefficiency as simple queries consume resources similar to complex multi-hop tasks. We present SymRAG, a framework that introduces adaptive query routing via real-time complexity and load assessment to select symbolic, neural, or hybrid pathways. SymRAG’s neuro-symbolic approach adjusts computational pathways based on both query characteristics and system load, enabling efficient resource allocation across diverse query types. By combining linguistic and structural query properties with system load metrics, SymRAG allocates resources proportional to reasoning requirements. Evaluated on 2,000 queries across HotpotQA (multi-hop reasoning) and DROP (discrete reasoning) using Llama-3.2-3B and Mistral-7B models, SymRAG achieves competitive accuracy (97.6–100.0% exact match) with efficient resource utilization (3.6–6.2% CPU utilization, 0.985–3.165s processing). Disabling adaptive routing increases processing time by 169–1151%, showing its significance for complex models. These results suggest adaptive computation strategies are more sustainable and scalable for hybrid AI systems that use dynamic routing and neuro-symbolic frameworks.} }
Endnote
%0 Conference Paper %T SymRAG: Efficient Neuro-Symbolic Retrieval Through Adaptive Query Routing %A Safayat Bin Hakim %A Muhammad Adil %A Alvaro Velasquez %A Houbing Herbert Song %B Proceedings of The 19th International Conference on Neurosymbolic Learning and Reasoning %C Proceedings of Machine Learning Research %D 2025 %E Leilani H. Gilpin %E Eleonora Giunchiglia %E Pascal Hitzler %E Emile van Krieken %F pmlr-v284-hakim25a %I PMLR %P 540--564 %U https://proceedings.mlr.press/v284/hakim25a.html %V 284 %X Current Retrieval-Augmented Generation systems use uniform processing, causing inefficiency as simple queries consume resources similar to complex multi-hop tasks. We present SymRAG, a framework that introduces adaptive query routing via real-time complexity and load assessment to select symbolic, neural, or hybrid pathways. SymRAG’s neuro-symbolic approach adjusts computational pathways based on both query characteristics and system load, enabling efficient resource allocation across diverse query types. By combining linguistic and structural query properties with system load metrics, SymRAG allocates resources proportional to reasoning requirements. Evaluated on 2,000 queries across HotpotQA (multi-hop reasoning) and DROP (discrete reasoning) using Llama-3.2-3B and Mistral-7B models, SymRAG achieves competitive accuracy (97.6–100.0% exact match) with efficient resource utilization (3.6–6.2% CPU utilization, 0.985–3.165s processing). Disabling adaptive routing increases processing time by 169–1151%, showing its significance for complex models. These results suggest adaptive computation strategies are more sustainable and scalable for hybrid AI systems that use dynamic routing and neuro-symbolic frameworks.
APA
Hakim, S.B., Adil, M., Velasquez, A. & Song, H.H.. (2025). SymRAG: Efficient Neuro-Symbolic Retrieval Through Adaptive Query Routing. Proceedings of The 19th International Conference on Neurosymbolic Learning and Reasoning, in Proceedings of Machine Learning Research 284:540-564 Available from https://proceedings.mlr.press/v284/hakim25a.html.

Related Material