An LLM-Guided Query-Aware Inference System for GNN Models on Large Knowledge Graphs
Waleed Afandi, Hussein Abdallah, Ashraf Aboulnaga, Essam Mansour · 2026
Read An LLM-Guided Query-Aware Inference System for GNN Models on Large Knowledge Graphs by Waleed Afandi, Hussein Abdallah, Ashraf Aboulnaga, Essam Mansour (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-04545.txt.
Description
Afandi and colleagues speed graph neural network inference on huge knowledge graphs. Each query targets different nodes linked to subgraphs of varied density and shape. Old accelerators shrink models yet ignore per query structure and semantics. They load whole monolithic files instead of only relevant parts. KG-WISE splits trained models into fine parts loaded partially per queried subgraph. Language model templates pull relevant subgraphs. On six graphs, the largest with 42 million nodes and 166 million edges, it runs up to 28 times faster with 98 percent less memory, holding accuracy.
Catalog record
| CREATOR | Waleed Afandi, Hussein Abdallah, Ashraf Aboulnaga, Essam Mansour |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 64253 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.04545