Turning Semantics into Topology: LLM-Driven Attribute Augmentation for Collaborative Filtering
Junjie Meng, Ranxu Zhang, Wei Wu, Rui Zhang, Chuan Qin, Zhang Qi, Qi Liu, Hui Xiong, Chao Wang · 2026
Read Turning Semantics into Topology: LLM-Driven Attribute Augmentation for Collaborative Filtering by Junjie Meng, Ranxu Zhang, Wei Wu, Rui Zhang, Chuan Qin, Zhang Qi, Qi Liu, Hui Xiong, Chao Wang (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-2602-21099.txt.
Description
Language models know a lot about items and users, but turning that knowledge into tip vectors is awkward. Junjie Meng and eight co-authors bridge the gap in this 2026 arXiv paper using Topology-Augmented Graph Collaborative Filtering. The setup uses language models to infer contact intents and causal links from user-item pairs, coding them as trait nodes in a rich user-trait-item graph.
A tunable link-weighted graph conv then learns how much each link type matters, using link-own guessing nets. Tests across trials and joint-sifting bases show steady gains, including cold-start cases where new items lack past use. The work turns meaning knowledge into graph shape rather than bolting text traits onto vectors.
Catalog record
| CREATOR | Junjie Meng, Ranxu Zhang, Wei Wu, Rui Zhang, Chuan Qin, Zhang Qi, Qi Liu, Hui Xiong, Chao Wang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 49 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-21099