analiza aiqbench
This paper proposes ZeroPolar, a training-free method for zero-shot classification of linked web pages that uses LLM labels and text-prototype predictions to decide when links should push predictions toward or away from neighbours. Its reported gains are on eleven graphs, with ZeroPolar-T at 73.99% mean accuracy and a 6.01-point lead over LLM-BP. Readers should weigh these as benchmark results on a specific setup that relies on GPT-4o labels, and the method's value in production depends on how well it holds up beyond those graphs.
When Links Mislead: Label-Free Structural Polarity for Zero-Shot Classification of Web Pages
arXiv:2610.41464v1 Announce Type: new Abstract: Web platforms such as digital libraries, online stores and university sites must classify new pages before anyone labels them. Large language models (LLMs) can label each page from its text, but these labels are noisy, and links could correct them only where linked pages share classes. Existing zero-shot methods ignore the links, or pull a page toward its neighbours as label propagation and graph neural networks (GNNs) do, which hurts where linked pages differ in class. Others weight propagation by an LLM-estimated homophily that costs extra calls and drifts across LLMs. We find that the sign with which links should enter a prediction, which we call structural polarity, is readable without true labels. It is the excess agreement of LLM labels and text-prototype predictions along links. We thus propose ZeroPolar, which gates one signed propagation step by this statistic. Training-free ZeroPolar mixes prototype scores with any number of LLM labels, and ZeroPolar-T denoises the labels with a text classifier once half of the pages carry one. With one GPT-4o label per page, ZeroPolar-T reaches the highest mean accuracy on eleven graphs (73.99%), 6.01 points above LLM-BP and significantly above all eighteen baselines we compare. With LLM labels on only 5% of the pages, training-free ZeroPolar leads a GNN trained on these labels by 10.80 points.
Abstract:Web platforms such as digital libraries, online stores and university sites must classify new pages before anyone labels them. Large language models (LLMs) can label each page from its text, but these labels are noisy, and links could correct them only where linked pages share classes. Existing zero-shot methods ignore the links, or pull a page toward its neighbours as label propagation and graph neural networks (GNNs) do, which hurts where linked pages differ in class. Others weight propagation by an LLM-estimated homophily that costs extra calls and drifts across LLMs. We find that the sign with which links should enter a prediction, which we call structural polarity, is readable without true labels. It is the excess agreement of LLM labels and text-prototype predictions along links. We thus propose ZeroPolar, which gates one signed propagation step by this statistic. Training-free ZeroPolar mixes prototype scores with any number of LLM labels, and ZeroPolar-T denoises the labels with a text classifier once half of the pages carry one. With one GPT-4o label per page, ZeroPolar-T reaches the highest mean accuracy on eleven graphs (73.99%), 6.01 points above LLM-BP and significantly above all eighteen baselines we compare. With LLM labels on only 5% of the pages, training-free ZeroPolar leads a GNN trained on these labels by 10.80 points.
Machine Learning (cs.LG)
arXiv:2610.41464 [cs.LG]
(or arXiv:2610.41464v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2610.41464 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 309 cuvinte
Subiecte conexe
Traditional machine learning vs. deep learning from dynamic graph representations of proteins' 3D folds in the task of protein structure classification
arXiv · 15 iun. 2026
ReDiSC: A Reparameterized Masked Diffusion Model for Scalable Node Classification with Structured Predictions
arXiv · 30 iul. 2026
Linear-Core Surrogates: Smooth Loss Functions with Linear Rates for Classification and Structured Prediction
arXiv · 27 oct. 2026
Retrieving Relations, Detecting Fallacies: A RAG Approach to Political Debate Analysis
arXiv · 1 sept. 2026