Hound: Hunting Supervision Signals for Few and Zero Shot Node Classification on Text-attributed Graph

TL;DR

Hound enhances few- and zero-shot node classification on text-attributed graphs by introducing diverse supervision signals through innovative augmentation techniques, significantly improving accuracy over existing methods.

Contribution

The paper proposes Hound, a novel framework that leverages node perturbation, text matching, and semantics negation to provide additional supervision signals beyond node-text pairs.

Findings

Hound outperforms 13 state-of-the-art baselines on 5 datasets.

Accuracy improvements over the best baseline are typically over 5%.

The method effectively enhances classification performance in low-supervision scenarios.

Abstract

Text-attributed graph (TAG) is an important type of graph structured data with text descriptions for each node. Few- and zero-shot node classification on TAGs have many applications in fields such as academia and social networks. However, the two tasks are challenging due to the lack of supervision signals, and existing methods only use the contrastive loss to align graph-based node embedding and language-based text embedding. In this paper, we propose Hound to improve accuracy by introducing more supervision signals, and the core idea is to go beyond the node-text pairs that come with data. Specifically, we design three augmentation techniques, i.e., node perturbation, text matching, and semantics negation to provide more reference nodes for each text and vice versa. Node perturbation adds/drops edges to produce diversified node embeddings that can be matched with a text. Text matching retrieves texts with similar embeddings to match with a node. Semantics negation uses a negative prompt to construct a negative text with the opposite semantics, which is contrasted with the original node and text. We evaluate Hound on 5 datasets and compare with 13 state-of-the-art baselines. The results show that Hound consistently outperforms all baselines, and its accuracy improvements over the best-performing baseline are usually over 5%.