Is One Token All It Takes? Graph Pooling Tokens for LLM-based GraphQA

• URL: http://arxiv.org/abs/2604.00342v1
• Date: Wed, 01 Apr 2026 00:34:10 GMT
• Title: Is One Token All It Takes? Graph Pooling Tokens for LLM-based GraphQA
• Abstract summary: We show how to increase the bandwidth of the graph-to-LLM interface via multi-token pooling.<n>We also use global attention mechanisms to enhance the semantic quality of the graph encoder.<n>Our analysis reveals that the GraphQA benchmark suffers from representational saturation.
• Score: 1.2239546747355887
• License: http://creativecommons.org/licenses/by-nc-sa/4.0/
• Abstract: The integration of Graph Neural Networks (GNNs) with Large Language Models (LLMs) has emerged as a promising paradigm for Graph Question Answering (GraphQA). However, effective methods for encoding complex structural information into the LLM's latent space remain an open challenge. Current state-of-the-art architectures, such as G-Retriever, typically rely on standard GNNs and aggressive mean pooling to compress entire graph substructures into a single token, creating a severe information bottleneck. This work mitigates this bottleneck by investigating two orthogonal strategies: (1) increasing the bandwidth of the graph-to-LLM interface via multi-token pooling, and (2) enhancing the semantic quality of the graph encoder via global attention mechanisms. We evaluate a suite of hierarchical pruning and clustering-based pooling operators including Top-k, SAGPool, DiffPool, MinCutPool, and Virtual Node Pooling (VNPool) to project graph data into multiple learnable tokens. Empirically, we demonstrate that while pooling introduces significant instability during soft prompt tuning, the application of Low-Rank Adaptation (LoRA) effectively stabilizes specific hierarchical projections (notably VNPool and pruning methods), though dense clustering operators remain challenging. This stabilization allows compressed representations to rival full-graph baselines (achieving ~73% Hit@1 on WebQSP). Conceptually, we demonstrate that a Graph Transformer with VNPool implementation functions structurally as a single-layer Perceiver IO encoder. Finally, we adapt the FandE (Features and Edges) Score to the generative GraphQA domain. Our analysis reveals that the GraphQA benchmark suffers from representational saturation, where target answers are often highly correlated with isolated node features. The implementation is available at https://github.com/Agrover112/G-Retriever/tree/all_good/

Related papers

• Semi-supervised Instruction Tuning for Large Language Models on Text-Attributed Graphs [62.544129365882014]
  We propose a novel Semi-supervised Instruction Tuning pipeline for Graph Learning, named SIT-Graph.<n> SIT-Graph is model-agnostic and can be seamlessly integrated into any graph instruction tuning method that utilizes LLMs as the predictor.<n>Extensive experiments demonstrate that when incorporated into state-of-the-art graph instruction tuning methods, SIT-Graph significantly enhances their performance on text-attributed graph benchmarks.
  arXiv  Detail & Related papers  (2026-01-19T08:10:53Z)
• Enhancing Graph Classification Robustness with Singular Pooling [11.537806919295098]
  Graph Neural Networks (GNNs) have achieved strong performance across a range of graph representation learning tasks, yet their adversarial robustness in graph classification remains underexplored compared to node classification.<n>We present a theoretical analysis of standard flat pooling methods, deriving upper bounds on their adversarial risk and identifying their vulnerabilities under different attack scenarios and graph structures.<n>Motivated by these insights, we propose textitRobust Singular Pooling (RS-Pool), a novel pooling strategy that leverages the dominant singular vector of the node embedding matrix to construct a robust graph-level representation.
  arXiv  Detail & Related papers  (2025-10-26T12:07:06Z)
• GILT: An LLM-Free, Tuning-Free Graph Foundational Model for In-Context Learning [50.40400074353263]
  Graph Neural Networks (GNNs) are powerful tools for precessing relational data but often struggle to generalize to unseen graphs.<n>We introduce textbfGraph textbfIn-context textbfL textbfTransformer (GILT), a framework built on an LLM-free and tuning-free architecture.
  arXiv  Detail & Related papers  (2025-10-06T08:09:15Z)
• Beyond Message Passing: Neural Graph Pattern Machine [50.78679002846741]
  We introduce the Neural Graph Pattern Machine (GPM), a novel framework that bypasses message passing by learning directly from graph substructures.<n>GPM efficiently extracts, encodes, and prioritizes task-relevant graph patterns, offering greater expressivity and improved ability to capture long-range dependencies.
  arXiv  Detail & Related papers  (2025-01-30T20:37:47Z)
• Graph Pooling via Ricci Flow [1.1126342180866644]
  We introduce a graph pooling operator (ORC-Pool) which utilizes a characterization of the graph's geometry via Ollivier's discrete Ricci curvature and an associated geometric flow. ORC-Pool extends such clustering approaches to attributed graphs, allowing for the integration of geometric coarsening into Graph Neural Networks as a pooling layer.
  arXiv  Detail & Related papers  (2024-07-05T03:26:37Z)
• SPGNN: Recognizing Salient Subgraph Patterns via Enhanced Graph Convolution and Pooling [25.555741218526464]
  Graph neural networks (GNNs) have revolutionized the field of machine learning on non-Euclidean data such as graphs and networks. We propose a concatenation-based graph convolution mechanism that injectively updates node representations. We also design a novel graph pooling module, called WL-SortPool, to learn important subgraph patterns in a deep-learning manner.
  arXiv  Detail & Related papers  (2024-04-21T13:11:59Z)
• Multi-Granularity Graph Pooling for Video-based Person Re-Identification [14.943835935921296]
  graph neural networks (GNNs) are introduced to aggregate temporal and spatial features of video samples. Existing graph-based models, like STGCN, perform the textitmean/textitmax pooling on node features to obtain the graph representation. We propose the graph pooling network (GPNet) to learn the multi-granularity graph representation for the video retrieval.
  arXiv  Detail & Related papers  (2022-09-23T13:26:05Z)
• Dynamic Graph Message Passing Networks for Visual Recognition [112.49513303433606]
  Modelling long-range dependencies is critical for scene understanding tasks in computer vision. A fully-connected graph is beneficial for such modelling, but its computational overhead is prohibitive. We propose a dynamic graph message passing network, that significantly reduces the computational complexity.
  arXiv  Detail & Related papers  (2022-09-20T14:41:37Z)
• Higher-order Clustering and Pooling for Graph Neural Networks [77.47617360812023]
  Graph Neural Networks achieve state-of-the-art performance on a plethora of graph classification tasks. HoscPool is a clustering-based graph pooling operator that captures higher-order information hierarchically. We evaluate HoscPool on graph classification tasks and its clustering component on graphs with ground-truth community structure.
  arXiv  Detail & Related papers  (2022-09-02T09:17:10Z)
• VQ-GNN: A Universal Framework to Scale up Graph Neural Networks using Vector Quantization [70.8567058758375]
  VQ-GNN is a universal framework to scale up any convolution-based GNNs using Vector Quantization (VQ) without compromising the performance. Our framework avoids the "neighbor explosion" problem of GNNs using quantized representations combined with a low-rank version of the graph convolution matrix.
  arXiv  Detail & Related papers  (2021-10-27T11:48:50Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.