Last year has witnessed the re-flourishment of tag-aware recommender systems
supported by the LLM-enriched tags. Unfortunately, though large efforts have
been made, current solutions may fail to describe the diversity and uncertainty
inherent in user preferences with only tag-driven profiles. Recently, with the
development of geometry-based techniques, e.g., box embedding, diversity of
user preferences now could be fully modeled as the range within a box in high
dimension space. However, defect still exists as these approaches are incapable
of capturing high-order neighbor signals, i.e., semantic-rich multi-hop
relations within the user-tag-item tripartite graph, which severely limits the
effectiveness of user modeling. To deal with this challenge, in this paper, we
propose a novel algorithm, called BoxGNN, to perform the message aggregation
via combination of logical operations, thereby incorporating high-order
signals. Specifically, we first embed users, items, and tags as hyper-boxes
rather than simple points in the representation space, and define two logical
operations to facilitate the subsequent process. Next, we perform the message
aggregation mechanism via the combination of logical operations, to obtain the
corresponding high-order box representations. Finally, we adopt a volume-based
learning objective with Gumbel smoothing techniques to refine the
representation of boxes. Extensive experiments on two publicly available
datasets and one LLM-enhanced e-commerce dataset have validated the superiority
of BoxGNN compared with various state-of-the-art baselines. The code is
released online

提出了一种新的算法，称为 BoxGNN，通过组合逻辑操作来执行消息聚合，从而融合高阶信号，并通过基于体积的学习目标和 Gumbel 平滑技术来优化盒子的表示。在两个公开数据集和一个 LLM 增强的电子商务数据集上的大量实验验证了 BoxGNN 相对于各种最先进的基线方法的优越性。

当盒子遇见图神经网络时的标签感知推荐

When Box Meets Graph Neural Network in Tag-aware Recommendation

Graph neural networks (GNNs), and especially message-passing neural networks,
excel in various domains such as physics, drug discovery, and molecular
modeling. The expressivity of GNNs with respect to their ability to
discriminate non-isomorphic graphs critically depends on the functions employed
for message aggregation and graph-level readout. By applying signal propagation
theory, we propose a variance-preserving aggregation function (VPA) that
maintains expressivity, but yields improved forward and backward dynamics.
Experiments demonstrate that VPA leads to increased predictive performance for
popular GNN architectures as well as improved learning dynamics. Our results
could pave the way towards normalizer-free or self-normalizing GNNs.

通过应用信号传播理论，我们提出了一种保持表达能力但具有改进的前向和后向动力学的方差保持聚合函数 (VPA)，实验证明 VPA 能够提高流行的 GNN 架构的预测性能和学习动力学，这可能为无归一化或自归一化 GNN 铺平道路。

GNN-VPA：图神经网络的方差保持聚合策略

GNN-VPA: A Variance-Preserving Aggregation Strategy for Graph Neural  Networks

Graph Neural Networks (GNNs) have emerged as a revolutionary paradigm in the
realm of machine learning, offering a transformative approach to dissect
intricate relationships inherent in graph-structured data. The foundational
architecture of most GNNs involves the dissemination of information through
message aggregation and transformation among interconnected nodes, a mechanism
that has demonstrated remarkable efficacy across diverse applications
encompassing node classification, link prediction, and recommendation systems.
Nonetheless, their potential prowess encounters a restraint intrinsic to
scenarios necessitating extensive contextual insights. In certain contexts,
accurate predictions hinge not only upon a node's immediate local surroundings
but also on interactions spanning far-reaching domains. This intricate demand
for long-range information dissemination exposes a pivotal challenge recognized
as "over-squashing," wherein the fidelity of information flow from distant
nodes becomes distorted. This phenomenon significantly curtails the efficiency
of message-passing mechanisms, particularly for tasks reliant on intricate
long-distance interactions. In this comprehensive article, we illuminate the
prevalent constraint of over-squashing pervading GNNs. Our exploration entails
a meticulous exposition of the ongoing efforts by researchers to improve the
ramifications posed by this limitation. Through systematic elucidation, we
delve into strategies, methodologies, and innovations proposed thus far, all
aimed at mitigating the detriments of over-squashing. By shedding light on this
intricately woven issue, we aim to contribute to a nuanced understanding of the
challenges within the GNN landscape and the evolving solutions designed to
surmount them.

对图神经网络（GNN）中普遍存在的过烈抑制问题进行了细致剖析，探讨了研究者目前提出的改善方法和创新，旨在解决 GNN 领域面临的挑战。

图神经网络中的过度压缩现象：一项综合调查

Over-Squashing in Graph Neural Networks: A Comprehensive survey

Graph Convolutional Networks (GCNs) have been drawing significant attention
with the power of representation learning on graphs. Unlike Convolutional
Neural Networks (CNNs), which are able to take advantage of stacking very deep
layers, GCNs suffer from vanishing gradient, over-smoothing and over-fitting
issues when going deeper. These challenges limit the representation power of
GCNs on large-scale graphs. This paper proposes DeeperGCN that is capable of
successfully and reliably training very deep GCNs. We define differentiable
generalized aggregation functions to unify different message aggregation
operations (e.g. mean, max). We also propose a novel normalization layer namely
MsgNorm and a pre-activation version of residual connections for GCNs.
Extensive experiments on Open Graph Benchmark (OGB) show DeeperGCN
significantly boosts performance over the state-of-the-art on the large scale
graph learning tasks of node property prediction and graph property prediction.
Please visit this https URL for more information.

本文提出了一种名为 DeeperGCN 的新型 Graph Convolutional Networks，它有效地解决了深度较大的 GCNs 存在的梯度消失、过度平滑和过度拟合等问题，实现了对大规模图数据的成功且可靠的训练，并在 Open Graph Benchmark (OGB) 上进行的实验证明其在节点属性预测和图形属性预测的大规模图形学习任务中比现有技术有显著提高。