TL;DR
DeepIDDFS is introduced, a scalable node embedding designed for ER in large and dynamic graphs that employs an iterative deepening depth-first search (IDDFS), integrates BERT-based semantic embeddings to handle noisy and inconsistent attributes, and incorporates a time-aware aggregation mechanism that emphasizes recent interactions to better model evolving graph structures.
This work proposes H2Table (Hierarchical Hypergraph-Enhanced Table Reasoning), a novel framework that represents complex tables as hierarchical nested hypergraphs, and designs a tailored hypergraph encoder to facilitate message passing between hyperedges and nodes within complex tables.
This work addresses the need for efficient and reusable vector embeddings for large-scale RDF Knowledge Graphs, focusing on the widely used DrugBank dataset, and provides a comprehensive, reusable resource that facilitates research while promoting computational efficiency and environmental sustainability.
An adaptive community search framework ECHO is proposed, which consistently outperforms state-of-the-art methods in terms of community quality while achieving superior search efficiency.
TopoCompress is introduced, a training-free and model-agnostic framework that compresses long contexts by selecting coherent semantic spans by selecting coherent semantic spans and achieves performance comparable to the strongest baseline while using a 4x smaller compression budget.
A large-scale, universal, and joint multi-domain heterogeneous graph dataset named UniHG to facilitate heterogeneous graph representation learning and cross-domain knowledge mining and comprehensive benchmark results demonstrate that the model significantly out-performs existing methods with an accuracy improvement of...
HyMAGE models each node as an autonomous agent and leverages LLMs for local-level semantic selection, so that hyperedge formation and dissolution emerge from local semantic affinity and structural context in a self-organizing manner.
We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.
