Hui-Jun Zhang

Entity-relation joint extraction model based on multi-branch dilated convolutions and multi-head cross-attention mechanism

A joint extraction framework that combines multi-branch dilated convolutions with a multi-head cross-attention mechanism together with a learnable gated fusion mechanism for feature interaction and adaptive fusion, thereby improving relation representation.