Multi-hop question answering (MQA) requires models to perform multi-step reasoning and integrate multiple knowledge sources. However, existing methods combining pre-trained language models (PLMs) and graph neural networks (GNNs) often suffer from low computational efficiency, insufficient deep semantic fusion, and imbalanced modeling of heterogeneous relations. To solve these problems, we propose a Dynamic Hierarchical Adaptive Graph Convolution Network (DHACNet). First, to deal with the issues of insufficient computational efficiency and feature interpretability, we introduce Dynamic Sparse Activation (DSA). A trainable gate unit is used to generate importance masks for the encoder outputs, keeping only the task-relevant neurons. This greatly decreases the computational burden and enhances the interpretability of the model’s decisions. Second, to alleviate insufficient deep semantic fusion, we design a Hierarchical Feature Fusion (HFF) mechanism. It adaptively weights and fuses hidden states from different layers, enhancing the extraction and representation of deep textual semantics. Furthermore, for graph structure modeling, we present Adaptive Graph Convolution (AGC), which assigns learnable weights to different edge types in the graph, thereby improving heterogeneous relation modeling. Finally, hierarchical graph pooling is introduced, which integrates attention mechanism and Top-$K$ selection to achieve efficient and robust graph-level representation. The experimental results show that our proposed model maintains the symmetry between the text representation and graph representation through adaptive layered fusion and relational perceptual graph propagation. This symmetry-aware reasoning process encourages semantic consistency during multi-hop inference and makes knowledge integration more robust.

The purpose of few-shot relation extraction (RE) is to recognize the relationship between specific entity pairs in text when there are a limited number of labeled samples. A few-shot RE method based on a prototype network, which constructs relation prototypes by relying on the support set to assign labels to query samples, inherently leverages the symmetry between support and query processing. Although these methods have achieved remarkable results, they still face challenges such as the misjudging of noisy samples or outliers, as well as distinguishing semantic similarity relations. To address the aforementioned challenges, we propose a novel semantic enhanced prototype network, which can integrate the semantic information of relations more effectively to promote more expressive representations of instances and relation prototypes, so as to improve the performance of the few-shot RE. Firstly, we design a prompt encoder to uniformly process different prompt templates for instance and relation information, and then utilize the powerful semantic understanding and generation capabilities of large language models (LLMs) to obtain precise semantic representations of instances, their prototypes, and conceptual prototypes. Secondly, graph attention learning techniques are introduced to effectively extract specific-relation features between conceptual prototypes and isomorphic instances while maintaining structural symmetry. Meanwhile, a prototype-level contrastive learning strategy with bidirectional feature symmetry is proposed to predict query instances by integrating the interpretable features of conceptual prototypes and the intra-class shared features captured by instance prototypes. In addition, a clustering loss function was designed to guide the model to learn a discriminative metric space with improved relational symmetry, effectively improving the accuracy of the model’s relationship recognition. Finally, the experimental results on the FewRel1.0 and FewRel2.0 datasets show that the proposed approach delivers improved performance compared to existing advanced models in the task of few-shot RE.