Numerous investigations have demonstrated that integrating syntactic information into character - level embeddings can boost the performance of Chinese Named Entity Recognition (CNER). Yet, current methods mainly concentrate on using syntactic information to improve character - level semantic embeddings, frequently overlooking the interplay between syntactic and semantic information. To address this problem, we present a new method that leverages the Graph Convolutional Network (GCN) to facilitate a thorough fusion of syntactic and semantic information. Specifically, we initially extract semantic features with the pre - trained language model RoformerV2. Then, we create a syntactic tree with external tools and transform it into an adjacency matrix. Subsequently, GCN are employed to blend syntactic and semantic information. The fused information undergoes further semantic enhancement to yield feature vectors with dual syntactic and semantic reinforcement. Finally, we implement the GlobalPointer framework, which takes into account both the start and end locations of entities, to achieve accurate entity prediction. We carry out experiments on four standard datasets, and the outcomes confirm the efficacy of our proposed method. Additionally, considering the dearth of research on few-shot CNER for Chinese, we utilize the youku dataset to mimic a small-sample situation and evaluate the performance of our method, thereby showing its effectiveness in small-sample Chinese named entity recognition.