Active Few-Shot Learning (AFSL) is an effective paradigm for improving the performance of large language models under limited annotation budgets. To address the inefficiency of conventional fine-tuning objectives in AFSL, this paper proposes a supervised contrastive fine-tuning framework specifically designed for natural language processing (NLP) text classification tasks. By integrating Supervised Contrastive Learning (SCL) with Hard Negative Mining (HNM), the proposed framework optimizes the embedding space through an enhanced hybrid loss function, thereby improving the utilization efficiency of labeled samples. Extensive experiments on five benchmark datasets show that, under a fixed state-of-the-art (SOTA) query strategy, our method consistently outperforms baseline models in text classification performance, and exhibits strong generalizability across different backbone architectures and acquisition functions. These findings demonstrate that optimizing how to learn—through improved learning objectives—provides a complementary direction to existing query strategies in advancing AFSL.

Recently, prompt-tuning has achieved very significant results for few-shot tasks. The core idea of prompt-tuning is to insert prompt templates into the input, thus converting the classification task into a masked language modeling problem. However, for few-shot relation extraction tasks, how to mine more information from limited resources becomes particularly important. In this paper, we first construct a global relation graph based on label consistency to optimize the feature representation of samples between different relations. Then the global relation graph is further divided to form a local relation subgraph for each relation type to optimize the feature representation of samples within the same relation. This fully uses the limited supervised information and improves the tuning efficiency. In addition, the existence of rich semantic knowledge in relation labels cannot be ignored. For this reason, this paper incorporates the knowledge in relation labels into prompt-tuning. Specifically, the potential knowledge implicit in relation labels is injected into constructing learnable prompt templates. In this paper, we conduct extensive experiments on four datasets under low-resource settings, showing that this method achieves significant results.