The extensive range of food safety standards poses a significant challenge to efficiently accessing specific information within this domain, necessitating innovative solutions to streamline the process. In response, researchers are focusing on constructing a knowledge graph based on food safety standards to facilitate efficient associative querying. Named entity recognition is a pivotal element in this endeavor due to its critical impact on the accuracy and quality of the knowledge graph. To address the nuanced challenges of accurately identifying nested entity boundaries and rectifying entity class imbalances in food safety standards, we present PGD-GP, a novel Chinese named entity recognition model. This model is based on Projected Gradient Descent for adversarial training and Global Pointer. The model innovatively refines the Chinese Bert model at the encoding layer, employing the adversarial training method PGD to iteratively introduce perturbations to character vectors, thereby significantly enhancing the model's robustness and adaptability to texts. The decoding layer leverages Global Pointer to accurately determine dependencies and relative positional relationships between characters, thus facilitating more precise recognition of entity boundaries. To combat the issue of class imbalance, Circle Loss is utilized as the loss function. We developed and annotated the Food Safety Standard Dataset using a specifically tailored ontology rule for food safety standards. Comparative experiments conducted on the Food Safety Standard Dataset and the public Resume dataset demonstrate that PGD-GP surpasses six mainstream baseline models in performance, thereby validating the effectiveness and robustness of PGD-GP. Building upon the foundation of PGD-GP and the Food Safety Standard Dataset, we implemented a prototype system that integrates a food safety standard-based knowledge graph with associated queries. This system serves as an efficient, accurate, and comprehensive intelligent assistant, enabling researchers to effectively acquire food safety standard information.

In recent years, projection-based no-reference point cloud quality metrics have shown remarkable performance on public datasets. These metrics involve projecting the point cloud onto a 2D space from multiple viewpoints and applying traditional image quality metrics to the resulting projected images. However, these existing methods suffer from a measurable projection strategy, which would induce information loss or redundancy leading to training bias and reduced generalizability. To overcome these limitations, we propose AKSNet, a novel no-reference point cloud quality metric that utilizes an adaptive keyframe selection strategy. In particular, we treat the process of point cloud projection as video sequences captured from successive viewpoints and design a keyframe selection module to pick the most representative frames. Based on the selected frames, we can subsequently predict the quality score of the point cloud. Experimental results demonstrate that our method achieves competitive performance in comparison with state-of-the-art methods. While the proposed adaptive keyframe selecting strategy can effectively promote the generalization performance.