Search

Urban traffic prediction is essential for intelligent transportation systems. However, traffic data often exhibit highly complex spatio-temporal correlations, posing challenges for accurate forecasting. Graph neural networks have demonstrated an outstanding ability in capturing spatial correlations and are now extensively applied to traffic prediction. However, many graph-based methods neglect the dynamic spatial features between road segments and the continuity of spatial features across adjacent time steps, leading to subpar predictive performance. This paper proposes a Dynamic Spatio-Temporal Graph Fusion Convolutional Network (DSTGFCN) to enhance the accuracy of traffic prediction. Specifically, we designed a dynamic graph fusion module without prior road spatial information, which extracts dynamic spatial information among roads from observed data. Subsequently, we fused the dynamic spatial features of the current time step and adjacent time steps to generate a dynamic graph for each time step. The graph convolutional gated recurrent network was employed to model the spatio-temporal correlations jointly. Additionally, residual connections were added to the model to enhance the ability to extract long-term temporal relationships. Finally, we conducted experiments on six publicly available traffic datasets, and the results demonstrated that DSTGFCN outperforms the baseline models with state-of-the-art predictive performance. [ABSTRACT FROM AUTHOR]

Named entity recognition involves two main types: nested named entity recognition and flat named entity recognition. The span-based approach treats nested entities and flat entities uniformly by classifying entities on a span representation. However, the span-based approach ignores the local features within the entities and the relative position features between the head and tail tokens, which affects the performance of entity recognition. To address these issues, we propose a nested entity recognition model using a convolutional block attention module and rotary position embedding for local features and relative position features enhancement. Specifically, we apply rotary position embedding to the sentence representation and capture the semantic information between the head and tail tokens using a biaffine attention mechanism. Meanwhile, the convolution module captures the local features within the entity to generate the span representation. Finally, the two parts of the representation are fused for entity classification. Extensive experiments were conducted on five widely used benchmark datasets to demonstrate the effectiveness of our proposed model. [ABSTRACT FROM AUTHOR]

Deep learning techniques have demonstrated significant advancements in the task of text classification. Regrettably, the majority of these techniques necessitate a substantial corpus of annotated data to achieve optimal performance. Meta-learning has yielded intriguing outcomes in few-shot learning tasks, showcasing its potential in advancing the field. However, the current meta-learning methodologies are susceptible to overfitting due to the mismatch between a small number of samples and the complexity of the model. To mitigate this concern, we propose a Prompt-based Graph Convolutional Adversarial (PGCA) meta-learning framework, aiming to improve the adaptability of complex models in a few-shot scenario. Firstly, leveraging prompt learning, we generate embedding representations that bridge the downstream tasks. Then, we design a meta-knowledge extractor based on a graph convolutional neural network (GCN) to capture inter-class dependencies through instance-level interactions. We also integrate the adversarial network architecture into a meta-learning framework to extend sample diversity through adversarial training and improve the ability of the model to adapt to new tasks. Specifically, we mitigate the impact of extreme samples by introducing external knowledge to construct a list of class prototype extensions. Finally, we conduct a series of experiments on four public datasets to demonstrate the effectiveness of our proposed method. [ABSTRACT FROM AUTHOR]

Few-shot named entity recognition requires sufficient prior knowledge to transfer valuable knowledge to the target domain with only a few labeled examples. Existing Chinese few-shot named entity recognition methods suffer from inadequate prior knowledge and limitations in feature representation. In this paper, we utilize enhanced Span and Label semantic representations for Chinese few-shot Named Entity Recognition (SLNER) to address the problem. Specifically, SLNER utilizes two encoders. One encoder is used to encode the text and its spans, and we employ the biaffine attention mechanism and self-attention to obtain enhanced span representations. This approach fully leverages the internal composition of entity mentions, leading to more accurate feature representations. The other encoder encodes the full label names to obtain label representations. Label names are broad representations of specific entity categories and share similar semantic meanings with entities. This similarity allows label names to offer valuable prior knowledge in few-shot scenarios. Finally, our model learns to match span representations with label representations. We conducted extensive experiments on three sampling benchmark Chinese datasets and a self-built food safety risk domain dataset. The experimental results show that our model outperforms the F1 scores of 0.20–6.57% of previous state-of-the-art methods in few-shot settings. [ABSTRACT FROM AUTHOR]

The primary task of few-shot relation extraction is to quickly learn the features of relation classes from a few labelled instances and predict the semantic relations between entity pairs in new instances. Most existing few-shot relation extraction methods do not fully utilize the relation information features in sentences, resulting in difficulties in improving the performance of relation classification. Some researchers have attempted to incorporate external information, but the results have been unsatisfactory when applied to different domains. In this paper, we propose a method that utilizes triple information for data augmentation, which can alleviate the issue of insufficient instances and possesses strong domain adaptation capabilities. Firstly, we extract relation and entity pairs from the instances in the support set, forming relation triple information. Next, the sentence information and relation triple information are encoded using the same sentence encoder. Then, we construct an interactive attention module to enable the query set instances to interact separately with the support set instances and relation triple instances. The module pays greater attention to highly interactive parts between instances and assigns them higher weights. Finally, we merge the interacted support set representation and relation triple representation. To our knowledge, we are the first to propose a method that utilizes triple information for data augmentation in relation extraction. In our experiments on the standard datasets FewRel1.0 and FewRel2.0 (domain adaptation), we observed substantial improvements without including external information. [ABSTRACT FROM AUTHOR]

Food safety is closely related to human health. Therefore, named entity recognition technology is used to extract named entities related to food safety, and building a regulatory knowledge graph in the field of food safety can help relevant authorities to regulate food safety issues and mitigate the hazards caused by food safety problems. However, there is no publicly available named entity recognition dataset in the food safety domain. In contrast, the non-standardized Chinese short texts generated from user comments on the web contain rich implicit information that can help identify named entities in specific domains (e.g., food safety domain) where the corpus is scarce. Therefore, in this paper, named entities related to food safety are extracted from these unstandardized texts on the web. However, the existing Chinese named entity identification methods are mainly for standardized texts. Meanwhile, these unstandardized texts have the following problems: (1) their corpus size is small; (2) there are various new and wrong words and noise; (3) and they do not follow strict syntactic rules. These problems make the recognition of Chinese named entities for online texts more challenging. Therefore, this paper proposes the ERNIE-Adv-BiLSTM-Att-CRF model to improve the recognition of food safety domain entities in unstandardized texts. Specifically, adversarial training is added to the model training as a regularization method to alleviate the influence of noise on the model, while self-attention is added to the BiLSTM-CRF model to capture features that significant impact entity classification and improve the accuracy of entity classification. This paper conducts experiments on the public dataset Weibo NER and the self-built food domain dataset Food. The experimental results show that our model achieves a SOTA performance of 72.64% and a good performance of 69.68% for F1 values on the public and self-built datasets, respectively. The validity and reasonableness of our model are verified. In addition, the paper further analyses the impact of various components and settings on the model. The study has practical implications in the field of food safety. [ABSTRACT FROM AUTHOR]