Maritime search and rescue (SAR) operations play a crucial role in reducing fatalities and mitigating human suffering. Compared to short-range maritime SAR, long-range maritime SAR (LRMSAR) is more challenging due to the far distance from the shore, changeful weather, and less available resources. Such an operation put high requirements on decision makers to timely assign multiple resources, such as aircraft and vessels to deal with the emergency. However, most current researches pay attention to assign only one kind of resource, while practically, multiple resources are necessary for LRMSAR. Thus, a method is proposed to provide support for decision makers to allocate multiple resources in dealing with LRMSAR problem; to ensure the sustainable use of resources. First, by analyzing the factors involved in the whole process, we formulated the problem as a multi-objective optimization problem, the objective of which was to maximize both the probability of completing the tasks and the utilities of allocated resources. Based on the theory of search, an integer nonlinear programming (INLP) model was built for different tasks. Second, in order to solve the non-deterministic polynomial-time hardness (NP-hard) model, by constructing a rule base, candidate solutions can be found to improve the calculation efficiency. Furthermore, in order to obtain the optimal scheme, the Non-dominated Sorting Genetic Algorithm II (NSGA-II) was applied to the candidate solution sets to approximate Pareto fronts. Finally, an emergency case of Chinese Bohai Sea was used to demonstrate the effectiveness of the proposed model. In the study, 11 resource allocation schemes were obtained to respond to the emergency, and calculation processes of schemes were further analyzed to demonstrate our model’s rationality. Results showed that the proposed models provide decision-makers with scientific decision support on different emergency tasks.

Aiming at the characteristics of multivariate time series, such as high dimensionality, the strong correlation of variables, and unequal length of sequences, this paper proposes the MSN-WDTW similarity measure for multivariate time series based on multi-dimensional segmentation norm representation and weighted dynamic time warping. First, a hierarchical detection algorithm is proposed to gradually detect the important points of the multivariate time series, and uniformly segment the sequences through an error rate-based method. On the basis of the obtained segmentation, the Frobenius norm is introduced to uniformly represent each segmentation matrix and its covariance matrix, then the multivariate time series is transformed into a univariate time series. Besides, the weighted dynamic time warping algorithm is used to measure the similarity of the compressed sequences of unequal length. Finally, both the similarity search and nearest neighbor classification experiments are carried out and verify the effectiveness and feasibility of the proposed MSN-WDTW similarity measure for MTS.

The growth of e-government has rendered automated element extraction from official documents a critical bottleneck for administrative efficiency. The core challenge lies in unifying deep semantic understanding with the structured domain knowledge required to interpret complex formats and specialized terminology. To address the limitations of existing methods, we propose a hybrid framework. Our approach leverages a BERT-CRF model for robust sequence labeling, a knowledge graph (KG)-driven retrieval system to ground the model in verifiable facts, and a large language model (LLM) as a reasoning engine to resolve ambiguities and identify complex relationships. Validated on the DovDoc-CN dataset, our framework achieves a macro-average F1 score of 0.850, outperforming the BiLSTM-CRF baseline by 2.41 percentage points, and demonstrates high consistency, with a weighted F1 score of 0.984. The low standard deviation in the validation set further indicates the model’s stable performance across different subsets. These results confirm that our integrated approach provides an efficient and reliable solution for intelligent document processing, effectively handling the format diversity and specialized knowledge characteristic of government documents.

Different from numerical attributes, measuring the similarity between categorical attributes is more complex due to their non-inherently ordered characteristic, especially in an unsupervised scheme. This work, therefore, presents a new method, Heterogeneous Graph-based Similarity measure (HGS), to measure the similarity between categorical data for unsupervised learning. In order to capture the possible complex relationships hidden among attributes, a heterogeneous weighted graph is creatively constructed by extracting the information from categorical data. Both objects and attribute values are represented as nodes and their occurrence and co-occurrence relationships are shown as edges. Based on a derived node-pair graph, three rules are used to iteratively update the similarity scores between object pairs and attribute-value pairs until the scores converge. We also analyze its complexities and validate the metric properties and convergence. In experiment validation, five state-of-the-art measures are compared with HGS based on 20 UCI datasets and 6 high-dimensional datasets in the medical domain in both k-modes and spectral clustering and similarity search experiments. The results show although no measure can outperform all other measures on all datasets, HGS can perform better in both clustering and similarity search tasks on the whole. Finally, six studies further discuss the convergence, time cost, and parameter sensitivity of the HGS, explore its application to imbalanced class distribution, and compare it with its variants by different initialization and graph construction.

Perception model is an important research field of CGF behavior modeling. As the input of the reasoning module, whether the information obtained from the perception model is precise is of great significance for the simulation fidelity. This paper first investigates the related researches on CGF perception model, and focuses on analyzing the attention mechanism, action time, and action way. Besides, a virtual human perception framework based on attention mechanism is proposed by constructing the knowledge base of the attention and modeling of information feedback mechanism. Compared to the traditional perception model, the proposed framework is more effective to capture and deal with the information from simulation environment.