In light of the rapid expansion of maritime trade, the maritime transportation industry has experienced burgeoning growth and complexity. The deployment of trajectory prediction technology is paramount in safeguarding navigational safety. Due to limitations in design complexity and the high costs of data fusion, current deep learning methods struggle to effectively integrate high-level semantic cues, such as vessel type, geographical identifiers, and navigational states, within predictive frameworks. Yet, these data contain abundant information regarding vessel categories or operational scenarios. Inspired by the robust semantic comprehension exhibited by large language models (LLMs) in natural language processing, this study introduces a trajectory prediction method leveraging LLMs. Initially, Automatic Identification System (AIS) data undergoes processing to eliminate incomplete entries, thereby selecting trajectories of high quality. Distinct from prior research that concentrated solely on vessel position and velocity, this study integrates ship identity, spatiotemporal trajectory, and navigational information through prompt engineering, empowering the LLM to extract multidimensional semantic features of trajectories from comprehensive natural language narratives. Thus, the LLM can amalgamate multi-source semantics with zero marginal cost, significantly enhancing its understanding of complex maritime environments. Subsequently, a supervised fine-tuning approach rooted in Low-Rank Adaptation (LoRA) is applied to train the chosen LLMs. This enables rapid adaptation of the LLM to specific maritime areas or vessel classifications by modifying only a limited subset of parameters, thereby appreciably diminishing both data requirements and computational costs. Finally, representative metrics are utilized to evaluate the efficacy of the model training and to benchmark its performance against prevailing advanced models for ship trajectory prediction. The results indicate that the model demonstrates notable performance in short-term predictions fFor instance, with a prediction step of 1 h, the average distance errors for VTLLM and TrAISformer are 5.26 nmi and 6.12 nmi, respectively, resulting in a performance improvement of approximately 14.05%), having identified certain patterns and features, such as linear movements and turns, from the training data.

With the rapid growth in maritime traffic, navigational safety has become a pressing concern. Some vessels deliberately manipulate their type information to evade regulatory oversight, either to circumvent legal sanctions or engage in illicit activities. Such practices not only undermine the accuracy of maritime supervision but also pose significant risks to maritime traffic management and safety. Therefore, accurately identifying vessel types is essential for effective maritime traffic regulation, combating maritime crimes, and ensuring safe maritime transportation. However, the existing methods fail to fully exploit the long-term sequential dependencies and intricate mobility patterns embedded in vessel trajectory data, leading to suboptimal identification accuracy and reliability. To address these limitations, we propose MESTR, a Multi-Task Enhanced Ship-Type Recognition model based on Automatic Identification System (AIS) data. MESTR leverages a Transformer-based deep learning framework with a motion-pattern-aware trajectory segment masking strategy. By jointly optimizing two learning tasks—trajectory segment masking prediction and ship-type prediction—MESTR effectively captures deep spatiotemporal features of various vessel types. This approach enables the accurate classification of six common vessel categories: tug, sailing, fishing, passenger, tanker, and cargo. Experimental evaluations on real-world maritime datasets demonstrate the effectiveness of MESTR, achieving an average accuracy improvement of 12.04% over the existing methods.

With the rapid growth of global maritime traffic, some vessels manipulate their identities to evade regulatory oversight, posing significant challenges to maritime security. Existing anomaly detection methods, however, make limited use of navigational semantics, reducing their effectiveness in complex maritime environments. To address this, we propose MSCE—a Multimodal Semantic Cognition-Enhanced framework for vessel identity anomaly detection. MSCE leverages Large Language Models (LLMs) to analyze semantically enriched multimodal trajectories, capturing critical insights into vessel navigation preferences. Meanwhile, it employs multi-task representation learning to extract personalized spatio-temporal features closely linked to vessel identity. By integrating semantic and spatio-temporal features through a progressive cross-fusion mechanism, MSCE effectively accounts for both global navigational semantic consistency and local spatio-temporal similarity, thereby enhancing anomaly detection in dynamic maritime environments. Experiments on real-world maritime trajectory datasets show that MSCE surpasses existing methods, achieving improvements of 13.29% in accuracy, 10.53% in precision, 5.69% in recall, and 11.79% in F1-score.

Spatio-temporal association analysis has attracted attention in various fields, such as urban computing and crime analysis. The proliferation of positioning technology and location-based services has facilitated the expansion of association analysis across spatio-temporal scales. However, existing methods inadequately consider the scale differences among spatio-temporal objects during analysis, leading to suboptimal precision in association analysis results. To remedy this issue, we propose a multiscale spatio-temporal object representation method, STO2Vec, for association analysis. This method comprises of two parts: graph construction and embedding. For graph construction, we introduce an adaptive hierarchical discretization method to distinguish the varying scales of local features. Then, we merge the embedding method for spatio-temporal objects with that for discrete units, establishing a heterogeneous graph. For embedding, to enhance embedding quality for homogeneous and heterogeneous data, we use biased sampling and unsupervised models to capture the association strengths between spatio-temporal objects. Empirical results using real-world open-source datasets show that STO2Vec outperforms other models, improving accuracy by 16.25% on average across diverse applications. Further case studies indicate STO2Vec effectively detects association relationships between spatio-temporal objects in a range of scenarios and is applicable to tasks such as moving object behavior pattern mining and trajectory semantic annotation.