Accurate interpretation of Notices To Airmen (NOTAMs) is critical for aviation safety, yet their condensed and cryptic language poses significant challenges to both manual and automated processing. Existing automated systems are typically limited to "Shallow Parsing," failing to extract the actionable intelligence needed for operational decisions. We formalize the complete interpretation task as "Deep Parsing," a dual-reasoning challenge requiring both dynamic knowledge grounding (linking the NOTAM to evolving real-world aeronautical data) and schema-based inference (applying static domain rules to deduce operational status). To tackle this challenge, we propose NOTAM-Evolve, a self-evolving framework that enables a Large Language Model (LLM) to autonomously master complex NOTAM interpretation. Leveraging a knowledge graph-enhanced retrieval module for data grounding, the framework introduces a crucial closed-loop learning process where the LLM progressively improves from its own outputs, minimizing the need for extensive human-annotated reasoning traces. In conjunction with this framework, we introduce a new benchmark dataset of 10,000 expert-annotated NOTAMs. Our experiments demonstrate that NOTAM-Evolve achieves a 30.4% absolute accuracy improvement over the base LLM, establishing a new state-of-the-art on the task of structured NOTAM interpretation.

Multimedia recommendation systems leverage user–item interactions and multimodal information to capture user preferences, enabling more accurate and personalized recommendations. Despite notable advancements, existing approaches still face two critical limitations: first, shallow modality fusion often relies on simple concatenation, failing to exploit rich synergic intra- and inter-modal relationships; second, asymmetric feature treatment—where users are only characterized by interaction IDs while items benefit from rich multimodal content—hinders the learning of a shared semantic space. To address these issues, we propose a C ross-Modal R ecursive A ttention N etwork with Dual Graph E mbedding (CRANE) . To tackle shallow fusion, we design a core Recursive Cross-Modal Attention (RCA) mechanism that iteratively refines modality features based on cross-correlations in a joint latent space, effectively capturing high-order intra- and inter-modal dependencies. For symmetric multimodal learning, we explicitly construct users’ multimodal profiles by aggregating features of their interacted items. Furthermore, CRANE integrates a symmetric dual graph framework—comprising a heterogeneous user–item interaction graph and a homogeneous item–item semantic graph—unified by a self-supervised contrastive learning objective to fuse behavioral and semantic signals. Despite these complex modeling capabilities, CRANE maintains high computational efficiency. Theoretical and empirical analyses confirm its scalability and high practical efficiency, achieving faster convergence on small datasets and superior performance ceilings on large-scale ones. Comprehensive experiments on four public real-world datasets validate an average 5% improvement in key metrics over state-of-the-art baselines. Our code is publicly available at https://github.com/MKC-Lab/CRANE .

Understanding surface congestion is essential for improving taxiing efficiency and reducing carbon emissions in busy airports. Diverging from the traditional input-output and dynamics analysis, we propose a novel end-to-end framework to study the complete dynamic process of airport surface congestion and recovery. This framework employs a stochastic hybrid system to model multi-timescale dynamics, integrating continuous states and discrete modes of surface operations under uncertainty. First, the probabilistic reachable set for the input-output state is computed via a chance-constrained optimization program to represent the relationship between the number of aircraft taxiing out and the departure throughput. Next, the discrete modes are divided based on taxiing efficiency and traffic load, utilizing tailored congestion contour regression and density-based clustering, respectively. Finally, the transition trajectory incorporating mode information is constructed to depict the complete process from congestion formation to subsequent recovery, followed by employing an unsupervised algorithm to identify representative patterns. The proposed framework is verified using two years of real-world datasets from Chengdu Shuangliu International Airport, China. Experimental results demonstrate the superiority of our approach compared with the baselines. Moreover, this work also reveals some intriguing findings, such as the diverse multi-timescale dynamical phenomena, and their implications for practical airport surface operations.

Due to the significance of air traffic delay propagation in the operational robustness of air transport systems, it has recently received increased attention from both industry and academia. The purpose of delay propagation analysis is to capture the dynamic traveling trajectories of root delays in air traffic networks. Few studies have investigated delay propagation while taking consideration of temporal context information, leading to incomplete traveling trajectories of root delays. In this paper, a novel systematic framework based on temporal causal inference is proposed to model delay propagation in airport networks. More specifically, considering the lagged nature caused by the transition time in delay propagation, a fully-connected delay propagation network based on transfer entropy is developed. Additionally, a causal biased random walk is embedded to explore the temporal cascading effect of root delays in air transport systems and to generate delay propagation trees for each airport. Extensive experiments on a real-world dataset indicate that our framework is able to effectively model the dynamic evolution of air traffic delays in an airport network with acceptable performance. This paper also presents a real-world case study of the Chinese airport network, which reveals that the traveling trajectories of air traffic delays differ significantly at various daily delay levels and airport throughput.

With the rapid rise of massive multi-modal Internet data, air crisis event recognition plays a significant role in accident information management. A key feature of this work is the open-world setting, where the new events can be detected via the model trained from the known events. A scenario-guided Transformer-enabled multi-modal unknown air crisis event classification is proposed. Firstly, we introduce a memory-augmented feature representation module to improve the cross-modal feature fusion in the basic Transformer network, where the textual and image features of the air crisis events are extracted through the pre-trained models. Then, a scenario-guided mechanism is proposed to pivot the pseudo event simulation to approximate the distribution of unknown events effectively. Specifically, an end-to-end scenario attention with a cross-event mask module is presented to select the valid samples of pseudo events and filter out the invalid ones. Finally, the unknown event classifier is designed to classify the known classes and recognize the unknown events simultaneously. Moreover, a specialized multi-modal dataset on air crisis events is proposed as a benchmark, named AirCrisisMMD. The extensive experiments are performed on the AirCrisisMMD and CrisisMMD datasets, where the latter is the publicly available multi-modal crisis event dataset. The results verify the superior performance boost of the proposed method compared with the highly related state-of-the-art baselines.

Online knowledge libraries refer to the online data warehouses that systematically organize and categorize the knowledge-based information about different kinds of concepts and entities. In the era of big data, the setup of online knowledge libraries is an extremely challenging and laborious task, in terms of efforts, time and expense required in the completion of knowledge entities. Especially nowadays, a large number of new knowledge entities, like movies, are keeping on being produced and coming out at a continuously accelerating speed, which renders the knowledge library setup and completion problem more difficult to resolve manually. In this paper, we will take the online movie knowledge libraries as an example, and study the "Multiple aligned ISomeric Online Knowledge LIbraries Completion problem" (MISO-KLIC) problem across multiple online knowledge libraries. Miso-Klic aims at identifying the missing entities for multiple knowledge libraries synergistically and ranking them for editing based on certain ranking criteria. To solve the problem, a thorough investigation of two isomeric online knowledge libraries, Douban and IMDB, have been carried out in this paper. Based on analyses results, a novel deep online knowledge library completion framework "Integrated Deep alignEd Auto-encoder" (IDEA) is introduced to solve the problem. By projecting the entities from multiple isomeric knowledge libraries to a shared feature space, IDEA solves the Miso-Klic problem via three steps: (1) entity feature space unification via embedding, (2) knowledge library fusion based missing entity identification, and (3) missing entity ranking. Extensive experiments done on the real-world online knowledge library dataset have demonstrated the effectiveness of IDEA in addressing the problem.