Contract clause retrieval is foundational to contract drafting because lawyers rarely draft contracts from scratch; instead, they locate and revise the most relevant precedent clauses. We introduce the Atticus Clause Retrieval Dataset (ACORD), the first expert-annotated benchmark specifically designed for contract clause retrieval to support contract drafting tasks. ACORD focuses on complex contract clauses such as Limitation of Liability, Indemnification, Change of Control, and Most Favored Nation. It includes 114 queries and over 126,000 query-clause pairs, each ranked on a scale from 1 to 5 stars. The task is to find the most relevant precedent clauses to a query. The bi-encoder retriever paired with pointwise LLMs re-rankers shows promising results. However, substantial improvements are still needed to manage the complex legal work typically undertaken by lawyers effectively. As the first expert-annotated benchmark for contract clause retrieval, ACORD can serve as a valuable IR benchmark for the NLP community.

Urinary tract infections (UTIs) are a significant health concern, particularly for people living with dementia (PLWD), as they can lead to severe complications if not detected and treated early. This study builds on previous work that utilised machine learning (ML) to detect UTIs in PLWD by analysing in-home activity and physiological data collected through low-cost, passive sensors. The current research focuses on improving the performance of previous models, particularly by refining the Multilayer Perceptron (MLP), to better handle variations in home environments and improve sex fairness in predictions by making use of concepts from multitask learning. This study implemented three primary model designs: feature clustering, loss-dependent clustering, and participant ID embedding which were compared against a baseline MLP model. The results demonstrated that the loss-dependent MLP achieved the most significant improvements, increasing validation precision from 48.92 % to 72.60 % and sensitivity from 27.44 % to 70.52 %, while also enhancing model fairness across sexes. These findings suggest that the refined models offer a more reliable and equitable approach to early UTI detection in PLWD, addressing participant-specific data variations and enabling clinicians to detect and screen for UTI risks more effectively, thereby facilitating earlier and more accurate treatment decisions. The code used in this study is available at the following link: https://github.com/Kexin-Fan/Multi-Source-Analysing.git.

With the advantage of robust resilience, large capacity, and strong adaptability, space-air-ground integrated network (SAGIN) can simultaneously support various task scenarios involving heterogeneous networks and diverse task demands. In this integrated network, the constraints from limited resources and dynamic environment pose challenges in fulfilling concurrent demands, and improper task scheduling strategy can lead to network resource wastage and task dissatisfaction. In this paper, we propose an adaptive solution for multiscenario joint scheduling in SAGIN. We construct a comprehensive task scheduling frame-work and propose the task relevance matrix for in-depth analysis. To achieve the goal of improving network resource utilization and task satisfaction, we formulate the joint optimization problem as a cooperative Markov game and propose a novel multi-scenario task scheduling algorithm based on heterogeneous-agent proximal policy optimization (HAPPO). Simulation results show that the proposed algorithm can achieve better performance by effectively improving resource utilization and reducing task delay, compared with two state-of-the-art multi-agent reinforcement learning algorithms and the random baseline.

Mega-constellation networks (MCNs), comprising an expansive array of orbits and satellites, will be a pivotal component of the prospective nonterrestrial network (NTN). Laser intersatellite links (LISLs) represent a promising technology for the establishment of satellite networks, offering high capacity and highly reliable communication links. Nevertheless, LISL still has some technical challenges, such as link establishment instability, satellite payload capacity, and topology design. For a considerable number of satellites and laser communication constraints, LISLs have elevated the complexity and difficulty of routing and topology construction. In this article, we focus on the LISL connecting stability and propose a novel method to evaluate the intersatellite link (ISL) selection based on the acquisition probability of the laser terminal system. Subsequently, a nonlinear optimization model is formulated for the laser link selection problem, where the terminal acquisition probability is maximized. Finally, an MCN topology design algorithm (MTDA) is proposed to establish greater stability and higher channel gain within the different access distances of the laser terminal. Using the satellite constellation for Phase I of Starlink, three different laser MCN topologies were constructed by MTDA, adapting to the maximum access distance of the laser terminal within 5000 km. The impact of these differing topologies on network latency and hop count was then analyzed in a full satellite period. Compared with the baseline, MTDA has a 15.7% latency advantage and 14.0% hop advantage under the same access distance of laser terminal. The numerical results show that the proposed algorithm has a positive impact on the average network latency, hop count and their respective variances.

Space-air-ground integrated network (SAGIN) can support a wide variety of task scenarios as a next-generation network. In this dynamic network, tasks with different characteristics and demands are usually simultaneously generated, which requires an efficient task scheduling method to meet diverse task demands, reduce system cost, and improve resource utilization. In this paper, we propose a scalable task scheduling and resource allocation solution suitable for various task scenarios in SAGIN. Specifically, we construct a generic task scheduling framework that considers diverse task types and demands in SAGIN. With the objective of minimizing the total system cost, we propose a novel joint task scheduling and resource allocation algorithm based on deep reinforcement learning to obtain the optimal solution. Extensive simulations are conducted in a civil aviation scenario to evaluate the performance of the proposed algorithm. Numerical results demonstrate that our proposed algorithm can efficiently reduce system cost and task drop rate compared with three benchmark schemes, and the improvement of resource allocation to task scheduling is verified through analysis.