Graded reading tailors text difficulty to learners’ proficiency by producing multiple versions of the same content—an approach long embraced in language education but still dependent on labor-intensive, expert-driven adaptation. In this paper, we introduce the task of C hinese G raded D ocument S implification (CGDS) for non-native learners, which seeks to automate the creation of multi-level reading materials in accordance with established proficiency standards. Guided by the three stages of the Hanyu Shuiping Kaoshi (HSK) 3.0 framework (Levels 1–3 for Advanced, Levels 4–6 for Intermediate, and Levels 7–9 for Beginner learners), we propose Large Language Model for Chinese Graded Document Simplification (LLM4CGDS), a rule-guided, large language model (LLM)-based framework that integrates HSK-level readability constraints and external knowledge retrieval to control document-level simplification without requiring supervised fine-tuning. To foster further research, we construct two complementary datasets: J ourney to the W est D ocument S implification (JWDS) and M ulti- D omain D ocument S implification (MDDS) that covering diverse genres and difficulty levels. Experimental evaluation on two datasets demonstrates that LLM4CGDS substantially outperforms direct prompting of state-of-the-art LLMs in both readability control and meaning preservation.

Monocular endoscope-based reconstruction of dynamic 3D surgical fields is beneficial for both intraoperative manual/robotic manipulation and post-operative surgical skills training. However, the natural characteristics of tissue deformations with instruments and blood obscuration bring great challenges to 3D scene awareness, especially under sparse viewpoints limited by laparoscopic movements. In this work, we propose Endo-4SRF, an effective Neural Radiance Field (NeRF)-based method that can reconstruct deformable tissues with instruments stealth by solely relying on monocular endoscopic image flows from a single viewpoint. Specifically, to enhance the 3D reconstruction accuracy under the deficiency of depth ground truth, we devised a dynamic Gaussian-based neural sampling strategy, leveraging the depth inherently obtained from NeRF and the conjunction information inferred by a prior learning-based depth estimation network. Besides, we integrated the Signed Distance Function (SDF) and resolved its singularity problem by furnishing additional geometric constraints for the neural radiance field, thereby achieving precise reconstruction of dynamic scenes devoid of depth ground truth supervision. Furthermore, adopting spherical harmonic functions for color fitting has significantly improved our model's computational efficiency and rendering quality. We extensively performed cross validation experiments to verify the performance using public and in-house datasets. Our quantitative and qualitative results demonstrate remarkable superiority over the state-of-the-art (SOTA) approaches concerning depth prediction accuracy, image rendering quality, model training efficiency, and 3D reconstruction outcomes.

Federated Learning (FL) is an emerging decentralized machine learning paradigm that addresses the data-silo problem through privacy-preserving collaborative model training, attracting significant attention from academia and industry. However, model trading in FL involves multiple stakeholders, including data owners, model requesters, and the cloud service platform, whose conflicting interests hinder the sustainability and stability of FL. To address these challenges, this paper considers the bounded rationality of the three parties involved in long-term dynamic decision-making and constructs a tripartite evolutionary game model based on evolutionary game theory, further taking into account collusion between data owners and cloud service platforms. We analyze the evolutionary dynamics involved, theoretically revealing the social dilemma of dishonesty in FL model trading. To prevent dishonest behaviors such as free-riding and false reporting, we apply the replicator dynamics and Lyapunov method to analyze the impact of rewards, punishments, and collusion costs on the evolutionary stable strategies of the three parties and propose incentive strategies. Simulation experimental results validate that our incentive model is effective in alleviating dishonest social dilemmas and improving social welfare.

Process discovery techniques analyze process logs to extract models that characterize the behavior of business processes. In real-life logs, however, noises exist and adversely affect the extraction and thus decrease the understandability of discovered models. In this paper, we propose a novel double granularity filtering method, executed on both the event and trace levels, to detect noises by analyzing the directly-following and parallel relations between events. Based on the probability of an event occurring in a sequence, the infrequent behaviors and redundant events in the logs can be filtered out. In addition, the missing events in parallel blocks are detected to further improve the performance of filtering. Experiments on synthetic logs and five real-life datasets demonstrate that our method significantly outperforms other state-of-the-art methods.

Remaining time predictive monitoring of business processes (BPs) is a key research issue in business process mining, which provides timely predictive information for stakeholders to take proactive corrective actions to reduce process execution risk such as exceeding time limit or to adjust the priority of activities. However, current researches on remaining time prediction only consider the impact of internal attributes of single process instance, but ignore the resource competition among multiple instances executed together. Therefore, this paper takes resource competition into consideration and characterizes several inter-instance attributes as the input of prediction. We also prioritize and select some key activities that strongly impact the execution time of BPs according to historical event logs and include them as input of the prediction. Meanwhile, in order to solve the instability of one single prediction model in complex scenarios, a multilayer hybrid model constructed from XGBoost and LightGBM models using stacking technique is proposed. Experiments on four real-life datasets show that our approach of considering attributes among instances and including key activities into a hybrid model outperforms other prediction methods.

Mining high-quality process models from real-life event logs remains a huge challenge owing to the presence of noises in the logs. Noises lead to unnecessary structures in process models, which further impede their precision and fitness. This paper proposed a novel double granularity method of filtering out noise logs based on event dependency. In this method, local dependency and global dependency of events in recorded logs are coupled as a mixed dependency to conduct fine-grained filtering of noise events based on statistical analysis. Furthermore, a coarse-grained filtering strategy is added to the method for removing specific noise traces that cannot be detected by fine-grained filtering. Comparison experiments on one synthetic and five real-life datasets with two other filtering algorithms verified the superior performance and high precision of the method.