Long-horizon steel-equipment inspection requires reasoning over heterogeneous records accumulated across repeated inspection cycles. Existing retrieval-augmented generation systems treat historical logs as a static corpus and retain records without estimating their diagnostic value, failing to report early risk. To this end, we propose ConMem, a contribution-aware memory framework for LLM-assisted equipment inspection, supporting a human-in-the-loop early-risk screening system. Specifically, our ConMem first segments inspection logs into functional evidence units, then estimates each memory unit's contribution to downstream diagnosis through a Shapley-style estimation, and finally retains high-value evidence under a constrained memory budget. In experiments, we evaluate ConMem on real-world dataset and ConMem achieves 76.0

We address the challenge of unsupervised coreset selection for few-shot in-context learning (ICL). The goal is to select a small subset of examples under a fixed annotation budget to yield effective prompts for large language models. Existing geometry-based methods often yield coresets that suffer from a skewed distribution, due to the oversampling of peripheral examples and high local redundancy. To address these issues, we propose a process-aware framework for coreset selection. It jointly optimizes the diversity and representativeness of selected samples via a submodular objective. It ensures representativeness by selecting samples based on local density awareness, while promoting diversity by imposing a redundancy penalty relative to the evolving selected set. Thus, it performs progress-aware balancing of representativeness and diversity based on the selection context. Extensive experiments on 7 NLP datasets demonstrate that our method consistently outperforms state-of-the-art coreset selection methods in downstream ICL performance. Further analysis validates that our approach better balances diversity and representativeness throughout the selection process, while retaining the theoretical guarantees of submodular optimization.

Encrypted traffic classification is essential for network management tasks such as quality-of-service controls, identifying malicious traffic, and enhancing cybersecurity. However, the scarcity of plaintext information and the significant reduction of payload characteristics in encrypted traffic present challenges to effective classification. To tackle these issues, we propose a novel time series model called TitNet, which models network traffic at the session level as a multivariate time series and effectively integrates periodic and spatial features inherent in time series data. Our TitNet contains a dynamic frequency selection strategy(DFSS) that facilitates the conversion of time series data into two-dimensional tensor representations, which is pivotal for accurately discerning the intricate patterns embedded in encrypted traffic. This approach enables TitNet to iteratively transform time series into 2D tensors, effectively exploiting the multi-period nesting characteristics of the data to improve classification performance. Experimental results on the ISCXTor2016 dataset (43 Tor/NonTor categories) robustly indicate that our TitNet excels in the detection, classification, and identification of applications within encrypted traffic, achieving 96.21 % accuracy while handling extreme class imbalance. Nonetheless, TitNet introduces additional computational overhead and relies on fixed session truncation, which may limit scalability and long-range modeling. Future work will explore lightweight variants and improved sequence aggregation strategies to address these challenges.

Trimming is an essential step in sequencing data processing. However, many existing trimming tools, such as Trimmomatic and Ktrim, are limited by suboptimal implementations and fail to fully leverage the computational power of modern multi-core platforms. To address this, we introduce RabbitTrim, a highly optimized and versatile trimming tool that fully supports the functionalities of Trimmomatic and Ktrim. RabbitTrim's performance is enhanced through efficient I/O strategies, parallel (de)compression engines, block-based memory pools, bitwise operations, and vectorization techniques. Compared to Trimmomatic, RabbitTrim (in trimmomatic mode) achieves speedups ranging from 1.8x to 6.0x for plain FASTQ files and 3.7x to 14.0x for gzip-compressed FASTQ files on a 48-core Intel server. Similarly, compared to Ktrim, RabbitTrim (in ktrim mode) achieves speedups ranging from 1.5x to 2.5x for plain FASTQ files and 2.7x to 5.6x for gzip-compressed FASTQ files on the same server. Moreover, RabbitTrim is able to process 101 GB gzip-compressed sequencing data in only 5 minutes while Trimmomatic requires at least 21 minutes.

Real-time perception of the geological condition is of great importance to efficient tunneling and hazard prevention in underwater shield tunneling. This study proposes a geological condition-shield machine mutual feedback perception method to address the issues of insufficient utilization of excavation data, lack of optimization of models, low prediction accuracy and efficiency in the current research on geological condition identification in soft soil shield tunneling. For implementation, first, the database of the slurry shield machine tunneling parameters containing 6 input features related to operation parameters were established, in which 269 tunneling cycles from a river-crossing shield tunnel in China were accommodated. Then, the outlier detection method is carried out to pre-process the data sample set and remove the outliers. Furthermore, the genetic algorithm is adapted to optimize the K-means clustering algorithm to cluster the geological conditions category. Four categories with better clustering performance were obtained. To obtain the identification model of the geological condition category with the best prediction performance, 75 % of the sample data is used for data learning, and the optimal training parameters of each model are determined through 10-fold cross-validation. The remaining 25 % of the data is used for validating the four classifiers' performance. The accuracy levels of the proposed models were assessed using four statistical indices, i.e., the Accuracy, F1 score, Precision, and Recall. The testing results revealed that the PSO-ELM algorithm can better characterize and predict the geological conditions in SPB shield tunnelling among all three recognition models. Finally, the synthetic minority oversampling technique (SMOTE) was used to process the database to eliminate the impact of category imbalance on the recognition performance and obtain the best prediction effect. The validation results indicated the four models have improved the overall prediction performance of the minority samples (type-III) by about 0-45 %. Moreover, the Accuracy of (increased by 35-44 %), Recall of (increased by 0-31 %), F1 score of (increased by 25-37 %) and Precision of (increased by 35-44 %), respectively, for testing stages of the PSO-ELM model confirmed that this hybrid model is a powerful and applicable technique addressing problems related to shield tunnelling performance with a high level of accuracy using the proposed investigation flow.

This study examines the protective and risk factors for suicide among nursing home residents as well as strategies to prevent suicide. In this study, semi-structured interviews are used to interpret the experiences and perceptions of nursing home residents regarding suicide attempts. The research is conducted using the technique of content analysis with Nvivo. Researchers have assessed this study for clarity. The experiences of nursing home residents with suicidal ideation were divided into three categories, nine sub-themes, and relevant theme clusters. The three categories were: (a) protective factors for suicide prevention; (b) risk factors for suicide attempts; and (c) suicide prevention strategies. These factors and strategies were found to be associated with suicide ideation and attempts. Factors and strategies influencing suicidal ideation and attempts among nursing home residents revealed some new issues and problems during COVID-19, allowing healthcare providers to better understand nursing home residents and improve intervention strategies in practice.

Large language models (LLMs) have attracted a lot of attention in various fields due to their superior performance, aiming to train hundreds of millions or more parameters on large amounts of text data to understand and generate natural language. As the superior performance of LLMs becomes apparent, they are increasingly being applied to knowledge graph embedding (KGE)-related tasks to improve the processing results. Traditional KGE representation learning methods map entities and relations into a low-dimensional vector space, enabling the triples in the knowledge graph to satisfy a specific scoring function in the vector space. However, based on the powerful language understanding and semantic modeling capabilities of LLMs, which have recently been invoked to varying degrees in different types of KGE-related scenarios such as multi-modal KGE and open KGE according to their task characteristics, researchers are increasingly exploring how to integrate LLMs to enhance knowledge representation, improve generalization to unseen entities or relations, and support reasoning beyond static graph structures. In this paper, we investigate a wide range of approaches for performing LLMs-related tasks in different types of KGE scenarios. To better compare the various approaches, we summarize each KGE scenario in a classification. In the article we also discuss the applications in which the methods are mainly used and suggest several forward-looking directions for the development of this new research area.

Knowledge summarization task aims to summarize the knowledge scattered in multiple documents for answering a query. In this paper, we adopt the concept relation knowledge base ConceptNet for the task and propose the ConceptNet integrated summarization method CNSum, where the concepts are adopted as a bridge to find the latent associations between the query and segments in documents. Besides, a semantic mixture mechanism is introduced to combine the concept-centered associations with the contextual semantics of segments for summarization. To evaluate the knowledge in summary without references, we introduce a Question Answer (QA) based knowledge labeling method to construct training samples. The training samples are used for training a neural evaluation model. We compare CNSum with multiple methods and large language models (LLMs). The results show that CNSum outperforms these baselines. We also evaluate the knowledge in the generated summaries by human evaluation and our neural evaluation. The results show that CNSum is also better than baselines on knowledge completeness. Besides, these two evaluation results are highly correlated.

Current inference systems for Mixture-of-Experts (MoE) models primarily employ static parallelization strategies. However, these static approaches cannot consistently achieve optimal performance across different inference scenarios, as they lack the flexibility to adapt to varying computational requirements. In this work, we propose HAP (Hybrid Adaptive Parallelism), a novel method that dynamically selects hybrid parallel strategies to enhance MoE inference efficiency. The fundamental innovation of HAP lies in hierarchically decomposing MoE architectures into two distinct computational modules: the Attention module and the Expert module, each augmented with a specialized inference latency simulation model. This decomposition promotes the construction of a comprehensive search space for seeking model parallel strategies. By leveraging Integer Linear Programming (ILP), HAP could solve the optimal hybrid parallel configurations to maximize inference efficiency under varying computational constraints. Our experiments demonstrate that HAP consistently determines parallel configurations that achieve comparable or superior performance to the TP strategy prevalent in mainstream inference systems. Compared to the TP-based inference, HAP-based inference achieves speedups of 1.68x, 1.77x, and 1.57x on A100, A6000, and V100 GPU platforms, respectively. Furthermore, HAP showcases remarkable generalization capability, maintaining performance effectiveness across diverse MoE model configurations, including Mixtral and Qwen series models.

As an emerging communication technology, time-sensitive networking (TSN) holds the potential to enable real-time and deterministic interactions for streams within the Industrial Internet of Things. However, effectively and promptly scheduling large-scale streams in the TSN network poses a significant challenge due to high computational complexity. In this article, we conduct a schedulability analysis to preprocess the stream set with given routing paths, avoiding invalid searches and providing optimized guidance for stream routing. To accelerate the feasibility validation of potential solutions, an efficient stream conflict detection approach is proposed leveraging stream grouping with correlation analysis to compress the detection space. Integrating the above preprocess and efficient conflict detection, we develop a scalable scheduling algorithm with an incremental schedule synthesis to enhance scalability while ensuring low slot occupancy for all links. Evaluation results demonstrate that the proposed algorithm significantly reduces synthesis time and achieves low slot occupancy of all links compared to existing scheduling methods.

Knowledge graph embedding (KGE) methods map entities and relations from knowledge graphs to continuous vector spaces, simplifying their representations and enhancing performance across various tasks (e.g., link prediction, question answering). As concerns about personal privacy rise, machine unlearning (MU), an emerging artificial intelligence technology that enables models to eliminate the influence of specific data, has garnered increasing attention from the academic community. The existing KGE unlearning works mainly achieve MU through data obfuscation and adjustments to the model’s training loss. Furthermore, existing approaches lack generalization ability across different unlearning tasks. In this paper, we propose a Meta-Learning-Based Knowledge Graph Embedding Unlearning framework (MetaEU), specifically designed for KGE unlearning. With the help of meta-learning, the model can discover the inherent relationships between different unlearning tasks, thereby avoiding the need to start from scratch for each unlearning task and achieving better generalization across various task scenarios. A thorough experimental study on benchmark datasets shows that MetaEU demonstrates promising performance in the knowledge graph embedding unlearning task.

Knowledge-graph-based recommendation aims to provide personalized recommendation services to users based on their historical interaction information, which is of great significance for shopping transaction rates and other aspects. With the rapid growth of online shopping, the knowledge graph constructed from users’ historical interaction data now incorporates multiattribute information, including timestamps, images, and textual content. The information of multiple modalities is difficult to effectively utilize due to their different representation structures and spaces. The existing methods attempt to utilize the above information through simple embedding representation and aggregation, but ignore targeted representation learning for information with different attributes and learning effective weights for aggregation. In addition, existing methods are not sufficient for effectively modeling temporal information. In this article, we propose MTR, a knowledge graph recommendation framework based on mixture of experts network. To achieve this goal, we use a mixture-of-experts network to learn targeted representations and weights of different product attributes for effective modeling and utilization. In addition, we effectively model the temporal information during the user shopping process. A thorough experimental study on popular benchmarks validates that MTR can achieve competitive results.

Trimmomatic is a de-facto standard trimmer for Illumina sequencing data. However, limited by its sub-optimal implementation, it cannot fully exploit the computational power of common multi-core platforms. Therefore, we propose RabbitTrim, a highly optimized implementation of Trimmomatic based on efficient I/O strategies, parallel (de)compression engines, block-based memory pools, bitwise operations and vectorization techniques. RabbitTrim achieves speedups between 1.5x and 3.3x (3.7x and 8.0x) when processing plain (gzip-compressed) FASTQ files on a 48-core Intel server. Overall, RabbitTrim is able to process 101 GB gzip-compressed sequencing data in only 5 min while Trimmomatic requires at least 21 min. The source code is available at https://github.com/RabbitBio/RabbitTrim.

Diverse opinion summarization aims to generate a summary that captures multiple opinions in texts. Although large language models (LLMs) have become the main choice for this task, the performance is highly depend on prompts. In this paper, we propose a self-evaluation based prompt calibration framework to stimulate LLM for generating high quality summary. It adopts the reinforcement learning mechanism to calibrate prompts for maximizing the reward of summary. The framework contains three parts. In the prompt construction part, we design the prompt that contains topic, task instruction and key opinion reference. The topic indicates the main focus of documents, the instruction describes the task with natural language and the key opinion reference is the explicit constraint on the expected opinions. In the reward part, for each summary, its coverage score and diversity score are used to represent the semantic coverage to the source documents and the inter opinion differences, respectively. The prompt calibration part selects the sentences in generated summaries to calibrate the prompts for the next iteration. With this framework, we use a LLM with 7B parameters to generate summaries, which outperforms large GPT-4 and multiple strong baselines. The ablation studies indicate the effectiveness of the iterative calibration process. We analyze the opinion difference in terms of the tendencies of sentences in summaries and use the Natural Language Inference (NLI)-based method to evaluate the faithfulness of summaries. Experiment results show that our method generates summaries with high opinion difference and faithfulness.

Shield tunneling processes can result in accidents like water inrush and collapse due to instability in the tunnel face. This paper conducts a systematic investigation of the deformation features and failure mechanism of subsea shield tunnels with different burial depths crossing fault zone. The research entails combing theoretical analysis, numerical simulation, and data mining with the second subsea tunnel in Jiaozhou Bay, Qingdao. The impact of the shallow and deep burial depths on the damage modes of the shield tunnel face is analyzed from the perspective of the pressure arch effect. When the tunnel burial depth exceeds the limit, the overlying rock of the tunnel will form a pressure arch that supports the overlying rock. The study investigates the impact of various factors on the tunnel face's failure modes using a fluid-structure interaction approach. Among them, as the thickness of the overlying bearing stratum increased, the displacement of the tunnel face initially decreased but later increased. The increase of cohesion and internal friction angle contributed to decreasing the displacement of the tunnel face, while seawater depth and tunnel diameter exhibited diverse effects. In the grey correlation analysis, the seawater depth exhibited the most sensitive impact on the pressure arch height.

The construction of large open knowledge bases (OKBs) is integral to many knowledge-driven applications on the world wide web such as web search. However, noun phrases in OKBs often suffer from redundancy and ambiguity, which calls for the investigation on OKB canonicalization. Current solutions address OKB canonicalization by devising advanced clustering algorithms and using knowledge graph embedding (KGE) to further facilitate the canonicalization process. Nevertheless, these works fail to fully exploit the synergy between clustering and KGE learning, and the methods designed for these sub-tasks are sub-optimal. To this end, we put forward a multi-task learning framework, namely MulCanon, to tackle OKB canonicalization. Specifically, diffusion model is used in the soft clustering process to improve the noun phrase representations with neighboring information, which can lead to more accurate representations. MulCanon unifies the learning objective of diffusion model, KGE model, side information and cluster assignment, and adopts a two-stage multi-task learning paradigm for training. A thorough experimental study on popular OKB canonicalization benchmarks validates that MulCanon can achieve competitive canonicalization results.

As a kind of protective engineering, breakwater plays a very important role in the safety of the protected objects, and scholars from all over the world have done a lot of research on the characteristics of breakwater. Reviews are a quick way to see how research is progressing. In addition to the traditional breakwaters for the purpose of wave suppression, there are many breakwaters with power generation functions. In order to supplement and enrich the research of breakwater, adapt to the development of The Times, the breakwater is reviewed. In this paper, breakwaters are divided into traditional breakwaters and wave energy converter(WEC) breakwaters according to their functions. The traditional breakwaters for wave absorbing, while the WEC breakwater is breakwater added wave energy converter, which can convert wave energy into electrical energy, so WEC breakwater has two functions of wave dissipation and power generation. According to the fixed type, the traditional breakwater is divided into fixed breakwater and floating breakwater. In addition to adding WEC breakwaters in the classification, this paper takes the transmission coefficient kt =0.5 as the reference line and compares the optimal transmission coefficient of the corresponding breakwater structure with the reference line. Finally, the structure of the breakwater is summarized and suggested. Hope to play a guiding role in scholars' research and engineering construction.

In this paper an improvement of the response surface method is proposed. An iterative strategy is used to determine a response surface that is able to fit the limit state function in the neighborhood of the design point. The locations of the sample points used to evaluate the free parameters of the response surface are chosen according to the importance sensitivity of each random variable. Several analytical and structural examples are considered to demonstrate the advantages of the proposed improvement.

Fixed-point decimal operations in databases with arbitrary-precision arithmetic refer to the ability to store and operate decimal fraction numbers with an arbitrary length of digits. This type of operation has become a requirement for many applications, including scientific databases, financial data processing, geometric data processing, and cryptography. However, the state-of-the-art fixed-point decimal technology either provides high performance for low-precision operations or supports arbitrary-precision arithmetic operations at low performance. In this paper, we present a design and implementation of a framework called UltraPrecise which supports arbitrary-precision arithmetic for databases on GPU, aiming to gain high performance for arbitrary-precision arithmetic operations. We build our framework based on the just-in-time compilation technique and optimize its performance via data representation design, PTX acceleration, and expression scheduling. UltraPrecise achieves comparable performance to other high-performance databases for low-precision arithmetic operations. For high-precision, we show that UltraPrecise consistently outperforms existing databases by two orders of magnitude, including workloads of RSA encryption and trigonometric function approximation.