Yinqing Wang , Xiangchun Li , Chunli Yang , Tao Yang , Fengchun Lan , Xin Tian
{"title":"A dynamic Bayesian network risk assessment model for coal-fired power plants based on grey correlation and triangular fuzzy theory","authors":"Yinqing Wang , Xiangchun Li , Chunli Yang , Tao Yang , Fengchun Lan , Xin Tian","doi":"10.1016/j.csi.2025.104001","DOIUrl":"10.1016/j.csi.2025.104001","url":null,"abstract":"<div><div>This article adopts analysis methods such as Bayesian network model, grey correlation analysis, analytic hierarchy process, and three-level fuzzy method to establish a risk assessment index system and dynamic risk assessment model for coal-fired power plants. The results show that starting from the perspective of \"human machine environment management\",relatively complete risk assessment index system for coal-fired power plants has been determined using the grey correlation analysis method, with a total of 4 first level indicators, 15 second level indicators, and 63 third level indicators; Using Analytic Hierarchy Process to determine the weights of indicators at all levels, the calculation results are incorporated into subsequent Bayesian network models, transforming the Bayesian model from a static model to a dynamic model; Then bring the calculated results into NETICA software for calculation, obtain the original Bayesian network model, and perform reverse inference analysis and sensitivity analysis; Through comparative analysis, B1 (management personnel), B8 (equipment and facility management), B10 (work environment hazards), B14 (dual prevention mechanism management), X4 (personnel \"three violations\" situation), X20 (equipment change management), and X43 (vibration hazards) have significant changes. Therefore, management should be strengthened in practical operations, providing a theoretical basis for the actual production management of power plants.</div></div>","PeriodicalId":50635,"journal":{"name":"Computer Standards & Interfaces","volume":"94 ","pages":"Article 104001"},"PeriodicalIF":4.1,"publicationDate":"2025-03-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143684748","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Jie Ou, Yueming Chen, Buyao Xiong, Zhaokun Wang, Wenhong Tian
{"title":"Accelerating Mixture-of-Experts language model inference via plug-and-play lookahead gate on a single GPU","authors":"Jie Ou, Yueming Chen, Buyao Xiong, Zhaokun Wang, Wenhong Tian","doi":"10.1016/j.csi.2025.103996","DOIUrl":"10.1016/j.csi.2025.103996","url":null,"abstract":"<div><div>The widespread adoption of large language models (LLMs) has encouraged researchers to explore strategies for running these models more efficiently, such as the mixture of experts (MoE) method, which aims to increase the knowledge capacity of the model without substantially increasing its computational costs, as only a fraction of the model components are active for each token. However, this approach also increases the size of the model, which makes it challenging to run these models even on high-end GPUs. Quantization and offloading strategies have been used to enable the execution of MoE in resource-constrained environments, however, the time overhead introduced by offloading remains a bottleneck. In this paper, we propose a plug-and-play lookahead gate that predicts in advance the experts to be used in the next few layers. Furthermore, to mitigate the misalignment problem arising from cross-layer prediction, we introduce an alignment training method, layer-wise gate alignment, enhancing the prediction hit rate while maintaining low resource requirements. Moreover, we present a speculative expert scheduling strategy to accelerate the end-to-end inference process of MoE models. To validate our approach, we established an inference framework for quantized MoE and conducted extensive experiments. The results demonstrate the effectiveness of our proposed methods, with throughput improvements of 57.72%, 60.00%, and 62.26% under 4, 3, and 2-bit quantization conditions for experts, respectively, compared with the Mixtral-offloading method.</div></div>","PeriodicalId":50635,"journal":{"name":"Computer Standards & Interfaces","volume":"94 ","pages":"Article 103996"},"PeriodicalIF":4.1,"publicationDate":"2025-03-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143611213","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Verifiable and auditable multi-authority attribute-based encryption","authors":"Xingwang Wang, Peng Zeng, Jiaying Luo","doi":"10.1016/j.csi.2025.103991","DOIUrl":"10.1016/j.csi.2025.103991","url":null,"abstract":"<div><div>Attribute-based encryption (ABE) is crucial for cloud computing security as it enables fine-grained non-interactive access control and allows efficient and secure data sharing without relying on trusted third parties. Multi-authority ABE (MABE) is an improvement to traditional single-authority ABE because the former can remove the key escrow problem existed in the latter and thus is more appropriate for access control of some special applications. Recently, Zhou et al. proposed an auditable MABE (A-MABE) scheme in which the authorities are categorized into a trusted authority and several attribute authorities. For the confirmation and protection of data ownership, we propose verifiable A-MABE (VA-MABE) schemes in this paper. The additional feature of verifiability is achieved by our newly proposed zero-knowledge proof protocol which enables a data sender (i.e. data owner) to prove the ownership of the encrypted data without leaking sensitive information. Our proposed VA-MABE schemes also afford authorities the flexibility to join or exit the system with a minimal cost. Addressing concerns about centralized audit power, we incorporate technologies that decentralize this role, thereby enhancing the system’s resilience.</div></div>","PeriodicalId":50635,"journal":{"name":"Computer Standards & Interfaces","volume":"94 ","pages":"Article 103991"},"PeriodicalIF":4.1,"publicationDate":"2025-03-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143592208","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Application of retrieval-augmented generation for interactive industrial knowledge management via a large language model","authors":"Lun-Chi Chen , Mayuresh Sunil Pardeshi , Yi-Xiang Liao , Kai-Chih Pai","doi":"10.1016/j.csi.2025.103995","DOIUrl":"10.1016/j.csi.2025.103995","url":null,"abstract":"<div><div>Industrial data processing and retrieval are necessary for adoption in Industry 5.0. Large Language Model (LLMs) revolutionize natural language process (NLP) but face challenges in domain-specific applications due to specialized terminology and context. Artificial Intelligence (AI) assistants for industrial-related work enquiry and customer support services are necessary for increasing demand and quality of service (QoS). Our research aims to design a novel customized model with a retrieval-augmented generation (RAG)-based LLM as a sustainable solution for industrial integration with AI. The goal is to provide an interactive industrial knowledge management (IIKM) system that can be applied to technical services: assisting technicians in the search for precise technical repair details and company internal regulation searches: personnel can easily inquire about regulations, such as business trips and leave requirements. The IIKM model architecture consists of BM25 and embedding sequence processing in the chroma database, where the top k-chunks are selected by the BAAI ranker to respond effectively to the queries. A group of documents of 234 MB size and pdf, pptx, docx, csv and txt formats are used for the experimental analysis. The designed interactive knowledge management system has a mean reciprocal rank (MRR) of 88 %, a recall of 85 % and a mean average precision (mAP) of 75 % in technical service. The internal regulatory documents have a generation-based retrieval evaluation prediction of recall of 91.62 %, MRR of 97.97 % and mAP of 91.12 %. We conclude with insights gained and experiences shared from IIKM deployment with Sakura incorporation, highlighting the importance of the hybrid approach integrating RAG-based generative pretrained transformer (GPT) models for customized solutions.</div></div>","PeriodicalId":50635,"journal":{"name":"Computer Standards & Interfaces","volume":"94 ","pages":"Article 103995"},"PeriodicalIF":4.1,"publicationDate":"2025-03-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143549225","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Ronghao Pan, José Antonio García-Díaz, Rafael Valencia-García
{"title":"Spanish MTLHateCorpus 2023: Multi-task learning for hate speech detection to identify speech type, target, target group and intensity","authors":"Ronghao Pan, José Antonio García-Díaz, Rafael Valencia-García","doi":"10.1016/j.csi.2025.103990","DOIUrl":"10.1016/j.csi.2025.103990","url":null,"abstract":"<div><div>The rise of digital communication has exacerbated the challenge of tackling harmful speech online, particularly hate speech, which dehumanises individuals or groups on the basis of traits such as race, gender or ethnicity. This study highlights the urgent need for fine-grained detection methods that take into account several subtasks of hate speech detection, including its intensity, determining the groups to which hate speech is directed, and whether the target is an individual or a group. Furthermore, there is a gap in comprehensive Spanish language corpora that cover these subtasks of hate speech detection. Therefore, we created a novel corpus entitled Spanish MTLHateCorpus 2023 to facilitate the analysis of hate speech in these subtasks and evaluated the effectiveness of the multi-task learning strategy evaluating mBART and T5, comparing its results with other Large Language Models using Zero-Shot Learning as a lower bound and an ensemble based on the mode of several Fine-Tuning as an upper bound. The results achieved by the Multi-Task Learning strategy demonstrated its potential to increase model versatility, allowing a single model to effectively tackle multiple tasks while achieving competitive results, particularly in target group recognition. However, the ensemble learning slightly outperforms the Multi-Task Learning strategy.</div></div>","PeriodicalId":50635,"journal":{"name":"Computer Standards & Interfaces","volume":"94 ","pages":"Article 103990"},"PeriodicalIF":4.1,"publicationDate":"2025-03-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143563566","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Huali Ren , Anli Yan , Lang Li , Zhenxin Zhang , Ning Li , Chong-zhi Gao
{"title":"Are you copying my prompt? Protecting the copyright of vision prompt for VPaaS via watermarking","authors":"Huali Ren , Anli Yan , Lang Li , Zhenxin Zhang , Ning Li , Chong-zhi Gao","doi":"10.1016/j.csi.2025.103992","DOIUrl":"10.1016/j.csi.2025.103992","url":null,"abstract":"<div><div>Visual Prompt Learning (VPL) reduces resource consumption by avoiding updates to pre-trained model parameters and instead learns input perturbations, a visual prompts, added to downstream task data for predictions. Designing high-quality prompts requires significant expertise and time-consuming optimization, leading to the emergence of Visual Prompts as a Service (VPaaS), where developers monetize well-crafted prompts by providing them to authorized customers.However, in cloud computing environments, prompts can be easily copied and redistributed, posing serious risks to the intellectual property (IP) of VPaaS developers.</div><div>To address this, we propose WVPrompt, the first method for protecting visual prompts via watermarking in a black-box setting. WVPrompt consists of two components: prompt watermarking and prompt verification. Specifically, it utilizes a poison-only backdoor attack method to embed a watermark into the prompt, and then employs a hypothesis-testing approach for remote verification of prompt ownership. Extensive experiments were conducted on three well-known benchmark datasets and three popular pre-trained models: RN50, BiT-M, and Instagram. The experimental results demonstrate that WVPrompt is efficient, harmless, and robust to various adversarial operations, making it a reliable solution for securing visual prompts in cloud-based applications.</div></div>","PeriodicalId":50635,"journal":{"name":"Computer Standards & Interfaces","volume":"94 ","pages":"Article 103992"},"PeriodicalIF":4.1,"publicationDate":"2025-03-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143563567","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Can LLMs revolutionize text mining in chemistry? A comparative study with domain-specific tools","authors":"Madhavi Kumari , Rohit Chauhan , Prabha Garg","doi":"10.1016/j.csi.2025.103997","DOIUrl":"10.1016/j.csi.2025.103997","url":null,"abstract":"<div><div>The exponential growth of chemical literature necessitates advanced tools for efficient data extraction and utilization. This study investigates the performance of Large Language Models (LLMs) in Chemical Named Entity Recognition (CNER), comparing them against traditional domain-specific tools. We fine-tuned the LLaMA-2 model using the NLM-Chem corpus and integrated a Retrieval-Augmented Generation (RAG) pipeline to enhance performance. The results revealed that fine-tuned LLaMA-2 models, particularly those incorporating RAG, achieved an F1 score of 0.82, surpassing the score of traditional CNER tools. Furthermore, LLMs demonstrated superior generalizability across different datasets. The study also explores the dependency of LLMs size for CNER tasks. A practical case study highlighting the application of these models in chemical entity extraction from pharmaceutical literature, achieving high accuracy in identifying drug and their interactions. These findings establish LLMs as a robust and adaptable alternative to traditional CNER tools, paving the way for transformative applications in chemoinformatics.</div></div>","PeriodicalId":50635,"journal":{"name":"Computer Standards & Interfaces","volume":"94 ","pages":"Article 103997"},"PeriodicalIF":4.1,"publicationDate":"2025-03-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143549224","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pei-Gen Ye , Zhengdao Li , Zuopeng Yang , Pengyu Chen , Zhenxin Zhang , Ning Li , Jun Zheng
{"title":"Periodic watermarking for copyright protection of large language models in cloud computing security","authors":"Pei-Gen Ye , Zhengdao Li , Zuopeng Yang , Pengyu Chen , Zhenxin Zhang , Ning Li , Jun Zheng","doi":"10.1016/j.csi.2025.103983","DOIUrl":"10.1016/j.csi.2025.103983","url":null,"abstract":"<div><div>Large Language Models (LLMs) have become integral in advancing content understanding and generation, leading to the proliferation of Embedding as a Service (EaaS) within cloud computing platforms. EaaS leverages LLMs to offer scalable, on-demand linguistic embeddings, enhancing various cloud-based applications. However, the proprietary nature of EaaS makes it a target for model extraction attacks, where the timing of such infringements often remains elusive. This paper introduces TimeMarker, a novel framework that enhances temporal traceability in cloud computing environments by embedding distinct watermarks at different sub-periods, marking the first attempt to identify the timing of model extraction attacks. TimeMarker employs an adaptive watermark strength method based on information entropy and frequency domain transformations to refine the detection accuracy of model extraction attacks within cloud infrastructures. The granularity of time frame identification for theft improves as more sub-periods are used. Furthermore, our approach investigates single sub-period theft and extends to multi-sub-period theft scenarios where attackers steal data across many sub-periods to train their models in cloud settings. Validated across five widely used datasets, TimeMarker is capable of detecting model extraction over various sub-periods and assessing its impact on the accuracy and robustness of large models deployed in the cloud. The results demonstrate that TimeMarker effectively identifies different periods of extraction attacks, enhancing EaaS security within cloud computing and extending traditional watermarking to offer copyright protection for LLMs.</div></div>","PeriodicalId":50635,"journal":{"name":"Computer Standards & Interfaces","volume":"94 ","pages":"Article 103983"},"PeriodicalIF":4.1,"publicationDate":"2025-02-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143452760","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Shujiang Xu , Shishi Dong , Lianhai Wang , Miodrag J. Mihaljevié , Shuhui Zhang , Wei Shao , Qizheng Wang
{"title":"Blockchain-based secure data sharing with overlapping clustering and searchable encryption","authors":"Shujiang Xu , Shishi Dong , Lianhai Wang , Miodrag J. Mihaljevié , Shuhui Zhang , Wei Shao , Qizheng Wang","doi":"10.1016/j.csi.2025.103979","DOIUrl":"10.1016/j.csi.2025.103979","url":null,"abstract":"<div><div>In the digital age, the importance of data sharing has significantly increased as it accelerates the release of value from data elements. Nevertheless, data confronts substantial security threats, including potential leakage during outsourcing and sharing procedures. To ensure the security of shared data, plaintext data is often replaced by encrypted data for sharing, and searchable encryption algorithms are used to improve the efficiency of sharing. However, due to the inherent limitations of searchable encryption schemes, existing secure data-sharing approaches frequently encounter inefficient search capabilities and privacy violations. This paper proposes a secure data-sharing scheme for encrypted data that integrates blockchain with an overlapping clustering technique to tackle these challenges. The proposed scheme combines blockchain with an attribute-based searchable encryption method to guarantee data transparency, trustworthiness, and confidentiality. Furthermore, the scheme significantly enhances search efficiency and accuracy by incorporating overlapping clustering and keyword relevance-based ranking strategies. Experimental results show that this scheme effectively enhances search efficiency and provides robust privacy protection.</div></div>","PeriodicalId":50635,"journal":{"name":"Computer Standards & Interfaces","volume":"93 ","pages":"Article 103979"},"PeriodicalIF":4.1,"publicationDate":"2025-02-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143387204","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Yingpan Kuang, Qiwen Wu, Riqing Chen, Xiaolong Liu
{"title":"Blockchain based lightweight authentication scheme for internet of things using lattice encryption algorithm","authors":"Yingpan Kuang, Qiwen Wu, Riqing Chen, Xiaolong Liu","doi":"10.1016/j.csi.2025.103981","DOIUrl":"10.1016/j.csi.2025.103981","url":null,"abstract":"<div><div>With the rapid development of the Internet of Things (IoT), robust and secure authentication among interconnected devices has become increasingly significant. Existing cryptographic methods, despite their effectiveness, face challenges in scalability, quantum vulnerability, and high computational demands, which are particularly problematic for resource-constrained IoT devices. This paper proposes a novel and lightweight authentication scheme for IoT devices that combines the decentralization of blockchain with the efficiency of lattice-based cryptography to address these security concerns. The proposed scheme employs a decentralized identity management model built on blockchain, eliminating vulnerable central points and enhancing system resilience. For user and device authentication, an efficient lattice-based protocol is introduced, utilizing simplified hash operations and matrix–vector multiplication for key negotiation and authentication. This approach significantly reduces both computational complexity and communication overhead compared to traditional methods such as ECC-based schemes. Specifically, at a 100-bit security level, our scheme achieves authentication and key agreement in approximately <span><math><mrow><mn>257</mn><mo>.</mo><mn>401</mn><mspace></mspace><mi>μ</mi><mi>s</mi></mrow></math></span> and maintains a communication cost of 1052 bits per authentication session. Comprehensive performance analyses demonstrate that the proposed scheme can withstand typical cryptographic attacks and offers advantages in quantum computing resistance. Additionally, the blockchain-based design ensures high scalability, making the scheme ideal for large-scale IoT deployments without performance degradation. Experimental results further validate the scheme’s practical applicability in resource-constrained IoT environments, highlighting its superior computational response times and lower communication costs compared to existing IoT authentication solutions.</div></div>","PeriodicalId":50635,"journal":{"name":"Computer Standards & Interfaces","volume":"93 ","pages":"Article 103981"},"PeriodicalIF":4.1,"publicationDate":"2025-02-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143377470","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}