Complex & Intelligent Systems最新文献

筛选
英文 中文
Spatiotemporal decoupling attention transformer for 3D skeleton-based driver action recognition
IF 5.8 2区 计算机科学
Complex & Intelligent Systems Pub Date : 2025-02-28 DOI: 10.1007/s40747-025-01811-1
Zhuoyan Xu, Jingke Xu
{"title":"Spatiotemporal decoupling attention transformer for 3D skeleton-based driver action recognition","authors":"Zhuoyan Xu, Jingke Xu","doi":"10.1007/s40747-025-01811-1","DOIUrl":"https://doi.org/10.1007/s40747-025-01811-1","url":null,"abstract":"<p>Driver action recognition is crucial for in-vehicle safety. We argue that the following factors limit the related research. First, spatial constraints and obstructions in the vehicle restrict the range of motion, resulting in similar action patterns and difficulty collecting the full body posture. Second, in skeleton-based action recognition, establishing the joint dependencies by the self-attention computation is always limited to a single frame, ignoring the effect of body spatial structure on dependence weights and inter-frame. Common convolution in temporal flow only focuses on frame-level temporal features, ignoring motion pattern features at a higher semantic level. Our work proposed a novel spatiotemporal decoupling attention transformer (SDA-TR). The SDA module uses a spatiotemporal decoupling strategy to decouple the weight computation according to body structure and directly establish joint dependencies between multiple frames. The TFA module aggregates sub-action-level and frame-level temporal features to improve similar recognition accuracy. On the Driver Action Recognition dataset Drive&amp;Act using driver upper body skeletons, SDA-TR achieves state-of-the-art performance. SDA-TR also achieved 92.2%/95.8% accuracy under the CS/CV benchmarks of NTU RGB+D 60, 88.6%/89.8% accuracy under the CS/CSet benchmarks of NTU RGB+D 120, on par with other state-of-the-art methods. Our method demonstrates great scalability and generalization for action recognition.</p>","PeriodicalId":10524,"journal":{"name":"Complex & Intelligent Systems","volume":"32 1","pages":""},"PeriodicalIF":5.8,"publicationDate":"2025-02-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143518626","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}
引用次数: 0
BSAformer: bidirectional sequence splitting aggregation attention mechanism for long term series forecasting
IF 5.8 2区 计算机科学
Complex & Intelligent Systems Pub Date : 2025-02-28 DOI: 10.1007/s40747-025-01794-z
QingBo Zhu, JiaLin Han, Sheng Yang, ZhiQiang Xie, Bo Tian, HaiBo Wan, Kai Chai
{"title":"BSAformer: bidirectional sequence splitting aggregation attention mechanism for long term series forecasting","authors":"QingBo Zhu, JiaLin Han, Sheng Yang, ZhiQiang Xie, Bo Tian, HaiBo Wan, Kai Chai","doi":"10.1007/s40747-025-01794-z","DOIUrl":"https://doi.org/10.1007/s40747-025-01794-z","url":null,"abstract":"<p>Time series forecasting plays a crucial role across various sectors, including energy, transportation, meteorology, and epidemiology. However, existing models often struggle with capturing long-term dependencies and managing computational efficiency when handling complex and extensive time series data. To address these challenges, this paper introduces the BSAformer model, which leverages a unique combination of frequency-domain Sequence Progressive Split-Aggregation (SPSA) and Bidirectional Splitting-Agg Attention (BSAA) mechanisms. The SPSA module decomposes sequences into seasonal and trend components, enhancing the model’s ability to identify cyclical patterns, while the BSAA mechanism captures forward and backward dependencies, providing a comprehensive understanding of temporal dynamics. Extensive experiments conducted on seven benchmark datasets demonstrate the BSAformer model's superior performance, with notable improvements in accuracy and efficiency over state-of-the-art models. Specifically, the BSAformer achieves significant Mean Squared Error (MSE) reductions of 63.7% on the ECL dataset, 28.1% on the Traffic dataset, and 49.8% on the ILI dataset. These results validate the model’s robustness and its adaptability across diverse time series forecasting scenarios. The insights gained from this study contribute to the advancement of time series forecasting by providing a model that improves both accuracy and computational efficiency, especially in handling long-term dependencies and complex temporal patterns. </p>","PeriodicalId":10524,"journal":{"name":"Complex & Intelligent Systems","volume":"7 1","pages":""},"PeriodicalIF":5.8,"publicationDate":"2025-02-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143518616","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}
引用次数: 0
Trust-aware privacy-preserving QoS prediction with graph neural collaborative filtering for internet of things services
IF 5.8 2区 计算机科学
Complex & Intelligent Systems Pub Date : 2025-02-28 DOI: 10.1007/s40747-025-01824-w
Weiwei Wang, Wenping Ma, Kun Yan
{"title":"Trust-aware privacy-preserving QoS prediction with graph neural collaborative filtering for internet of things services","authors":"Weiwei Wang, Wenping Ma, Kun Yan","doi":"10.1007/s40747-025-01824-w","DOIUrl":"https://doi.org/10.1007/s40747-025-01824-w","url":null,"abstract":"<p>The booming development of the Internet of Things (IoT) has led to an explosion of web services, making it more inconvenient for users to choose satisfactory services among numerous options. Therefore, ensuring quality of service (QoS) in a service-oriented IoT environment is crucial, highlighting QoS prediction as a prominent research focus. However, issues related to information credibility, user data privacy, and prediction accuracy in QoS prediction for IoT services have become significant challenges in current research. To tackle these issues, we propose TPP-GNCF, a trust-aware privacy-preserving QoS prediction framework that integrates graph neural networks with collaborative filtering methods. In TPP-GNCF, we filter out untrustworthy QoS values provided by users for certain services to select credible QoS values. Then, a message-passing graph neural network (MP-GNN) is utilized to effectively capture information transmission and relationships in the graph structure, while differential privacy is used to protect user node information. In addition, we use a similarity calculation method based on weight function in collaborative filtering to mine implicit embedded features that graph neural networks cannot directly utilize. Finally, the final missing QoS values are achieved by fusing graph neural predicted QoS and feature collaborative filtering predicted QoS. We conducted extensive experiments on the well-known WS-DREAM dataset. The results demonstrate that the TPP-GNCF framework not only surpasses existing schemes in performance but also effectively addresses issues of information credibility and user privacy.</p>","PeriodicalId":10524,"journal":{"name":"Complex & Intelligent Systems","volume":"22 1","pages":""},"PeriodicalIF":5.8,"publicationDate":"2025-02-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143518623","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}
引用次数: 0
Metalinguist: enhancing hate speech detection with cross-lingual meta-learning
IF 5.8 2区 计算机科学
Complex & Intelligent Systems Pub Date : 2025-02-27 DOI: 10.1007/s40747-025-01808-w
Ehtesham Hashmi, Sule Yildirim Yayilgan, Mohamed Abomhara
{"title":"Metalinguist: enhancing hate speech detection with cross-lingual meta-learning","authors":"Ehtesham Hashmi, Sule Yildirim Yayilgan, Mohamed Abomhara","doi":"10.1007/s40747-025-01808-w","DOIUrl":"https://doi.org/10.1007/s40747-025-01808-w","url":null,"abstract":"<p>The rise of social media has led to an increase in hate speech. Hate speech is generally described as a deliberate act of aggression aimed at a particular group, intended to harm or marginalize them based on specific attributes of their identity. While positive interactions in diverse communities can greatly enhance confidence, it is important to acknowledge that negative remarks such as hate speech can weaken community unity and present a significant impact on people’s well-being. This highlights the need for improved monitoring and guidelines on social media platforms to protect individuals from discriminatory and harmful actions. Despite extensive research on resource-rich languages, such as English and German, the detection and analysis of hate speech in less-resourced languages, such as Norwegian, remains underexplored. Addressing this gap, our study leverages a metalinguistic approach that uses advanced meta-learning techniques to enhance the detection capabilities across bilingual texts, effectively linking technical advancements directly to the pressing social issue of hate speech. In this study, we introduce techniques that adapt models that deal with hate speech detection within the same languages (intra-lingual), across different languages (cross-lingual), and techniques that adapt models to new languages with minimal extra training, independent of the model type (cross-lingual model-agnostic meta-learning-based approaches) for bilingual text analysis in Norwegian and English. Our methodology incorporates attention mechanisms (components that help the model focus on relevant parts of the text) and adaptive learning rate schedulers (tools that adjust the learning speed based on performance). Our methodology incorporates components that help the model focus on relevant parts of the text (attention mechanisms) and tools that adjust the learning speed based on performance (adaptive learning rate schedulers). We conducted various experiments using language-specific and multilingual transformers. Among these, the combination of Nor-BERT and LSTM with zero-shot and few-shot model-agnostic meta-learning achieved remarkable F1 scores of 79% and 90%, highlighting the effectiveness of our proposed framework.</p>","PeriodicalId":10524,"journal":{"name":"Complex & Intelligent Systems","volume":"24 1","pages":""},"PeriodicalIF":5.8,"publicationDate":"2025-02-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143506979","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}
引用次数: 0
A novel three-way distance-based fuzzy large margin distribution machine for imbalance classification
IF 5.8 2区 计算机科学
Complex & Intelligent Systems Pub Date : 2025-02-19 DOI: 10.1007/s40747-025-01797-w
Li Liu, Jinrui Guo, Ziqi Yin, Rui Chen, Guojun Huang
{"title":"A novel three-way distance-based fuzzy large margin distribution machine for imbalance classification","authors":"Li Liu, Jinrui Guo, Ziqi Yin, Rui Chen, Guojun Huang","doi":"10.1007/s40747-025-01797-w","DOIUrl":"https://doi.org/10.1007/s40747-025-01797-w","url":null,"abstract":"<p>Class imbalance is a prevalent issue in practical applications, which poses significant challenges for classifiers. The large margin distribution machine (LDM) introduces the margin distribution of samples to replace the traditional minimum margin, resulting in extensively enhanced classification performance. However, the hyperplane of LDM tends to be skewed toward the minority class, due to the optimization property for margin means. Moreover, the absence of non-deterministic options and measurement of the confidence level of samples further restricts the capability to manage uncertainty in imbalanced classification tasks. To solve these problems, we propose a novel three-way distance-based fuzzy large margin distribution machine (3W-DBFLDM). Specifically, we introduce a distance-based factor to mitigate the impact of sample size imbalance on classification results by increasing the distance weights of the minority class. Additionally, three-way decision model is introduced to deal with uncertainty, and the model’s robustness is further enhanced by utilizing the fuzzy membership degree that reflects the importance level of each input point. Comparative experiments conducted on UCI datasets demonstrate that the 3W-DBFLDM model surpasses other models in classification accuracy, stability, and robustness. Furthermore, the cost comparison experiment validate that the 3W-DBFLDM model reduces the overall decision cost.</p>","PeriodicalId":10524,"journal":{"name":"Complex & Intelligent Systems","volume":"12 1","pages":""},"PeriodicalIF":5.8,"publicationDate":"2025-02-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143443464","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}
引用次数: 0
Chaos-enhanced metaheuristics: classification, comparison, and convergence analysis
IF 5.8 2区 计算机科学
Complex & Intelligent Systems Pub Date : 2025-02-19 DOI: 10.1007/s40747-025-01791-2
Abdelhadi Limane, Farouq Zitouni, Saad Harous, Rihab Lakbichi, Aridj Ferhat, Abdulaziz S. Almazyad, Pradeep Jangir, Ali Wagdy Mohamed
{"title":"Chaos-enhanced metaheuristics: classification, comparison, and convergence analysis","authors":"Abdelhadi Limane, Farouq Zitouni, Saad Harous, Rihab Lakbichi, Aridj Ferhat, Abdulaziz S. Almazyad, Pradeep Jangir, Ali Wagdy Mohamed","doi":"10.1007/s40747-025-01791-2","DOIUrl":"https://doi.org/10.1007/s40747-025-01791-2","url":null,"abstract":"<p>Chaos theory, with its unique blend of randomness and ergodicity, has become a powerful tool for enhancing metaheuristic algorithms. In recent years, there has been a growing number of chaos-enhanced metaheuristic algorithms (CMAs), accompanied by a notable scarcity of studies that analyze and organize this field. To respond to this challenge, this paper comprehensively analyzes recent advances in CMAs from 2013 to 2024, proposing a novel classification scheme that systematically organizes prevalent and practical approaches for integrating chaos theory into metaheuristic algorithms based on their strategic roles. In addition, a list of 27 standard chaotic maps is explored, and a summary of the application domains where CMAs have demonstrably improved performance is provided. To experimentally demonstrate the capability of chaos theory to enhance metaheuristic algorithms that face common issues such as susceptibility to local optima, non-smooth transitions between global and local search phases, and decreased diversity, we developed a chaotic variant of the recently proposed RIME optimizer, which also encounters these challenges to some extent. We tested C-RIME on the CEC2022 benchmark suite, rigorously analyzing numerical results using statistical metrics. Non-parametric statistical tests, including the Friedman and Wilcoxon signed-rank tests, were also used to validate the findings. The results demonstrated promising performance, with 14 out of 21 chaotic variants outperforming the non-chaotic variant, whereas the piecewise map-based variant achieved the best results. In addition, C-RIME outperformed ten state-of-the-art metaheuristic algorithms regarding solution quality and convergence speed.</p>","PeriodicalId":10524,"journal":{"name":"Complex & Intelligent Systems","volume":"16 1","pages":""},"PeriodicalIF":5.8,"publicationDate":"2025-02-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143443465","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}
引用次数: 0
Control strategy of robotic manipulator based on multi-task reinforcement learning
IF 5.8 2区 计算机科学
Complex & Intelligent Systems Pub Date : 2025-02-19 DOI: 10.1007/s40747-025-01816-w
Tao Wang, Ziming Ruan, Yuyan Wang, Chong Chen
{"title":"Control strategy of robotic manipulator based on multi-task reinforcement learning","authors":"Tao Wang, Ziming Ruan, Yuyan Wang, Chong Chen","doi":"10.1007/s40747-025-01816-w","DOIUrl":"https://doi.org/10.1007/s40747-025-01816-w","url":null,"abstract":"<p>Multi-task learning is important in reinforcement learning where simultaneously training across different tasks allows for leveraging shared information among them, typically leading to better performance than single-task learning. While joint training of multiple tasks permits parameter sharing between tasks, the optimization challenge becomes crucial—identifying which parameters should be reused and managing potential gradient conflicts arising from different tasks. To tackle this issue, instead of uniform parameter sharing, we propose an adjudicate reconfiguration network model, which we integrate into the Soft Actor-Critic (SAC) algorithm to address the optimization problems brought about by parameter sharing in multi-task reinforcement learning algorithms. The decision reconstruction network model is designed to achieve cross-network layer information exchange between network layers by dynamically adjusting and reconfiguring the network hierarchy, which can overcome the inherent limitations of traditional network architecture in handling multitasking scenarios. The SAC algorithm based on the decision reconstruction network model can achieve simultaneous training in multiple tasks, effectively learning and integrating relevant knowledge of each task. Finally, the proposed algorithm is evaluated in a multi-task environment of the Meta-World, a benchmark for multi-task reinforcement learning containing robotic manipulation tasks, and the multi-task MUJOCO environment.</p>","PeriodicalId":10524,"journal":{"name":"Complex & Intelligent Systems","volume":"5 1","pages":""},"PeriodicalIF":5.8,"publicationDate":"2025-02-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143443462","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}
引用次数: 0
Tailored meta-learning for dual trajectory transformer: advancing generalized trajectory prediction
IF 5.8 2区 计算机科学
Complex & Intelligent Systems Pub Date : 2025-02-19 DOI: 10.1007/s40747-025-01802-2
Feilong Huang, Zide Fan, Xiaohe Li, Wenhui Zhang, Pengfei Li, Ying Geng, Keqing Zhu
{"title":"Tailored meta-learning for dual trajectory transformer: advancing generalized trajectory prediction","authors":"Feilong Huang, Zide Fan, Xiaohe Li, Wenhui Zhang, Pengfei Li, Ying Geng, Keqing Zhu","doi":"10.1007/s40747-025-01802-2","DOIUrl":"https://doi.org/10.1007/s40747-025-01802-2","url":null,"abstract":"<p>Trajectory prediction has become increasingly critical in various applications such as autonomous driving and robotic navigation. However, due to the significant variations in trajectory patterns across different scenarios, models trained in known environments often falter in unseen ones. To learn a generalized model that can directly handle unseen domains without requiring any model updating, we propose a novel tailored meta-learning-based trajectory prediction model called DTM. Our approach integrates a dual trajectory transformer (Dual_TT) equipped with an agent-consistency loss, facilitating a comprehensive exploration of both individual intentions and group dynamics across diverse scenarios. Building on this, we propose a tailored meta-learning framework (TMG) to simulate the generalization process between source and target domains during the training phase. In the task construction phase, we employ multi-dimensional labels to precisely define and distinguish between different domains. During the dual-phase parameter update, we partially fix crucial attention mechanism parameters and apply an attention alignment loss to harmonize domain-invariant and specific features. We also incorporate a Serial and Parallel Training (SPT) strategy to significantly enhance task processing and the model’s adaptability to domain shifts. Extensive testing across various domains demonstrates that our DTM model not only outperforms existing top-performing baselines on real-world datasets but also validates the effectiveness of our design through ablation studies.</p>","PeriodicalId":10524,"journal":{"name":"Complex & Intelligent Systems","volume":"1 1","pages":""},"PeriodicalIF":5.8,"publicationDate":"2025-02-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143443461","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}
引用次数: 0
Explainable and secure framework for autism prediction using multimodal eye tracking and kinematic data
IF 5.8 2区 计算机科学
Complex & Intelligent Systems Pub Date : 2025-02-19 DOI: 10.1007/s40747-025-01790-3
Ahmad Almadhor, Areej Alasiry, Shtwai Alsubai, Abdullah Al Hejaili, Urban Kovac, Sidra Abbas
{"title":"Explainable and secure framework for autism prediction using multimodal eye tracking and kinematic data","authors":"Ahmad Almadhor, Areej Alasiry, Shtwai Alsubai, Abdullah Al Hejaili, Urban Kovac, Sidra Abbas","doi":"10.1007/s40747-025-01790-3","DOIUrl":"https://doi.org/10.1007/s40747-025-01790-3","url":null,"abstract":"<p>Autism Spectrum Disorder (ASD) is a complex neurodevelopmental condition marked by difficulties in social skills, repetitive behaviours, and communication. Early and accurate diagnosis is essential for effective intervention and support. This paper proposes a secure and privacy-preserving framework for diagnosing ASD by integrating multimodal kinematic and eye movement sensory data, Deep Neural Networks (DNN), and Explainable Artificial Intelligence (XAI). Federated Learning (FL), a distributed machine learning approach, is utilized to ensure data privacy by training models across multiple devices without centralizing sensitive data. In our evaluation, we employ FL using a shallow DNN as the shared model and Federated Averaging (FedAvg) as the aggregation algorithm. We conduct experiments across two scenarios for each dataset: the first using FL with all features and the second using FL with features selected by XAI. The experiments, conducted with three clients over three rounds of training, show that the L_General dataset produces the best results, with Client 2 achieving an accuracy of 99.99% and Client 1 achieving 88%. This study underscores FL’s potential to preserve privacy and security while maintaining high diagnostic accuracy, making it a viable solution for healthcare applications involving sensitive data.</p>","PeriodicalId":10524,"journal":{"name":"Complex & Intelligent Systems","volume":"13 1","pages":""},"PeriodicalIF":5.8,"publicationDate":"2025-02-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143443463","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}
引用次数: 0
Cl2sum: abstractive summarization via contrastive prompt constructed by LLMs hallucination
IF 5.8 2区 计算机科学
Complex & Intelligent Systems Pub Date : 2025-02-19 DOI: 10.1007/s40747-025-01795-y
Xiang Huang, Qiong Nong, Xiaobo Wang, Hongcheng Zhang, Kunpeng Du, Chunlin Yin, Li Yang, Bin Yan, Xuan Zhang
{"title":"Cl2sum: abstractive summarization via contrastive prompt constructed by LLMs hallucination","authors":"Xiang Huang, Qiong Nong, Xiaobo Wang, Hongcheng Zhang, Kunpeng Du, Chunlin Yin, Li Yang, Bin Yan, Xuan Zhang","doi":"10.1007/s40747-025-01795-y","DOIUrl":"https://doi.org/10.1007/s40747-025-01795-y","url":null,"abstract":"<p>The rise of Large Language Models (LLMs) has further led to the development of text summarization techniques and also brought more attention to the problem of hallucination in the research of text summarization. Existing work in current text summarization research based on LLMs typically uses In-Context Learning (ICL) to supply accurate (document-summary) pairs of samples to the model, thus allowing the model to be more explicit in predicting the target. However, in this way, models can only determine what to do, without explicitly prohibiting what models cannot do. It is highly likely to lead to increased hallucinations due to excessive model-free play. In this paper, to alleviate the problem of hallucination in text summarization based on LLMs, we propose CL2Sum, a method that combines Contrastive Learning (CL) and ICL for summarization. After analysing the generated summaries of LLMs and summarising their hallucination types, we provided the models with accurate summaries and summaries containing hallucinations as ICL instances, either automatically or artificially. It aims to guide the model to make accurate predictions according to positive samples while also avoiding hallucinations similar to those in negative samples. Finally, a series of comparative experiments were conducted on summary datasets of different lengths and languages. The results show that CL2Sum effectively alleviates the hallucination problem of text summaries while also improving the overall quality of the generated summaries. Moreover, it can be widely adapted to text summarization tasks in different scenarios with a certain degree of robustness.</p>","PeriodicalId":10524,"journal":{"name":"Complex & Intelligent Systems","volume":"13 1","pages":""},"PeriodicalIF":5.8,"publicationDate":"2025-02-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143443478","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}
引用次数: 0
0
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
相关产品
×
本文献相关产品
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信