Decision Support Systems最新文献_第4页

“Language is the dress of thought”: A new method for automatic detection of AI-generated text “语言是思想的外衣”：一种人工智能生成文本自动检测的新方法

IF 6.8 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Decision Support Systems

Pub Date : 2025-11-25 DOI: 10.1016/j.dss.2025.114578

Zhenhua Wang , Guang Xu , Ming Ren

While AI technologies have garnered widespread attention for their revolutionary text generation capabilities, concerns have arisen regarding the risks associated with AI-generated text (AIGT), especially when used maliciously. Motivated by the recognition that AIGT is generated based on high-probability tokens, a process that inherently differs from the biological-based thought processes underlying human-written text (HWT), we trace and build upon theories of the language latent level to explore the fundamental differences between AIGT and HWT, particularly in terms of potentiality, logicality, and complexity. A novel method named LA2HDetect is proposed for automatic AIGT detection. Specifically, we discover that HWT exhibits higher potentiality than AIGT; AIGT and HWT each possesses unique characteristics in terms of logicality and complexity. These human-AI differences collectively form the decision-making mechanism of LA2HDetect. Extensive experiments on general domain datasets confirm the competitiveness and robustness of LA2HDetect, which outperforms existing methods. In addition, we evaluate the extensibility of LA2HDetect in multiple vertical domains, and explore the insights across progressively advanced AI models.

虽然人工智能技术因其革命性的文本生成能力而受到广泛关注，但人们也开始关注与人工智能生成文本（AIGT）相关的风险，尤其是在恶意使用时。由于认识到AIGT是基于高概率令牌生成的，这一过程本质上不同于人类书面文本（HWT）背后基于生物的思维过程，我们追踪并建立了语言潜在水平的理论，以探索AIGT和HWT之间的根本差异，特别是在潜力、逻辑性和复杂性方面。提出了一种新的AIGT自动检测方法LA2HDetect。具体而言，我们发现HWT比AIGT表现出更高的电位；AIGT和HWT在逻辑性和复杂性方面各具特色。这些人类与人工智能的差异共同形成了LA2HDetect的决策机制。在一般领域数据集上的大量实验证实了LA2HDetect的竞争力和鲁棒性，优于现有方法。此外，我们评估了LA2HDetect在多个垂直领域的可扩展性，并探索了跨逐步先进的人工智能模型的见解。

{"title":"“Language is the dress of thought”: A new method for automatic detection of AI-generated text","authors":"Zhenhua Wang , Guang Xu , Ming Ren","doi":"10.1016/j.dss.2025.114578","DOIUrl":"10.1016/j.dss.2025.114578","url":null,"abstract":"<div><div>While AI technologies have garnered widespread attention for their revolutionary text generation capabilities, concerns have arisen regarding the risks associated with AI-generated text (AIGT), especially when used maliciously. Motivated by the recognition that AIGT is generated based on high-probability tokens, a process that inherently differs from the biological-based thought processes underlying human-written text (HWT), we trace and build upon theories of the language latent level to explore the fundamental differences between AIGT and HWT, particularly in terms of <em>potentiality</em>, <em>logicality</em>, and <em>complexity</em>. A novel method named LA2HDetect is proposed for automatic AIGT detection. Specifically, we discover that HWT exhibits higher <em>potentiality</em> than AIGT; AIGT and HWT each possesses unique characteristics in terms of <em>logicality</em> and <em>complexity</em>. These human-AI differences collectively form the decision-making mechanism of LA2HDetect. Extensive experiments on general domain datasets confirm the competitiveness and robustness of LA2HDetect, which outperforms existing methods. In addition, we evaluate the extensibility of LA2HDetect in multiple vertical domains, and explore the insights across progressively advanced AI models.</div></div>","PeriodicalId":55181,"journal":{"name":"Decision Support Systems","volume":"201 ","pages":"Article 114578"},"PeriodicalIF":6.8,"publicationDate":"2025-11-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145598777","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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 method for testing adverse selection with IoT data: Evidence from China's auto insurance market 用物联网数据测试逆向选择的新方法：来自中国车险市场的证据

IF 6.8 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Decision Support Systems

Pub Date : 2025-11-19 DOI: 10.1016/j.dss.2025.114576

Esther Yanfei Jin , Wei Jiang , Zhiqiang Zheng

Adverse selection remains a significant challenge in the insurance industry, often resulting in substantial financial losses for insurers. The primary hurdle in addressing the issue lies in accurately identifying and quantifying adverse selection. Traditional methods often fail to adequately account for the heterogeneity of insurance purchasers and the endogenous nature of their insurance decisions. This study introduces an innovative approach that integrates the Gaussian Mixture Model and the regression-based model from Dionne et al. [18] to assess adverse selection, addressing the limitations of previous methods. Through comprehensive simulations, we demonstrate that our method yields unbiased estimates, outperforming existing approaches. Applied to China's automobile insurance market, this method leverages IoT-based telematics data to capture risk heterogeneity among policyholders more effectively than relying solely on traditional policy information. The results offer robust evidence of adverse selection, in contrast to conventional methods that fail to detect this phenomenon due to their inability to account for underlying risk and insurance choice heterogeneity. Our approach offers insurers a robust framework for identifying information asymmetries in the market, thereby enabling the development of more targeted policy interventions and risk management strategies.

逆向选择仍然是保险业面临的一个重大挑战，通常会给保险公司带来巨大的经济损失。解决这个问题的主要障碍在于准确地识别和量化逆向选择。传统的方法往往不能充分考虑保险购买者的异质性和他们的保险决策的内生性质。本研究引入了一种创新的方法，该方法集成了高斯混合模型和Dionne等人的基于回归的模型来评估逆向选择，解决了以前方法的局限性。通过综合模拟，我们证明了我们的方法产生无偏估计，优于现有的方法。该方法应用于中国车险市场，利用基于物联网的远程信息处理数据，比仅仅依靠传统的保单信息更有效地捕捉投保人之间的风险异质性。研究结果为逆向选择提供了有力的证据，而传统方法由于无法解释潜在风险和保险选择异质性而无法检测到这一现象。我们的方法为保险公司提供了一个强有力的框架，用于识别市场中的信息不对称，从而能够制定更有针对性的政策干预和风险管理策略。

{"title":"A novel method for testing adverse selection with IoT data: Evidence from China's auto insurance market","authors":"Esther Yanfei Jin , Wei Jiang , Zhiqiang Zheng","doi":"10.1016/j.dss.2025.114576","DOIUrl":"10.1016/j.dss.2025.114576","url":null,"abstract":"<div><div>Adverse selection remains a significant challenge in the insurance industry, often resulting in substantial financial losses for insurers. The primary hurdle in addressing the issue lies in accurately identifying and quantifying adverse selection. Traditional methods often fail to adequately account for the heterogeneity of insurance purchasers and the endogenous nature of their insurance decisions. This study introduces an innovative approach that integrates the Gaussian Mixture Model and the regression-based model from Dionne et al. [18] to assess adverse selection, addressing the limitations of previous methods. Through comprehensive simulations, we demonstrate that our method yields unbiased estimates, outperforming existing approaches. Applied to China's automobile insurance market, this method leverages IoT-based telematics data to capture risk heterogeneity among policyholders more effectively than relying solely on traditional policy information. The results offer robust evidence of adverse selection, in contrast to conventional methods that fail to detect this phenomenon due to their inability to account for underlying risk and insurance choice heterogeneity. Our approach offers insurers a robust framework for identifying information asymmetries in the market, thereby enabling the development of more targeted policy interventions and risk management strategies.</div></div>","PeriodicalId":55181,"journal":{"name":"Decision Support Systems","volume":"200 ","pages":"Article 114576"},"PeriodicalIF":6.8,"publicationDate":"2025-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145553606","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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

Published content vs. live-streamed: Empirical from digital content activities in online healthcare communities 发布内容与直播：来自在线医疗保健社区数字内容活动的经验

IF 6.8 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Decision Support Systems

Pub Date : 2025-11-19 DOI: 10.1016/j.dss.2025.114577

Liuan Wang , Yuke Luo , Linan Zhang

Online Health Communities (OHCs) serve as a platform for individuals seeking health support, where the exchange of health information harbors the potential to generate substantial social value. Existing literature reveals that previous research has primarily focused on the incentives for doctors' information sharing. However, the mechanism through which this information sharing translates into tangible benefits for hospitals remains unclear. To address these gaps, this study interprets it as a kind of digital content activity (DCA) and delves into its impact on hospitals' online demand in OHC. Analyzing data from over 2000 active hospitals on a leading OHC in China, our findings indicate that hospitals' published content activity consistently increases their online demand in OHC. In contrast, live-streamed content activity decreases online demand in the short term, yet this impact turns positive in the long term. Furthermore, a hospital's organizational capital enhances the impact of live-streamed content activity, while the hospital's reputation strengthens the long-term positive impact of published content activity. This study offers a novel perspective for understanding knowledge sharing within OHCs, providing practical insights for OHCs and hospitals.

在线卫生社区（OHCs）是寻求卫生支持的个人的平台，其中卫生信息的交换具有产生巨大社会价值的潜力。现有文献显示，以往的研究主要集中在医生信息共享的激励机制上。然而，这种信息共享转化为医院切实利益的机制尚不清楚。为了解决这些差距，本研究将其解释为一种数字内容活动（DCA），并深入研究其对OHC中医院在线需求的影响。通过分析中国一家领先的OHC上2000多家活跃医院的数据，我们的研究结果表明，医院发布的内容活动持续增加了他们对OHC的在线需求。相比之下，直播内容活动在短期内会减少在线需求，但从长期来看，这种影响会转为积极。此外，医院的组织资本增强了直播内容活动的影响力，而医院的声誉增强了发布内容活动的长期积极影响。本研究为理解OHCs内部的知识共享提供了一个新的视角，为OHCs和医院提供了实用的见解。

{"title":"Published content vs. live-streamed: Empirical from digital content activities in online healthcare communities","authors":"Liuan Wang , Yuke Luo , Linan Zhang","doi":"10.1016/j.dss.2025.114577","DOIUrl":"10.1016/j.dss.2025.114577","url":null,"abstract":"<div><div>Online Health Communities (OHCs) serve as a platform for individuals seeking health support, where the exchange of health information harbors the potential to generate substantial social value. Existing literature reveals that previous research has primarily focused on the incentives for doctors' information sharing. However, the mechanism through which this information sharing translates into tangible benefits for hospitals remains unclear. To address these gaps, this study interprets it as a kind of digital content activity (DCA) and delves into its impact on hospitals' online demand in OHC. Analyzing data from over 2000 active hospitals on a leading OHC in China, our findings indicate that hospitals' published content activity consistently increases their online demand in OHC. In contrast, live-streamed content activity decreases online demand in the short term, yet this impact turns positive in the long term. Furthermore, a hospital's organizational capital enhances the impact of live-streamed content activity, while the hospital's reputation strengthens the long-term positive impact of published content activity. This study offers a novel perspective for understanding knowledge sharing within OHCs, providing practical insights for OHCs and hospitals.</div></div>","PeriodicalId":55181,"journal":{"name":"Decision Support Systems","volume":"200 ","pages":"Article 114577"},"PeriodicalIF":6.8,"publicationDate":"2025-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145559693","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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

What makes a well-performing NFT collection initial offering campaign: Evidence from OpenSea Drop 是什么让NFT系列的首次发行活动表现良好：来自OpenSea的证据掉落

IF 6.8 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Decision Support Systems

Pub Date : 2025-11-19 DOI: 10.1016/j.dss.2025.114575

Zhichao Wu , Xi Zhao , Xiaoni Lu

The Drop feature on OpenSea provides creators with a standardized tool for designing NFT collection (NFTC) initial offering campaigns. This study examines the impact of campaign design elements on sales performance. Analyzing 693 NFTCs, we reveal an inverted U-shaped relationship between target size and sales outcomes, attributable to the balance between social proof and scarcity. Additionally, we observe a positive effect of incorporating pre-sale stages, which is driven by social proof. Notably, OpenSea's official certification, as a significant credibility signal, moderates these effects. This research advances the understanding of social proof theory within the Web3.0 context, offering actionable insights for NFT creators to optimize campaign strategies and for platform managers to enhance the effectiveness of the Drop feature.

OpenSea的Drop功能为创建者提供了一个标准化的工具来设计NFTC首次发行活动。本研究探讨活动设计元素对销售绩效的影响。通过对693个国家的分析，我们发现目标规模与销售结果之间存在倒u型关系，这可归因于社会认同与稀缺性之间的平衡。此外，我们观察到加入预售阶段的积极影响，这是由社会认同驱动的。值得注意的是，OpenSea的官方认证作为一个重要的可信度信号，缓和了这些影响。这项研究促进了对Web3.0背景下社会认同理论的理解，为NFT创作者优化活动策略和平台管理者提高Drop功能的有效性提供了可操作的见解。

引用次数: 0

Breaking boundaries: Investigating the formation of cross-domain collaboration on social media platforms 打破边界：调查社交媒体平台上跨领域协作的形成

IF 6.8 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Decision Support Systems

Pub Date : 2025-11-13 DOI: 10.1016/j.dss.2025.114574

Mengxiao Zhu , Lin Liu , Chunke Su

Creators on social media platforms are increasingly engaging in collaborative content generation. Given the recognized value of integrating diverse perspectives and expertise from different domains, such as fostering innovation, improving content quality, and expanding audience engagement, this study aims to investigate the decision-making dynamics among creators involved in cross-domain collaboration. Drawing on social identity theory, we examine the effect of content domain differentiation on the formation of collaborative relationships and how creators' attributes of content diversity and influencing power alter these effects. Our data were collected from Bilibili, one of the largest Chinese video-sharing platforms, which offers a joint submission feature allowing multiple creators to publish their generated videos. We employ exponential random graph models (ERGMs) to analyze the formation of a collaboration network comprising 2490 creators. The findings reveal that content domain differentiation is negatively related to the formation of collaborative relationships, indicating that cross-domain collaborative relationships are less likely to occur compared to within-domain ones on social media. Furthermore, content diversity mitigates the negative effect of content domain differentiation, suggesting that creators with higher content diversity are more inclined to engage in cross-domain collaborations. Regarding influencing power, creators with less reach and activeness are more likely to participate in cross-domain collaboration. Interestingly, creators with institutional authority are less likely to form cross-domain collaborations, whereas those with individual authority are more likely, compared to non-authority creators. This study highlights the challenges in fostering cross-domain collaborative relationships on social media and elucidates actionable strategies to promote such collaborations.

社交媒体平台上的创作者越来越多地参与到协作内容生成中。鉴于整合来自不同领域的不同观点和专业知识的公认价值，例如促进创新、提高内容质量和扩大受众参与度，本研究旨在调查参与跨领域合作的创作者之间的决策动态。本文以社会认同理论为基础，考察了内容领域分化对协作关系形成的影响，以及创作者的内容多样性属性和影响力如何改变这些影响。我们的数据来自Bilibili，这是中国最大的视频分享平台之一，该平台提供联合提交功能，允许多个创作者发布自己制作的视频。我们使用指数随机图模型（ergm）来分析由2490个创建者组成的协作网络的形成。研究发现，内容领域分化与协作关系的形成呈负相关，表明在社交媒体上，跨领域的协作关系比领域内的协作关系更不容易发生。此外，内容多样性可以缓解内容领域分化的负面影响，表明内容多样性越高的创作者更倾向于进行跨领域合作。在影响力方面，覆盖面和活跃度较低的创作者更有可能参与跨领域合作。有趣的是，与非权威的创造者相比，拥有机构权威的创造者不太可能形成跨领域合作，而拥有个人权威的创造者则更有可能形成跨领域合作。本研究强调了在社交媒体上培养跨领域合作关系所面临的挑战，并阐明了促进这种合作的可行策略。

{"title":"Breaking boundaries: Investigating the formation of cross-domain collaboration on social media platforms","authors":"Mengxiao Zhu , Lin Liu , Chunke Su","doi":"10.1016/j.dss.2025.114574","DOIUrl":"10.1016/j.dss.2025.114574","url":null,"abstract":"<div><div>Creators on social media platforms are increasingly engaging in collaborative content generation. Given the recognized value of integrating diverse perspectives and expertise from different domains, such as fostering innovation, improving content quality, and expanding audience engagement, this study aims to investigate the decision-making dynamics among creators involved in cross-domain collaboration. Drawing on social identity theory, we examine the effect of content domain differentiation on the formation of collaborative relationships and how creators' attributes of content diversity and influencing power alter these effects. Our data were collected from Bilibili, one of the largest Chinese video-sharing platforms, which offers a joint submission feature allowing multiple creators to publish their generated videos. We employ exponential random graph models (ERGMs) to analyze the formation of a collaboration network comprising 2490 creators. The findings reveal that content domain differentiation is negatively related to the formation of collaborative relationships, indicating that cross-domain collaborative relationships are less likely to occur compared to within-domain ones on social media. Furthermore, content diversity mitigates the negative effect of content domain differentiation, suggesting that creators with higher content diversity are more inclined to engage in cross-domain collaborations. Regarding influencing power, creators with less reach and activeness are more likely to participate in cross-domain collaboration. Interestingly, creators with institutional authority are less likely to form cross-domain collaborations, whereas those with individual authority are more likely, compared to non-authority creators. This study highlights the challenges in fostering cross-domain collaborative relationships on social media and elucidates actionable strategies to promote such collaborations.</div></div>","PeriodicalId":55181,"journal":{"name":"Decision Support Systems","volume":"200 ","pages":"Article 114574"},"PeriodicalIF":6.8,"publicationDate":"2025-11-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145520732","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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

ProMatch: A novel dynamic process-unpacking approach for two-way proactive recruitment ProMatch：一种新颖的动态流程拆解方法，用于双向主动招聘

IF 6.8 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Decision Support Systems

Pub Date : 2025-11-13 DOI: 10.1016/j.dss.2025.114564

Xiaowei Shi , Cong Wang , Qiang Wei

Online recruitment platforms have revolutionized labor markets by enabling bidirectional engagement between job seekers and employers, but this transformation has also introduced complex decision-making challenges due to information overload and parallel decision processes. Existing research and algorithms often focus on static and one-way models, neglecting the dynamic feedback loops and preference adjustments inherent in two-way proactive recruitment. This study introduces ProMatch, a novel person-job matching approach designed to support decision-making for both sides. ProMatch formalizes recruitment as a multi-stage process involving intention formation, preference updates, and bilateral matching, capturing the sequential dependencies between decision outcomes. It also incorporates a dynamic preference learning mechanism grounded in self-regulation theory, which iteratively refines preferences using textual profiles, historical interactions, and feedback. Validation using a real-world IT enterprise dataset and a two-week field experiment demonstrates ProMatch’s effectiveness. Results show a 9% increase in click-through rates and a 20% improvement in interview-through rates, highlighting its ability to enhance prediction accuracy by dynamically modeling evolving preferences. ProMatch’s innovations offer actionable decision support for both job seekers and employers, ultimately improving recruitment efficiency and cost-effectiveness in modern recruitment ecosystems.

在线招聘平台通过实现求职者和雇主之间的双向互动，彻底改变了劳动力市场，但这种转变也带来了复杂的决策挑战，因为信息过载和决策过程并行。现有的研究和算法往往侧重于静态和单向模型，而忽略了双向主动招聘中固有的动态反馈循环和偏好调整。本研究引入ProMatch，一种新颖的个人-工作匹配方法，旨在支持双方的决策。ProMatch将招聘形式化为一个多阶段的过程，包括意向形成、偏好更新和双边匹配，捕捉决策结果之间的顺序依赖关系。它还结合了基于自我调节理论的动态偏好学习机制，该机制使用文本概要、历史交互和反馈迭代地改进偏好。使用真实的IT企业数据集和为期两周的现场实验验证了ProMatch的有效性。结果显示，点击率提高了9%，采访通过率提高了20%，突出了通过动态建模不断变化的偏好来提高预测准确性的能力。ProMatch的创新为求职者和雇主提供可操作的决策支持，最终提高现代招聘生态系统的招聘效率和成本效益。

{"title":"ProMatch: A novel dynamic process-unpacking approach for two-way proactive recruitment","authors":"Xiaowei Shi , Cong Wang , Qiang Wei","doi":"10.1016/j.dss.2025.114564","DOIUrl":"10.1016/j.dss.2025.114564","url":null,"abstract":"<div><div>Online recruitment platforms have revolutionized labor markets by enabling bidirectional engagement between job seekers and employers, but this transformation has also introduced complex decision-making challenges due to information overload and parallel decision processes. Existing research and algorithms often focus on static and one-way models, neglecting the dynamic feedback loops and preference adjustments inherent in two-way proactive recruitment. This study introduces ProMatch, a novel person-job matching approach designed to support decision-making for both sides. ProMatch formalizes recruitment as a multi-stage process involving intention formation, preference updates, and bilateral matching, capturing the sequential dependencies between decision outcomes. It also incorporates a dynamic preference learning mechanism grounded in self-regulation theory, which iteratively refines preferences using textual profiles, historical interactions, and feedback. Validation using a real-world IT enterprise dataset and a two-week field experiment demonstrates ProMatch’s effectiveness. Results show a 9% increase in click-through rates and a 20% improvement in interview-through rates, highlighting its ability to enhance prediction accuracy by dynamically modeling evolving preferences. ProMatch’s innovations offer actionable decision support for both job seekers and employers, ultimately improving recruitment efficiency and cost-effectiveness in modern recruitment ecosystems.</div></div>","PeriodicalId":55181,"journal":{"name":"Decision Support Systems","volume":"200 ","pages":"Article 114564"},"PeriodicalIF":6.8,"publicationDate":"2025-11-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145531617","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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

Effects of artificial intelligence usage and knowledge-based dynamic capabilities on organizational innovation: A configurational approach 人工智能使用和基于知识的动态能力对组织创新的影响：一种配置方法

IF 6.8 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Decision Support Systems

Pub Date : 2025-11-06 DOI: 10.1016/j.dss.2025.114573

Meng An , Jiabao Lin , Jose Benitez

Many antecedents of organizational innovation have been examined in isolation, overlooking their synergistic and threshold effects. To address this gap, this study draws on resource orchestration theory to investigate how AI usage and knowledge-based dynamic capabilities, i.e., knowledge generation capability, knowledge acquisition capability, and market-sensing capability, jointly drive exploratory and exploitative innovation. Using survey data from 218 Chinese firms, we apply fuzzy-set qualitative comparative analysis (fsQCA) to identify multiple sufficient configurations that generate high innovation, highlighting heterogeneous pathways shaped by firm size and industry context. To complement these findings, we conduct necessary condition analysis (NCA), which reveals critical threshold levels for AI usage and knowledge capabilities that should be met regardless of the chosen configuration. Furthermore, we map fsQCA results with three types of interdependencies among AI usage and knowledge-based capabilities—complementarity, contingency, and substitution—to form configurations that lead to different organizational innovations. This study enriches configurational theory on organizational innovation, expands the theoretical boundaries of AI-enabled innovation, and provides actionable decision support for resource allocation and capability development under digital transformation.

许多组织创新的前因被孤立地考察，忽略了它们的协同效应和门槛效应。为了解决这一差距，本研究借鉴资源编排理论，探讨人工智能的使用和基于知识的动态能力，即知识生成能力、知识获取能力和市场感知能力，如何共同推动探索性和开发性创新。利用218家中国企业的调查数据，我们运用模糊集定性比较分析（fsQCA）来识别产生高创新的多种充分配置，突出了由企业规模和行业背景形成的异质路径。为了补充这些发现，我们进行了必要条件分析（NCA），揭示了人工智能使用和知识能力的关键阈值水平，无论所选择的配置如何，都应该满足这些阈值水平。此外，我们将fsQCA结果与人工智能使用和基于知识的能力之间的三种相互依赖关系——互补性、偶然性和替代性——进行映射，以形成导致不同组织创新的配置。本研究丰富了组织创新的构型理论，拓展了人工智能创新的理论边界，为数字化转型下的资源配置和能力发展提供了可操作的决策支持。

{"title":"Effects of artificial intelligence usage and knowledge-based dynamic capabilities on organizational innovation: A configurational approach","authors":"Meng An , Jiabao Lin , Jose Benitez","doi":"10.1016/j.dss.2025.114573","DOIUrl":"10.1016/j.dss.2025.114573","url":null,"abstract":"<div><div>Many antecedents of organizational innovation have been examined in isolation, overlooking their synergistic and threshold effects. To address this gap, this study draws on resource orchestration theory to investigate how AI usage and knowledge-based dynamic capabilities, i.e., knowledge generation capability, knowledge acquisition capability, and market-sensing capability, jointly drive exploratory and exploitative innovation. Using survey data from 218 Chinese firms, we apply fuzzy-set qualitative comparative analysis (fsQCA) to identify multiple sufficient configurations that generate high innovation, highlighting heterogeneous pathways shaped by firm size and industry context. To complement these findings, we conduct necessary condition analysis (NCA), which reveals critical threshold levels for AI usage and knowledge capabilities that should be met regardless of the chosen configuration. Furthermore, we map fsQCA results with three types of interdependencies among AI usage and knowledge-based capabilities—complementarity, contingency, and substitution—to form configurations that lead to different organizational innovations. This study enriches configurational theory on organizational innovation, expands the theoretical boundaries of AI-enabled innovation, and provides actionable decision support for resource allocation and capability development under digital transformation.</div></div>","PeriodicalId":55181,"journal":{"name":"Decision Support Systems","volume":"200 ","pages":"Article 114573"},"PeriodicalIF":6.8,"publicationDate":"2025-11-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145461300","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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

AI nudging and decision quality: Evidence from randomized experiments in online recommendation setting 人工智能推动和决策质量：来自在线推荐设置随机实验的证据

IF 6.8 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Decision Support Systems

Pub Date : 2025-11-04 DOI: 10.1016/j.dss.2025.114565

Yuxiao Luo , Nanda Kumar , Adel Yazdanmehr

This study explores the impacts of AI nudging on customer purchase decisions. Digital nudging is a well-established technique used to alter people's behaviors in a predictable way. With the rapid development of Artificial Intelligence/Machine Learning (AI/ML) and the widespread integration of the “black box” algorithm in the digital choice architecture, personalized targeting nudges can vastly influence individual and collective behaviors and lead to undesired consequences. AI nudge refers to the situation when human outsources developing and implementing nudges to AI/ML systems. Drawing upon the literature on nudge and recommendation agents/systems in IS, this study investigated the impact of two types of recommendation badges on user decision quality: AI nudge (e.g., Amazon's Choice) and non-AI nudge (e.g., Best Seller). We found that these two badges can lead to different user perceptions of transparency and thus affect the choice confidence of product selection. In addition, the effect of perceived transparency on choice confidence is contingent upon the mismatch/match between the recommendation and users' preferences, with perceived transparency exerting significantly higher influence on choice confidence in the preference match condition. We tested our research model using a randomized experiment and post-task survey data collected from 837 US-based college students with online shopping experience. This is the first empirical study examining the impact of AI nudging on user decision-making on e-commerce platforms and will contribute to the nudge literature and biased recommendation research in IS. The study also brings ethical implications to the use of AI/ML models and calls for careful oversight on delegating the power of nudging to AI in guiding online user behavior.

本研究探讨了人工智能推动对客户购买决策的影响。数字轻推是一种成熟的技术，用于以可预测的方式改变人们的行为。随着人工智能/机器学习（AI/ML）的快速发展以及“黑箱”算法在数字选择架构中的广泛集成，个性化的目标推动可以极大地影响个人和集体的行为，并导致意想不到的后果。AI助推是指人类将开发和实施助推外包给AI/ML系统的情况。借鉴IS中助推和推荐代理/系统的文献，本研究调查了两种类型的推荐徽章对用户决策质量的影响：人工智能助推（例如亚马逊的选择）和非人工智能助推（例如畅销书）。我们发现这两个徽章可以导致不同的用户对透明度的感知，从而影响产品选择的选择信心。此外，感知透明度对选择信心的影响取决于推荐与用户偏好之间的不匹配/匹配，在偏好匹配条件下，感知透明度对选择信心的影响显著更高。我们使用随机实验和从837名有网购经历的美国大学生中收集的任务后调查数据来测试我们的研究模型。这是第一个研究人工智能助推对电子商务平台用户决策影响的实证研究，将有助于is领域的助推文献和偏见推荐研究。该研究还对人工智能/机器学习模型的使用提出了伦理问题，并呼吁对委托人工智能指导在线用户行为的权力进行仔细监督。

{"title":"AI nudging and decision quality: Evidence from randomized experiments in online recommendation setting","authors":"Yuxiao Luo , Nanda Kumar , Adel Yazdanmehr","doi":"10.1016/j.dss.2025.114565","DOIUrl":"10.1016/j.dss.2025.114565","url":null,"abstract":"<div><div>This study explores the impacts of AI nudging on customer purchase decisions. Digital nudging is a well-established technique used to alter people's behaviors in a predictable way. With the rapid development of Artificial Intelligence/Machine Learning (AI/ML) and the widespread integration of the “black box” algorithm in the digital choice architecture, personalized targeting nudges can vastly influence individual and collective behaviors and lead to undesired consequences. AI nudge refers to the situation when human outsources developing and implementing nudges to AI/ML systems. Drawing upon the literature on nudge and recommendation agents/systems in IS, this study investigated the impact of two types of recommendation badges on user decision quality: AI nudge (e.g., <em>Amazon's Choice</em>) and non-AI nudge (e.g., <em>Best Seller</em>). We found that these two badges can lead to different user perceptions of transparency and thus affect the choice confidence of product selection. In addition, the effect of perceived transparency on choice confidence is contingent upon the mismatch/match between the recommendation and users' preferences, with perceived transparency exerting significantly higher influence on choice confidence in the preference match condition. We tested our research model using a randomized experiment and post-task survey data collected from 837 US-based college students with online shopping experience. This is the first empirical study examining the impact of AI nudging on user decision-making on e-commerce platforms and will contribute to the nudge literature and biased recommendation research in IS. The study also brings ethical implications to the use of AI/ML models and calls for careful oversight on delegating the power of nudging to AI in guiding online user behavior.</div></div>","PeriodicalId":55181,"journal":{"name":"Decision Support Systems","volume":"200 ","pages":"Article 114565"},"PeriodicalIF":6.8,"publicationDate":"2025-11-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145441667","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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

Leveraging large language models for enhanced process model comprehension 利用大型语言模型来增强流程模型的理解能力

IF 6.8 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Decision Support Systems

Pub Date : 2025-11-04 DOI: 10.1016/j.dss.2025.114563

Humam Kourani , Alessandro Berti , Jasmin Hennrich , Wolfgang Kratsch , Robin Weidlich , Chiao-Yun Li , Ahmad Arslan , Wil M.P. van der Aalst , Daniel Schuster

In Business Process Management (BPM), effectively comprehending process models is crucial yet poses significant challenges, particularly as organizations scale and processes become more complex. This paper introduces a novel framework utilizing the advanced capabilities of Large Language Models (LLMs) to enhance the comprehension of complex process models. We present different methods for abstracting business process models into a format accessible to LLMs, and we implement advanced prompting strategies specifically designed to optimize LLM performance within our framework. Additionally, we present a tool, AIPA, that implements our proposed framework and allows for conversational process querying. We evaluate our framework and tool through: i) an automatic evaluation comparing different LLMs, model abstractions, and prompting strategies; ii) a qualitative analysis assessing the ability to identify critical quality issues in process models; and iii) a user study designed to assess AIPA’s effectiveness comprehensively. Results demonstrate our framework’s ability to improve the comprehension and understanding of process models, pioneering new pathways for integrating AI technologies into the BPM field.

在业务流程管理（BPM）中，有效地理解流程模型是至关重要的，但也带来了重大挑战，特别是在组织规模扩大和流程变得更加复杂的情况下。本文介绍了一种利用大型语言模型（llm）的高级功能来增强对复杂过程模型的理解的新框架。我们提出了将业务流程模型抽象为LLM可访问的格式的不同方法，并实现了专门设计用于在我们的框架内优化LLM性能的高级提示策略。此外，我们还提供了一个工具AIPA，它实现了我们提出的框架，并允许会话过程查询。我们通过以下方式评估我们的框架和工具：i)比较不同llm、模型抽象和提示策略的自动评估；Ii)定性分析，评估识别过程模型中关键质量问题的能力；iii)旨在全面评估AIPA有效性的用户研究。结果表明，我们的框架能够提高对流程模型的理解和理解，为将AI技术集成到BPM领域开辟了新的途径。

{"title":"Leveraging large language models for enhanced process model comprehension","authors":"Humam Kourani , Alessandro Berti , Jasmin Hennrich , Wolfgang Kratsch , Robin Weidlich , Chiao-Yun Li , Ahmad Arslan , Wil M.P. van der Aalst , Daniel Schuster","doi":"10.1016/j.dss.2025.114563","DOIUrl":"10.1016/j.dss.2025.114563","url":null,"abstract":"<div><div>In Business Process Management (BPM), effectively comprehending process models is crucial yet poses significant challenges, particularly as organizations scale and processes become more complex. This paper introduces a novel framework utilizing the advanced capabilities of Large Language Models (LLMs) to enhance the comprehension of complex process models. We present different methods for abstracting business process models into a format accessible to LLMs, and we implement advanced prompting strategies specifically designed to optimize LLM performance within our framework. Additionally, we present a tool, AIPA, that implements our proposed framework and allows for conversational process querying. We evaluate our framework and tool through: i) an automatic evaluation comparing different LLMs, model abstractions, and prompting strategies; ii) a qualitative analysis assessing the ability to identify critical quality issues in process models; and iii) a user study designed to assess AIPA’s effectiveness comprehensively. Results demonstrate our framework’s ability to improve the comprehension and understanding of process models, pioneering new pathways for integrating AI technologies into the BPM field.</div></div>","PeriodicalId":55181,"journal":{"name":"Decision Support Systems","volume":"200 ","pages":"Article 114563"},"PeriodicalIF":6.8,"publicationDate":"2025-11-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145441665","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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

Artificial intelligence agents or human agents? Impact of online customer service agents on crowdfunding performance 人工智能代理还是人类代理？在线客服代理对众筹绩效的影响

IF 6.8 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Decision Support Systems

Pub Date : 2025-10-30 DOI: 10.1016/j.dss.2025.114562

Wei Wang , Yao Tong , Jian Mou

Although Artificial Intelligence (AI) agents are being increasingly deployed in crowdfunding platforms to address labor shortages, knowledge about their scope and limits is still limited. Across a secondary data analysis and three experiments (total N = 1027), we reveal that AI (vs. human) agents are more effective in reward-based (vs. donation-based) crowdfunding. This effect can be parallelly mediated by perceptions of warmth and competence, with AI agents evoking higher competence but weaker warmth perceptions. Importantly, anthropomorphic AI agents serve as an effective intervention to alleviate AI's negative impact on donation-based crowdfunding by enhancing warmth perceptions. Finally, we show that human agents outperform AI agents in boosting donation-based funding performance only for those with an interdependent versus independent self-construal. Overall, these findings expand the theoretical framework on AI applications in crowdfunding and offer actionable insights for fundraisers and platform operators to optimize agent deployment.

尽管人工智能（AI）代理越来越多地部署在众筹平台上，以解决劳动力短缺问题，但对其范围和限制的了解仍然有限。通过二次数据分析和三个实验（总N = 1027），我们发现人工智能（相对于人类）代理在基于奖励（相对于基于捐赠）的众筹中更有效。这种效应可以通过对温暖和能力的感知来平行调节，人工智能代理唤起更高的能力，但更弱的温暖感知。重要的是，拟人化人工智能代理通过增强温暖感知，可以有效地缓解人工智能对捐赠型众筹的负面影响。最后，我们表明，只有对于那些具有相互依赖与独立自我构造的人，人类代理在提高基于捐赠的融资绩效方面才优于人工智能代理。总的来说，这些发现扩展了人工智能在众筹中的应用的理论框架，并为筹款人和平台运营商优化代理部署提供了可操作的见解。

引用次数: 0