Decision Support Systems最新文献

Augmenting micro-moment recommendations with group and serendipity perspectives

IF 6.7 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Decision Support Systems

Pub Date : 2025-04-17 DOI: 10.1016/j.dss.2025.114454

Yi-Ling Lin , Yu-Xiang Zheng , Yi-Cheng Ku

With the pervasive integration of internet and mobile services, mobile devices have become integral to daily life. The concept of micro-moments, characterized by immediate intent within specific contexts, underscores the importance of timely and relevant information. Traditional RS, though effective in mitigating information overload, often fall short in addressing the dynamic and context-specific needs inherent in micromoments. This study investigates the enhancement of MMRS by incorporating group dynamics and serendipity, aiming to improve recommendation quality and user satisfaction. The research explores two primary objectives: the feasibility of a groupaugmented MMRS and the integration of serendipity into MMRS. Utilizing a design science approach, we conducted a two-phase iterative design involving preliminary studies and field experiments. The results indicate that integrating group recommendations based on social relationships and serendipity mechanisms significantly enhances user satisfaction and behavioral intentions. Close groups exhibited higher satisfaction and engagement compared to acquainted groups, emphasizing the importance of social relationships in recommendation strategies. Moreover, the serendipity mechanism, characterized by relevance, novelty, and unexpectedness, successfully mitigates overspecialization, enriching user experience by introducing unexpected yet relevant recommendations. Our findings contribute to the theoretical understanding of MMRS by demonstrating the viability of combining group dynamics and serendipity to cater to the evolving needs of mobile users in micro-moments. Practically, the study provides valuable insights for developing RS that are adaptive, context-aware, and capable of delivering engaging and satisfying user experiences. Future research should expand on diverse social relationships and longterm evaluations to refine the application of these mechanisms in various domains.

{"title":"Augmenting micro-moment recommendations with group and serendipity perspectives","authors":"Yi-Ling Lin , Yu-Xiang Zheng , Yi-Cheng Ku","doi":"10.1016/j.dss.2025.114454","DOIUrl":"10.1016/j.dss.2025.114454","url":null,"abstract":"<div><div>With the pervasive integration of internet and mobile services, mobile devices have become integral to daily life. The concept of micro-moments, characterized by immediate intent within specific contexts, underscores the importance of timely and relevant information. Traditional RS, though effective in mitigating information overload, often fall short in addressing the dynamic and context-specific needs inherent in micromoments. This study investigates the enhancement of MMRS by incorporating group dynamics and serendipity, aiming to improve recommendation quality and user satisfaction. The research explores two primary objectives: the feasibility of a groupaugmented MMRS and the integration of serendipity into MMRS. Utilizing a design science approach, we conducted a two-phase iterative design involving preliminary studies and field experiments. The results indicate that integrating group recommendations based on social relationships and serendipity mechanisms significantly enhances user satisfaction and behavioral intentions. Close groups exhibited higher satisfaction and engagement compared to acquainted groups, emphasizing the importance of social relationships in recommendation strategies. Moreover, the serendipity mechanism, characterized by relevance, novelty, and unexpectedness, successfully mitigates overspecialization, enriching user experience by introducing unexpected yet relevant recommendations. Our findings contribute to the theoretical understanding of MMRS by demonstrating the viability of combining group dynamics and serendipity to cater to the evolving needs of mobile users in micro-moments. Practically, the study provides valuable insights for developing RS that are adaptive, context-aware, and capable of delivering engaging and satisfying user experiences. Future research should expand on diverse social relationships and longterm evaluations to refine the application of these mechanisms in various domains.</div></div>","PeriodicalId":55181,"journal":{"name":"Decision Support Systems","volume":"193 ","pages":"Article 114454"},"PeriodicalIF":6.7,"publicationDate":"2025-04-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143854753","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

High efficiency or easy troubleshooting? Human use of autonomous Mobile healthcare robots

IF 6.7 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Decision Support Systems

Pub Date : 2025-04-15 DOI: 10.1016/j.dss.2025.114453

Tzu-Ling Huang , Gen-Yih Liao , Alan R. Dennis , Ching-I Teng

Among modern information technologies, robots help reduce the effort employees expend on tasks that are repetitive and physically demanding. When helping employees, robots may be required to display enhanced efficiency, but such a design can also increase employees' effort required for operational troubleshooting. It is not yet known whether effort saving (i.e., increasing nurses' time and energy saved) or reduced troubleshooting effort (i.e., reducing nurses' time and energy costs) is more important for enhancing users' perception that the robot is performing optimally (user-perceived robot performance) and positive workplace outcomes. This hinders robot providers from making optimal decisions on robot design. In a healthcare context, nurses comprise the largest workforce and thus we examined autonomous mobile robots that help nurses carry heavy equipment and materials to and from operating rooms to meet the demand of surgical operations. Hence, this study examined the relative influence of increased effort saving versus reduced troubleshooting effort on perceived robot performance, patient care, and nurse health. We collected responses from 331 operating room nurses through two waves of surveys. Compared with reduced troubleshooting effort, effort saving effectively increased nurse-perceived robot performance, patient care and nurse health, from 39 % to 77 %. Nurses' greater professional experience reduced the negative influence of troubleshooting effort on perceived robot performance. These findings showed that designing information technologies for high efficiency is more important than designing for ease of troubleshooting. This research contributes to decision-making of robot makers and hospitals by indicating that the effects of benefits and costs may depend on the features of users.

在现代信息技术中，机器人有助于减少员工在重复性和体力要求高的任务上所花费的精力。在帮助员工时，机器人可能需要显示出更高的效率，但这样的设计也会增加员工排除操作故障所需的精力。对于提高用户对机器人最佳性能的感知（用户感知的机器人性能）和积极的工作场所结果而言，究竟是省力（即增加护士节省的时间和精力）还是减少故障排除的工作量（即降低护士的时间和精力成本）更为重要，目前还不得而知。这就阻碍了机器人供应商在机器人设计方面做出最佳决策。在医疗保健领域，护士是最大的劳动力，因此我们研究了帮助护士搬运重型设备和材料往返手术室以满足手术需求的自主移动机器人。因此，本研究考察了增加省力与减少故障排除工作对感知机器人性能、患者护理和护士健康的相对影响。我们通过两轮调查收集了 331 名手术室护士的反馈。与减少故障排除工作相比，省力有效地提高了护士对机器人性能、病人护理和护士健康的感知，从 39% 提高到 77%。护士的专业经验越丰富，就越能减少故障排除工作对感知机器人性能的负面影响。这些研究结果表明，设计出高效率的信息技术比设计出易于排除故障的信息技术更为重要。这项研究表明，效益和成本的影响可能取决于用户的特点，从而有助于机器人制造商和医院的决策。

{"title":"High efficiency or easy troubleshooting? Human use of autonomous Mobile healthcare robots","authors":"Tzu-Ling Huang , Gen-Yih Liao , Alan R. Dennis , Ching-I Teng","doi":"10.1016/j.dss.2025.114453","DOIUrl":"10.1016/j.dss.2025.114453","url":null,"abstract":"<div><div>Among modern information technologies, robots help reduce the effort employees expend on tasks that are repetitive and physically demanding. When helping employees, robots may be required to display enhanced efficiency, but such a design can also increase employees' effort required for operational troubleshooting. It is not yet known whether effort saving (i.e., increasing nurses' time and energy saved) or reduced troubleshooting effort (i.e., reducing nurses' time and energy costs) is more important for enhancing users' perception that the robot is performing optimally (user-perceived robot performance) and positive workplace outcomes. This hinders robot providers from making optimal decisions on robot design. In a healthcare context, nurses comprise the largest workforce and thus we examined autonomous mobile robots that help nurses carry heavy equipment and materials to and from operating rooms to meet the demand of surgical operations. Hence, this study examined the relative influence of increased effort saving versus reduced troubleshooting effort on perceived robot performance, patient care, and nurse health. We collected responses from 331 operating room nurses through two waves of surveys. Compared with reduced troubleshooting effort, effort saving effectively increased nurse-perceived robot performance, patient care and nurse health, from 39 % to 77 %. Nurses' greater professional experience reduced the negative influence of troubleshooting effort on perceived robot performance. These findings showed that designing information technologies for high efficiency is more important than designing for ease of troubleshooting. This research contributes to decision-making of robot makers and hospitals by indicating that the effects of benefits and costs may depend on the features of users.</div></div>","PeriodicalId":55181,"journal":{"name":"Decision Support Systems","volume":"193 ","pages":"Article 114453"},"PeriodicalIF":6.7,"publicationDate":"2025-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143842455","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

Enhancing return forecasting using LSTM with agent-based synthetic data

IF 6.7 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Decision Support Systems

Pub Date : 2025-04-14 DOI: 10.1016/j.dss.2025.114452

Lijian Wei , Sihang Chen , Junqin Lin , Lei Shi

Financial markets, as complex adaptive systems, are characterized by historical data limitations, inherent evolution and non-stationarity, which challenge the effectiveness of deep learning models such as Long Short-Term Memory (LSTM). We address these challenges by generating synthetic data using Agent-Based Modeling (ABM) to simulate complex market conditions through “what-if” scenarios. Our method comprises three steps: (i) pre-training the LSTM model on historical data, (ii) generating synthetic data with the ABM using “what-if” scenarios, and (iii) fine-tuning the pre-trained LSTM with ABM-generated synthetic data. The results show that ABM-generated data significantly improve model performance across various statistical and economic metrics and are robust to diverse market environments, model architectures, and data frequencies. Our primary contribution is modeling the properties of complex adaptive systems with ABM-generated data, highlighting the need for new complex scenarios to better simulate future market conditions that are distinct from historical trends. We explore the potential of ABM in generating unique synthetic data, offering a framework to address the challenges imposed by the complex adaptive system properties of financial markets, particularly, improving the discriminative ability of forecasting models such as the LSTM model.

{"title":"Enhancing return forecasting using LSTM with agent-based synthetic data","authors":"Lijian Wei , Sihang Chen , Junqin Lin , Lei Shi","doi":"10.1016/j.dss.2025.114452","DOIUrl":"10.1016/j.dss.2025.114452","url":null,"abstract":"<div><div>Financial markets, as complex adaptive systems, are characterized by historical data limitations, inherent evolution and non-stationarity, which challenge the effectiveness of deep learning models such as Long Short-Term Memory (LSTM). We address these challenges by generating synthetic data using Agent-Based Modeling (ABM) to simulate complex market conditions through “what-if” scenarios. Our method comprises three steps: (i) pre-training the LSTM model on historical data, (ii) generating synthetic data with the ABM using “what-if” scenarios, and (iii) fine-tuning the pre-trained LSTM with ABM-generated synthetic data. The results show that ABM-generated data significantly improve model performance across various statistical and economic metrics and are robust to diverse market environments, model architectures, and data frequencies. Our primary contribution is modeling the properties of complex adaptive systems with ABM-generated data, highlighting the need for new complex scenarios to better simulate future market conditions that are distinct from historical trends. We explore the potential of ABM in generating unique synthetic data, offering a framework to address the challenges imposed by the complex adaptive system properties of financial markets, particularly, improving the discriminative ability of forecasting models such as the LSTM model.</div></div>","PeriodicalId":55181,"journal":{"name":"Decision Support Systems","volume":"193 ","pages":"Article 114452"},"PeriodicalIF":6.7,"publicationDate":"2025-04-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143834655","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

Can comments and dialogues make sense? The effect of two-way interactions on sales and followers in live streaming commerce 评论和对话有意义吗？直播商业中双向互动对销售和粉丝的影响

IF 6.7 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Decision Support Systems

Pub Date : 2025-04-05 DOI: 10.1016/j.dss.2025.114451

Xiaoping Lang , Sheng Lin , Xiangyang Ma , Tieshan Li

This research employs interaction ritual model to explore the two-way interaction between streamers and viewers on a live streaming commerce platform. To be specific, the study investigates the impact of viewers' real-time comments and streamer's dialogue on product sales and follower growth, using minute-level data for detailed analysis. The results show that real-time comments exhibit a nonlinear inverted U-shaped relationship with product sales and increment in followers. Such relationship indicates that the “overloaded comments” during live streams could cause a problem on live streaming commerce platforms. In addition, we find that streamer's dialogue has a positive effect on product sales and can mitigate the inverted U-shaped relationship between real-time comments and product sales. Furthermore, we identify that streamer's dialogue has a delayed effect on product sales, with sales typically occurring 2.5 to 5 min after the dialogue ends. The findings provide valuable guidance for optimizing the management of live streaming commerce platforms.

{"title":"Can comments and dialogues make sense? The effect of two-way interactions on sales and followers in live streaming commerce","authors":"Xiaoping Lang , Sheng Lin , Xiangyang Ma , Tieshan Li","doi":"10.1016/j.dss.2025.114451","DOIUrl":"10.1016/j.dss.2025.114451","url":null,"abstract":"<div><div>This research employs interaction ritual model to explore the two-way interaction between streamers and viewers on a live streaming commerce platform. To be specific, the study investigates the impact of viewers' real-time comments and streamer's dialogue on product sales and follower growth, using minute-level data for detailed analysis. The results show that real-time comments exhibit a nonlinear inverted U-shaped relationship with product sales and increment in followers. Such relationship indicates that the “overloaded comments” during live streams could cause a problem on live streaming commerce platforms. In addition, we find that streamer's dialogue has a positive effect on product sales and can mitigate the inverted U-shaped relationship between real-time comments and product sales. Furthermore, we identify that streamer's dialogue has a delayed effect on product sales, with sales typically occurring 2.5 to 5 min after the dialogue ends. The findings provide valuable guidance for optimizing the management of live streaming commerce platforms.</div></div>","PeriodicalId":55181,"journal":{"name":"Decision Support Systems","volume":"193 ","pages":"Article 114451"},"PeriodicalIF":6.7,"publicationDate":"2025-04-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143800434","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

Remote work in the metaverse: The impact of gamification and online social connectedness on job satisfaction

IF 6.7 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Decision Support Systems

Pub Date : 2025-04-04 DOI: 10.1016/j.dss.2025.114447

Khadija Ali Vakeel , Saurav Chakraborty , Lamont Black

This study explores the potential of the metaverse in enhancing job satisfaction for remote employees. With the increasing shift towards remote work, firms are investing more in the metaverse to create dynamic and immersive digital environments. Drawing upon media richness theory, we investigate the roles of gamification and online social connectedness within the metaverse, which are crucial factors shaping employee job satisfaction. Survey results show that remote employees perceive higher gamification in the metaverse but have similar online social connectedness to video conferencing platforms. Contrary to linear assumptions, polynomial regression analysis reveals an intriguing S-shaped relationship between gamification and job satisfaction in the metaverse, highlighting an optimal threshold beyond which excessive gamification in the metaverse may diminish job satisfaction. Additionally, online social connectedness in the metaverse significantly strengthens job satisfaction. This study contributes to theoretical and practical knowledge by expanding the application of the metaverse to remote work. Our findings provide valuable guidance for firms navigating the evolving landscape of remote work and technology adoption, paving the way for more engaging and satisfying remote work experiences.

{"title":"Remote work in the metaverse: The impact of gamification and online social connectedness on job satisfaction","authors":"Khadija Ali Vakeel , Saurav Chakraborty , Lamont Black","doi":"10.1016/j.dss.2025.114447","DOIUrl":"10.1016/j.dss.2025.114447","url":null,"abstract":"<div><div>This study explores the potential of the metaverse in enhancing job satisfaction for remote employees. With the increasing shift towards remote work, firms are investing more in the metaverse to create dynamic and immersive digital environments. Drawing upon media richness theory, we investigate the roles of gamification and online social connectedness within the metaverse, which are crucial factors shaping employee job satisfaction. Survey results show that remote employees perceive higher gamification in the metaverse but have similar online social connectedness to video conferencing platforms. Contrary to linear assumptions, polynomial regression analysis reveals an intriguing S-shaped relationship between gamification and job satisfaction in the metaverse, highlighting an optimal threshold beyond which excessive gamification in the metaverse may diminish job satisfaction. Additionally, online social connectedness in the metaverse significantly strengthens job satisfaction. This study contributes to theoretical and practical knowledge by expanding the application of the metaverse to remote work. Our findings provide valuable guidance for firms navigating the evolving landscape of remote work and technology adoption, paving the way for more engaging and satisfying remote work experiences.</div></div>","PeriodicalId":55181,"journal":{"name":"Decision Support Systems","volume":"193 ","pages":"Article 114447"},"PeriodicalIF":6.7,"publicationDate":"2025-04-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143790851","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

Unravelling the effects of two inconsistencies on online review helpfulness: Evidence from TripAdvisor

IF 6.7 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Decision Support Systems

Pub Date : 2025-04-04 DOI: 10.1016/j.dss.2025.114450

Dujuan Wang , Qianyang Xia , Yi Feng , T.C.E. Cheng

Facing the challenge of information overload, some travel websites have introduced systems for travelers to vote on helpful reviews, prompting researchers to focus on the determinants of review helpfulness. While evaluations from multiple reviews may provide travelers with more perspectives, inconsistent information within the reviews may cause confusion. Studies exploring the effects of multiple inconsistencies on review helpfulness are relatively rare. Grounded in the heuristic-systematic model, we explore the relationships between systematic cues, i.e., review and rating inconsistencies, and review helpfulness. We also investigate how reviewer expertise and hotel rank moderate these inconsistency-helpfulness links, serving as heuristic cues. Applied to a real-world hotel dataset collected from TripAdvisor, our findings show that review inconsistency negatively influences review helpfulness, while rating inconsistency positively affects it. Furthermore, we find that reviewer expertise negatively moderates the review and rating inconsistency-helpfulness links, while hotels that rank low positively moderate both links. These findings offer both theoretical insights for research and practical implications for consumers, reviewers, and platform managers.

{"title":"Unravelling the effects of two inconsistencies on online review helpfulness: Evidence from TripAdvisor","authors":"Dujuan Wang , Qianyang Xia , Yi Feng , T.C.E. Cheng","doi":"10.1016/j.dss.2025.114450","DOIUrl":"10.1016/j.dss.2025.114450","url":null,"abstract":"<div><div>Facing the challenge of information overload, some travel websites have introduced systems for travelers to vote on helpful reviews, prompting researchers to focus on the determinants of review helpfulness. While evaluations from multiple reviews may provide travelers with more perspectives, inconsistent information within the reviews may cause confusion. Studies exploring the effects of multiple inconsistencies on review helpfulness are relatively rare. Grounded in the heuristic-systematic model, we explore the relationships between systematic cues, i.e., review and rating inconsistencies, and review helpfulness. We also investigate how reviewer expertise and hotel rank moderate these inconsistency-helpfulness links, serving as heuristic cues. Applied to a real-world hotel dataset collected from TripAdvisor, our findings show that review inconsistency negatively influences review helpfulness, while rating inconsistency positively affects it. Furthermore, we find that reviewer expertise negatively moderates the review and rating inconsistency-helpfulness links, while hotels that rank low positively moderate both links. These findings offer both theoretical insights for research and practical implications for consumers, reviewers, and platform managers.</div></div>","PeriodicalId":55181,"journal":{"name":"Decision Support Systems","volume":"193 ","pages":"Article 114450"},"PeriodicalIF":6.7,"publicationDate":"2025-04-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143821319","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

Dynamic model selection in enterprise forecasting systems using sequence modeling

IF 6.7 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Decision Support Systems

Pub Date : 2025-03-29 DOI: 10.1016/j.dss.2025.114439

Jinhang Jiang , Kiran Kumar Bandeli , Karthik Srinivasan

Enterprise forecasting systems often involve modeling a large scale of heterogeneous time series using a pool of candidate algorithms, such as in the case of simultaneous sales forecasts of thousands of stock-keeping units. In such cases, it can be advantageous to automatically monitor and replace algorithms for each time series. We introduce TimeSpeaks, a framework that adapts sequence modeling in natural language processing to the problem of dynamic model selection in enterprise forecasting. We instantiate our framework using sequential (BiLSTM) and transformer-based (TimeXer) deep learning models to learn the temporal dependencies between candidate algorithms. We compare the performance of our framework with state-of-the-art forecasting models using two public benchmarking datasets. We further demonstrate its practical application on two retail case studies, while comparing them to alternative model selection scenarios. TimeSpeaks has superior predictive performance and scalability across different scenarios and datasets. Its ability to adapt to evolving data patterns and its minimal reliance on exogenous information make TimeSpeaks a suitable framework for large-scale enterprise forecasting applications.

{"title":"Dynamic model selection in enterprise forecasting systems using sequence modeling","authors":"Jinhang Jiang , Kiran Kumar Bandeli , Karthik Srinivasan","doi":"10.1016/j.dss.2025.114439","DOIUrl":"10.1016/j.dss.2025.114439","url":null,"abstract":"<div><div>Enterprise forecasting systems often involve modeling a large scale of heterogeneous time series using a pool of candidate algorithms, such as in the case of simultaneous sales forecasts of thousands of stock-keeping units. In such cases, it can be advantageous to automatically monitor and replace algorithms for each time series. We introduce TimeSpeaks, a framework that adapts sequence modeling in natural language processing to the problem of dynamic model selection in enterprise forecasting. We instantiate our framework using sequential (BiLSTM) and transformer-based (TimeXer) deep learning models to learn the temporal dependencies between candidate algorithms. We compare the performance of our framework with state-of-the-art forecasting models using two public benchmarking datasets. We further demonstrate its practical application on two retail case studies, while comparing them to alternative model selection scenarios. TimeSpeaks has superior predictive performance and scalability across different scenarios and datasets. Its ability to adapt to evolving data patterns and its minimal reliance on exogenous information make TimeSpeaks a suitable framework for large-scale enterprise forecasting applications.</div></div>","PeriodicalId":55181,"journal":{"name":"Decision Support Systems","volume":"193 ","pages":"Article 114439"},"PeriodicalIF":6.7,"publicationDate":"2025-03-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143748335","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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}

引用次数: 0

ESG-KIBERT: A new paradigm in ESG evaluation using NLP and industry-specific customization

IF 6.7 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Decision Support Systems

Pub Date : 2025-03-29 DOI: 10.1016/j.dss.2025.114440

Haein Lee , Jang Hyun Kim , Hae Sun Jung

This study presents a significant advancement in Environmental, Social, Governance (ESG) evaluation by addressing critical gaps in transparency, consistency, and industry-specific relevance. The ESG-Keyword integrated bidirectional encoder representations from transformers (ESG-KIBERT) model, developed using advanced natural language processing (NLP) techniques, enhances ESG classification performance and sets a new standard for automated ESG analysis. With robust performance metrics, it supports reliable and consistent assessments across industries. Additionally, incorporating Sustainability Accounting Standards Board's materiality map offers a customized evaluation framework that accounts for industry-specific factors affecting corporate sustainability. Furthermore, the integration of sentiment analysis enriches ESG evaluations by capturing market and investor perceptions, contributing to a more transparent assessment. This study offers a comprehensive, standardized ESG evaluation framework that improves both the methodological rigor and practical utility of corporate sustainability assessments, enabling more informed decision-making for companies, investors and policymakers.

{"title":"ESG-KIBERT: A new paradigm in ESG evaluation using NLP and industry-specific customization","authors":"Haein Lee , Jang Hyun Kim , Hae Sun Jung","doi":"10.1016/j.dss.2025.114440","DOIUrl":"10.1016/j.dss.2025.114440","url":null,"abstract":"<div><div>This study presents a significant advancement in Environmental, Social, Governance (ESG) evaluation by addressing critical gaps in transparency, consistency, and industry-specific relevance. The ESG-Keyword integrated bidirectional encoder representations from transformers (ESG-KIBERT) model, developed using advanced natural language processing (NLP) techniques, enhances ESG classification performance and sets a new standard for automated ESG analysis. With robust performance metrics, it supports reliable and consistent assessments across industries. Additionally, incorporating Sustainability Accounting Standards Board's materiality map offers a customized evaluation framework that accounts for industry-specific factors affecting corporate sustainability. Furthermore, the integration of sentiment analysis enriches ESG evaluations by capturing market and investor perceptions, contributing to a more transparent assessment. This study offers a comprehensive, standardized ESG evaluation framework that improves both the methodological rigor and practical utility of corporate sustainability assessments, enabling more informed decision-making for companies, investors and policymakers.</div></div>","PeriodicalId":55181,"journal":{"name":"Decision Support Systems","volume":"193 ","pages":"Article 114440"},"PeriodicalIF":6.7,"publicationDate":"2025-03-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143738406","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

Predicting stock price movement using social network analytics: Posts are sometimes less useful

IF 6.7 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Decision Support Systems

Pub Date : 2025-03-24 DOI: 10.1016/j.dss.2025.114438

Wanyun Li , Alvin Chung Man Leung , Ka Wai Choi (Stanley) , Shuk Ying Ho

Contemporary research has leveraged social network data as a predictive tool for decision-making process in the capital market. Yet, its effectiveness may be compromised by social contagion. This study addresses this problem by introducing conversation-level measures that capture how interactions among investors affect market predictions. Drawing on social contagion theory, we identified three conversation conditions—argument similarity, sentiment similarity, and conversation size—and examined their association with the likelihood of abrupt stock price changes, which indicate a loss of collective wisdom. Our analysis of 18 million StockTwits posts for 859 Initial Public Offerings (2008–2017) reveals that conversations with highly similar arguments, highly similar sentiments, and larger size are significantly associated with an increased likelihood of abrupt stock price changes in the subsequent week. Moreover, out-of-sample tests confirm that monitoring conversational dynamics enhances the predictive power of social network analytics, offering valuable guidance for investors and practitioners. Our study extends the theoretical framework of social contagion by highlighting the importance of the conversation level and provides practical recommendations for refining trading strategies based on social media data.

{"title":"Predicting stock price movement using social network analytics: Posts are sometimes less useful","authors":"Wanyun Li , Alvin Chung Man Leung , Ka Wai Choi (Stanley) , Shuk Ying Ho","doi":"10.1016/j.dss.2025.114438","DOIUrl":"10.1016/j.dss.2025.114438","url":null,"abstract":"<div><div>Contemporary research has leveraged social network data as a predictive tool for decision-making process in the capital market. Yet, its effectiveness may be compromised by social contagion. This study addresses this problem by introducing conversation-level measures that capture how interactions among investors affect market predictions. Drawing on social contagion theory, we identified three conversation conditions—argument similarity, sentiment similarity, and conversation size—and examined their association with the likelihood of abrupt stock price changes, which indicate a loss of collective wisdom. Our analysis of 18 million StockTwits posts for 859 Initial Public Offerings (2008–2017) reveals that conversations with highly similar arguments, highly similar sentiments, and larger size are significantly associated with an increased likelihood of abrupt stock price changes in the subsequent week. Moreover, out-of-sample tests confirm that monitoring conversational dynamics enhances the predictive power of social network analytics, offering valuable guidance for investors and practitioners. Our study extends the theoretical framework of social contagion by highlighting the importance of the conversation level and provides practical recommendations for refining trading strategies based on social media data.</div></div>","PeriodicalId":55181,"journal":{"name":"Decision Support Systems","volume":"192 ","pages":"Article 114438"},"PeriodicalIF":6.7,"publicationDate":"2025-03-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143725238","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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}

引用次数: 0

To disclose or not? The impact of prosocial behavior disclosure on the attainment of social capital on social networking sites

IF 6.7 1区计算机科学 Q1 COMPUTER SCIENCE, ARTIFICIAL INTELLIGENCE

Decision Support Systems

Pub Date : 2025-03-21 DOI: 10.1016/j.dss.2025.114437

Jiayuan Zhang , Koray Özpolat , Gulver Karamemis , Dara Schniederjans

While some donors and volunteers do not publicize their prosocial behaviors because of humility, many others fear that disclosing their prosocial behaviors may be perceived as bragging. With the rise of social networking sites (SNSs), this has become an essential issue with important business implications. As more companies encourage employees to volunteer a small portion of their work time and match their charitable contributions, disclosing these prosocial acts on social media platforms has become more common. Building upon social capital theory, we apply a mixed-method approach to investigate the relationship between the disclosure of prosocial behaviors and the attainment of social capital on SNSs. Our first exploratory study applies qualitative interviews to explore the factors that moderate the relationship between the disclosure of prosocial behaviors and the attainment of social capital. Our second study utilizes a randomized online experiment in the U.S. to test the causal effect of prosocial behavior disclosure on social capital attainment online, as well as two moderators of this relationship. A post-hoc replication study of our experiment is conducted in China. We find that the disclosure of prosocial behavior increases relational and structural social capital on SNSs but find no evidence of the impact of the disclosure of prosocial behavior on cognitive social capital. The effect becomes stronger when one's prosocial behavior is disclosed by others (rather than by oneself) in the U.S. sample. Our findings inform SNSs users to make informed decisions regarding disclosing prosocial behaviors to attain structural and relational social capital. Businesses encouraging their employees to donate/volunteer and charities on the receiving end could also benefit from our findings.

{"title":"To disclose or not? The impact of prosocial behavior disclosure on the attainment of social capital on social networking sites","authors":"Jiayuan Zhang , Koray Özpolat , Gulver Karamemis , Dara Schniederjans","doi":"10.1016/j.dss.2025.114437","DOIUrl":"10.1016/j.dss.2025.114437","url":null,"abstract":"<div><div>While some donors and volunteers do not publicize their prosocial behaviors because of humility, many others fear that disclosing their prosocial behaviors may be perceived as bragging. With the rise of social networking sites (SNSs), this has become an essential issue with important business implications. As more companies encourage employees to volunteer a small portion of their work time and match their charitable contributions, disclosing these prosocial acts on social media platforms has become more common. Building upon social capital theory, we apply a mixed-method approach to investigate the relationship between the disclosure of prosocial behaviors and the attainment of social capital on SNSs. Our first exploratory study applies qualitative interviews to explore the factors that moderate the relationship between the disclosure of prosocial behaviors and the attainment of social capital. Our second study utilizes a randomized online experiment in the U.S. to test the causal effect of prosocial behavior disclosure on social capital attainment online, as well as two moderators of this relationship. A post-hoc replication study of our experiment is conducted in China. We find that the disclosure of prosocial behavior increases relational and structural social capital on SNSs but find no evidence of the impact of the disclosure of prosocial behavior on cognitive social capital. The effect becomes stronger when one's prosocial behavior is disclosed by others (rather than by oneself) in the U.S. sample. Our findings inform SNSs users to make informed decisions regarding disclosing prosocial behaviors to attain structural and relational social capital. Businesses encouraging their employees to donate/volunteer and charities on the receiving end could also benefit from our findings.</div></div>","PeriodicalId":55181,"journal":{"name":"Decision Support Systems","volume":"192 ","pages":"Article 114437"},"PeriodicalIF":6.7,"publicationDate":"2025-03-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143706230","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