In organizational and social settings, people often work in teams, where conflicts are inevitable. Such conflicts can have a significant impact on team performance and cohesion. Thus, identifying and mitigating conflicts in team interactions is crucial to fostering productive activities and relationships. Prior conflict detection methods have relied mainly on classical machine learning models, focusing primarily on interpersonal (i.e., relationship) conflicts alone. However, conflicts are more varied, including not only interpersonal issues but also task conflicts that stem from differing perspectives or approaches towards the work itself. Detecting the distinct conflict types is crucial since they have differential impacts on team performance. In this research, we first develop a novel, synthetic team conflict dataset (NUSSWD) designed to distinguish between these conflict types and their varying levels. We then utilize advanced language models (BERT and RoBERTa) with multifeature embeddings as well as large language models (GPT-4o and Gemini 1.0 Pro) to propose a new approach to enhance team conflict detection and classification. The multifeature embeddings are also implemented for prior classical models. Experiments on four datasets (including ours) demonstrate the superior performance of our approach, with F1 scores of up to 0.95 for conflict type and 0.86 for conflict intensity, significantly improving upon the baselines. This sets a new benchmark in the field and offers novel insights into the detection and mitigation of team conflicts.

Medical question answer (QA) assistants respond to lay users' health-related queries by synthesizing information from multiple sources using natural language processing and related techniques. They can serve as vital tools to alleviate issues of misinformation, information overload, and complexity of medical language, thus addressing lay users' information needs while reducing the burden on healthcare professionals. QA systems, the engines of such assistants, have often used large language models (LLMs) or knowledge graphs (KG), though the approaches could be complementary. LLM-based QA systems excel at understanding complex questions and providing well-formed answers but are prone to factual mistakes. KG-based QA systems, which represent facts well, are mostly limited to answering short-answer questions with pre-created templates. While a few studies have used both LLM and KG for text-based QA, the approaches are still prone to incomplete or inaccurate answers. Extant QA systems also have limitations in terms of automation and performance. We address these challenges by designing a novel, automated disease QA system named Disease Guru-Long-Form Question Answer (DG-LFQA), which effectively utilizes both LLM and KG techniques through a joint reasoning approach to answer disease-related questions appropriate for lay users. Our evaluation of the system using a range of quality metrics demonstrates its efficacy over related baseline systems.

Organizations are increasingly crowdsourcing their tasks to unknown individual workers, i.e., solvers. Solvers' participation is critical to the success of crowdsourcing activities. However, challenges exist in attracting solvers to participate in crowdsourcing. In this regard, prior research has mainly investigated the influences of benefit factors on solvers’ intention to participate in crowdsourcing. Thus, there is a lack of understanding of the cost factors that influence actual participation behavior, in conjunction with the benefits. Additionally, the role of trust in the cost-benefit analysis remains to be explored. Motivated thus, based on social exchange theory and context-related literature, we develop a model to explain the impacts of benefit and cost factors as well as trust on solver participation behavior in crowdsourcing. The model was tested using survey and archival data from 156 solvers on a large crowdsourcing platform. As hypothesized, monetary reward, skill enhancement, work autonomy, enjoyment, and trust were found to positively affect solvers’ participation in crowdsourcing, while cognitive effort negatively affects their participation. In addition, it was found that monetary reward positively affects trust (trust partially mediates its effect on participation behavior), while loss of knowledge power negatively affects trust. The theoretical contributions and practical implications of the study are discussed.

Much attention is currently being paid in both the academic and practitioner literatures to the value that organisations could create through the use of big data and business analytics (Gillon et a...

There is significant interest among marketers and academics to understand how to segment online consumers to better fulfill their needs. Previous literature on brick-and-mortar shopping has advocated demographic variables as simple yet valuable indicators to understand consumers and segment them accordingly. However, research in the online shopping context has produced mixed findings about the effects of demographics, which limit their utility to online merchants. As an attempt to address the ambiguity, the study proposes a more comprehensive approach to investigate the effects of demographic variables in the online shopping context. This is done by considering the demographic variables in combinations and using clickstream data to more accurately derive online shoppers’ surfing behaviour for segmentation purposes. Following this approach, our study investigates the effects of the demographic variables of gender, age, income, and education, based on the theories of media naturalness and consumer trait and involvement. The results reveal various effects of age, income, and education on online consumers’ needs being reflected in their store visit strategies. Implications are suggested for e-commerce research and practice.

Subsidized ICT access is often provided in less-developed regions of the world to bridge the digital divide. While such efforts can be helpful, their effectiveness depends on targeted users’ willingness to utilize the ICT provided. To better understand the phenomenon from users’ perspective, this study examines the factors influencing individual’s intention and behavior of using public internet kiosks in Mauritius. Having witnessed rapid growth in ICT recently, Mauritius is an appropriate context for this study. Findings indicate that self efficacy, perceived ease of use, perceived usefulness, subjective norm, and perceived behavioral control are significant in explaining ICT use. Although these factors have been studied in the developed world, their perception and assessment in the developing world differ and deserve more attention. This study adds to the limited empirical research in this part of the world. Implications for research and suggestions for encouraging ICT use in such regions are discussed.

Knowledge sharing, which is critical for the strategic utilization of knowledge resources for the benefit of an organization, can only take place when both knowledge contribution and knowledge seeking exist. However, most previous research has focused on only one side of this process – knowledge contribution motivations. This is despite the fact that various barriers to knowledge seeking and reuse exist, such as the effort required to seek relevant knowledge and the cost of future obligation. In overcoming such barriers, norms related to collaboration are considered to be important. However, little is known of how these norms operate in conjunction with other antecedents to influence individuals' knowledge seeking behavior. Addressing the knowledge gap, this study explores how collaborative norms in an organization impact knowledge seeking with regard to a common knowledge management system type – the electronic knowledge repository (EKR). For this purpose, we have developed a model and tested it through a survey of EKR users in knowledge-intensive organizations. Our results indicate that collaborative norms positively impact individuals' knowledge seeking behavior through EKRs, both directly and through reducing the negative effect of future obligation on seeking. However, collaborative norms could also undermine the positive impact of perceived usefulness on knowledge seeking behavior. We identify other antecedents of knowledge seeking such as knowledge growth, resource-facilitating conditions, and self-efficacy. Implications for research and knowledge sharing practice are discussed.

Metrics are essential for the advancement of research and practice in an area. In knowledge management (KM), the process of measurement and development of metrics is made complex by the intangible nature of the knowledge asset. Further, the lack of standards for KM business metrics and the relative infancy of research on KM metrics points to a need for research in this area. This article reviews KM metrics for research and practice, and identifies areas where there is a gap in our understanding. It classifies existing research based on the units of evaluation such as user of knowledge management systems (KMS), KMS project, KM process, KM initiative, and organization as a whole. The article concludes by suggesting avenues for future research on KM and KMS metrics based on the gaps identified.

Abstract Knowledge is a critical resource that can help organizations to sustain strategic advantage in competitive environments. Organizations in Asia and elsewhere are turning to knowledge management (KM) initiatives and technologies to leverage their knowledge resources. As a key component of KM initiatives, electronic knowledge repositories (EKRs) are deployed by organizations to store codified knowledge for future reuse. Although EKRs have been used for some time, there is a lack of understanding of what motivate employees' usage of an EKR. This study formulates and empirically tests a theoretical model relating potential antecedents to EKR usage for knowledge seeking. The model was operationalized and the survey instrument subjected to a conceptual validation process. The survey was administered to 160 knowledge professionals in public‐sector organizations in Singapore who had accessed EKRs in the course of their work. Results reveal that perceived output quality directly affects EKR usage for knowledge seeking. Further, resource availability affects EKR usage for knowledge seeking particularly when task tacitness is low and incentives affect EKR usage particularly when task interdependence is high. Implications of these results for further research and improving EKR implementation are discussed.

We have developed a generalized alphanumeric character extraction algorithm that can efficiently and accurately locate and extract characters from complex scene images. A scene image may be complex due to the following reasons: (1) the characters are embedded in an image with other objects, such as structural bars, company logos and smears; (2) the characters may be painted or printed in any color including white, and the background color may differ only slightly from that of the characters; (3) the font, size and format of the characters may be different; and (4) the lighting may be uneven. The main contribution of this research is that it permits the quick and accurate extraction of characters in a complex scene. A coarse search technique is used to locate potential characters, and then a fine grouping technique is used to extract characters accurately. Several additional techniques in the postprocessing phase eliminate spurious as well as overlapping characters. Experimental results of segmenting characters written on cargo container surfaces show that the system is feasible under real-life constraints. The program has been installed as part of a vision system which verifies container codes on vehicles passing through the Port of Singapore.