3633 research outputs found
Sort by
Introduction to the minitrack on Artificial Intelligence-based assistants and platforms
Get PDFAssistants and platforms based on artificial intelligence (AI) have become a new general-purpose technology (Helpman, 1998) in the digital economy, with chatbots and virtual personal assistants as examples. AI-based assistants and platforms provide seamless and intuitive access to digital services and devices. They free humans from the burden of acquiring domain knowledge and resources, allowing them to focus on more complex tasks. AI-based assistants also create business value by automating processes, intensifying user interaction, reducing errors, and speeding up interactions. While the concept of digital assistants is not new, the diffusion of general-purpose assistants, such as Amazon’s Alexa, Apple’s Siri, or Google’s Assistant (Këpuska & Bohouta, 2018), has fundamentally changed the presence of assistants. The technology is constantly advancing with developments in voice processing (Sivapriyan et al., 2021) and generative technologies, such as ChatGPT, Google Bard (Gozalo-Brizuela & Garrido-Merchan, 2023), and large language models like Bloom (Scao et al., 2022)
Microscopic and spectroscopic study on phase separation in highly crosslinked biobased polyurethane thermosets
Get PDFIn this study a biobased polyurethane (PU) thermoset is investigated due to its turbidity. In contrary to the expectations, the turbidity increases with a higher amount of a low molecular weight crosslinker. Morphological aspects are investigated with SEM imaging and measurement of the effective scattering coefficient μ’s. FTIR spectroscopy is applied to study the influence of the chemical structure. This is combined with multivariate data analysis to identify the relevant peaks. SEM images show spherical precipitations with increasing turbidity and a simultaneous increase in the μ’s values. FTIR analysis shows a significant amount of unreacted isocyanate‐(NCO)groups and a low level of hydrogen bonding. No formation of typical hard and soft segments is detectable. Therefore, it can be concluded that the increase in polarity differences with increasing crosslinker amount disabled the mixture of the polyol and isocyanate components, resulting in the precipitation of the isocyanate. At the same time, the low molecular weight crosslinker (~200 g mol−1) can react with the NCO quickly, reducing the mobility of the polymer chain, with remaining, non‐reacted isocyanate. A proof for the correlation of the differences in the FTIR and the μ’s values was found by a regression analysis with an R2 of 0.94
How Salesforce built its platform business : Salesforce built an external developer platform that allowed third parties to develop cloud apps extending the company’s core CRM system.
No full textIn 2023, Salesforce was the provider of the CRM system with the biggest market share. Besides providing a CRM software product, the company also offered an external developer platform called Salesforce Platform. Salesforce Platform enabled CRM customers to customize their CRM system and allowed partners (third-party developers) to develop cloud apps that could be distributed via Salesforce’s AppExchange marketplace. Since its official launch in 2008, Salesforce Platform had become a major revenue contributor for the company, including from over 7,000 apps in AppExchange.
Building a successful platform business extended far beyond building a technology platform to nurturing an entire ecosystem around it. Based on interviews with twelve executives, this case study illustrates what it takes for a product company to build a platform business—developing and scaling the business model, engineering the technology platform, growing a vivid community of developers, and marketing and selling the platform—and how Salesforce addressed related challenges
Exploring Hybrid Project Management: a qualitative inquiry of organizational adoption
No full textHybrid Project Management (HPM) represents a blend of traditional and agile practices that many organizations are increasingly adopting to combine the best of both methodologies. However, academic literature lacks a comprehensive understanding of how and why organizations adopt HPM and how they execute projects under this combined approach. Utilizing a qualitative multiple-case study design, the study intends to develop a theoretical model that encapsulates the adoption process of HPM. Employing contingency theory, which suggests that the effectiveness of management practices is contingent upon the situational context, as the theoretical framework, we delve into the diverse organizational contexts influencing HPM adoption. The paper discusses the ongoing data collection and analysis process and anticipates valuable insights that could be critical for both academic discourse and practical implications in the project management field
Customer success als unternehmensweite Initiative
No full textCustomer-Success-Management revolutioniert das B2B-Geschäftsbeziehungsmanagement. Der empirische praxisorientierte Beitrag untersucht, wie die Umsetzung von Customer-Success-Management als eine unternehmensweite Initiative gefördert werden kann, um Kunden zu erfolgreichen, langfristigen Nutzern zu entwickeln. Die Ergebnisse zeigen, dass eine erfolgreiche Umsetzung von Customer-Success-Management insbesondere von fünf Qualitätsdimensionen abhängig ist, die es im Anbieterunternehmen zu optimieren gilt
Enhancing early depression detection with AI: a comparative use of NLP models
Get PDFOne of the most underdiagnosed medical conditions worldwide is depression. It has been demonstrated that the current classical procedures for early detection of depression are insufficient, which emphasizes the importance of seeking a more efficient approach to overcome this challenge. One of the most promising opportunities is arising in the field of Artificial Intelligence as AI-based models could have the capacity to offer a fast, widely accessible, unbiased and efficient method to address this problem. In this paper, we compared three natural language processing models, namely, BERT, GPT-3.5 and GPT-4 on three different datasets. Our findings show that different levels of efficacy are shown by fine-tuned BERT, GPT-3.5, and GPT-4 in identifying depression from textual data. By comparing the models on the metrics such as accuracy, precision, and recall, our results have shown that GPT-4 outperforms both BERT and GPT-3.5 models, even without previous fine-tuning, showcasing its enormous potential to be utilized for automated depression detection on textual data. In the paper, we present newly introduced datasets, fine-tuning and model testing processes, while also addressing limitations and discussing further considerations for future research
Enhancing the conversion of waste motor oil into diesel-like fuels using mineral-impregnated biochar catalysts
No full textThis study shows how the presence of metal-impregnated biochar influences the kinetics of the catalytic conversion of waste motor oil (WMO) into a diesel-like fuel (DLF). First, the initial metal concentration in the wet impregnation process was optimized, showing that high metal concentrations tend to clog the biochar pores, reducing the reaction’s conversion and speed. Zinc and calcium showed higher kinetic constants at the optimized conditions than nickel, while the yield- and selectivity-to-liquid products were unaffected. Then, a reduction method was developed for impregnating metal particles of calcium oxide to enhance the metal dispersion at the biochar surface. This material showed the best performance in the reaction due to the presence of small particles instead of metal agglomerates. An increase of about 280% in the kinetic constant at 420 .C and a much lower activation energy (246 kJ mol-1) compared to the thermal cracking (308 kJ mol-1) was obtained. Finally, the DLF from the best catalytic system was analyzed. Interestingly, the fuel showed similar rheological properties to commercial diesel with a similar hydrocarbon distribution but with the presence of smaller hydrocarbon chains
The application of machine learning for demand prediction under macroeconomic volatility: a systematic literature review
Get PDFIn a contemporary context characterised by shifts in macroeconomic conditions and global uncertainty, predicting the future behaviour of demanders is critical for management science disciplines such as marketing. Despite the recognised potential of Machine Learning, there is a lack of reviews of the literature on the application of Machine Learning in predicting demanders’ behaviour in a volatile environment. To fill this gap, the following systematic literature review provides an interdisciplinary overview of the research question: “How can Machine Learning be effectively applied to predict demand patterns under macroeconomic volatility?” Following a rigorous review protocol, a literature sample of studies (n = 64) is identified and analysed based on a hybrid methodological approach. The findings of this systematic literature review yield novel insights into the conceptual structure of the field, recent publication trends, geographic centres of scientific activity, as well as leading sources. The research also discusses whether and in which ways Machine Learning can be used for demand prediction under dynamic market conditions. The review outlines various implementation strategies, such as the integration of forward-looking data with economic indicators, demand modelling using the Coefficient of Variation, or the application of combined algorithms and specific Artificial Neural Networks for accurate demand predictions
Process optimization of the morphological properties of epoxy resin molding compounds using response surface design
Get PDFAn epoxy compound’s polymer structure can be characterized by the glass transition temperature (Tg) which is often seen as the primary morphological characteristic. Determining the Tg after manufacturing thermoset-molded parts is an important objective in material characterization. To characterize quantitatively the dependence of Tg on the degree of cure, the DiBenedetto equation is usually used. Monitoring polymer network formation during molding processes is therefore one of the most challenging tasks in polymer processing and can be achieved using dielectric analysis (DEA). In this study, the morphological properties of an epoxy resin-based molding compounds (EMC) were optimized for the molding process using response surface analysis. Processing parameters such as curing temperature, curing time, and injection rate were investigated according to a DoE strategy and analyzed as the main factors affecting Tg as well as the degree of cure. A new method to measure the Tg at a certain degree of cure was developed based on warpage analysis. The degree of cure was determined inline via dielectric analysis (DEA) and offline using differential scanning calorimetry (DSC). The results were used as the response in the DoE models. The use of the DiBenedetto equation to refine the response characteristics for a wide range of process parameters has significantly improved the quality of response surface models based on the DoE approach
Self-categorization theory with 3D emotional model for chatbot emotional support systems
No full textChatbots are increasingly being used for mental health care related to stress and other issues. With the advent of ChatGPT, conversations with chatbots have become commonplace, and a variety of support has become possible. However, future developments are expected to determine what kind of personality and persona mental healthcare chatbots will have and how they will express their emotions. Chatbots often have emotional models to show empathy to users, but only the psychological information of the users is considered, and there are few studies that cover physiological information as well. Therefore, in this study, to give the chatbot characteristics suitable for the user, we proposed a method for expressing emotions using three types of input information regarding active listening in emotional support