Contributions published at Information Management (Gerhard Schwabe)

The Master's thesis takes a closer look at the term Third Wave AI. The Defense Advanced Research Projects Agency (DARPA) introduced a framework, where the evolution of artificial intelligence (AI) is categorised with waves. According to DARPA, the next wave will be the third one, which is based on contextual adaptation. Based on the initial literature review, the issue of the legitimacy of DARPA's framework was raised. To address this topic, different types of analyses were performed. The public discourse analysis was used to identify central entities and dominating topics. They showed a lack of common definitions and intermixing of Third Wave AI related terms. Since the results revealed the need for adjustments, a new framework was suggested, challenged and revised in a sensemaking workshop. For further improvements, interviews with industry and AI experts were conducted. They validated the identified problems and many suggestions for enhancements were collected. With these insights, the framework's final version was created. The main component consists of a technical and functional layer, which are each segregated in dimensions representing the requirements. On the functional layer, the dimensions human-centered, explainable and contextual adaption were set as prerequisites. The technical layer consists of DARPA's requirements, which are perceiving, learning, abstracting and reasoning. Furthermore, the main component was expanded with a zoom-out and zoom-in. For the framework's application, a spider diagram is presented. By introducing a new approach with the corresponding definitions, the identified needs for action are remedied, and the legitimacy of Third Wave AI will be improved.

The present thesis resulted from a collaboration between the Information Management Re-search Group (IMRG) of the University of Zurich and Braingroup AG. Based on the «Design Science Research Methodology» the usage of a virtual assistant in the context of hybrid advi-sory services was explored. With the counsel of advisory experts, process steps of a traditional Face2Face advisory session have been identified, which can be performed autonomously by the client. These process steps were included in a Paper-Prototype of a self-advisory module for a retirement planning consultation, which serves to complement the in-person advisory session. Additionally, an IT-artifact has been developed, which covers one of the functionali-ties of the self-advisory module. It contains learning modules that can be utilised as a prepara-tion for an advisory service encounter. Thereby, a virtual assistant guides the user through the learning contents. Throughout the evaluation, this artifact has been compared to a traditional preparation. The testing participants rated the developed prototype significantly more posi-tive. Further, with the usage of the artifact, a higher increase in the objective knowledge has been measured. In addition, it is shown that the acceptance of a virtual assistant is given in the context of advisory services.

Conversational agents have long been introduced to private settings. The growing interest to apply these systems in a professional work context is subject to discussion in both scientific research as well as public discourse. This research seeks to provide a broad overview of the discussion of conversational agents by analyzing the discourses of common and scientific literature respectively focusing on agent collaboration and perception. Systematic qualitative literature analyses of both literatures are made with methods and procedures described and documented. The results of the processes are presented, and the discourses are categorized by different topics and compared. It is found that both literatures provide similar narratives about effects of conversational agents with the scientific literature providing more in-depth explanations.

The increasing proliferation of conversational agents in our daily life does not leave the business world untouched. New possibilities and use-cases of speech-based, virtual agents in the professional environment are constantly being explored. However, previous approaches to developing conversational agents for professional multi-party environments were confronted with the challenge of seamlessly embedding virtual agents into the natural dialog. In this respect, the lack of context-awareness of today's dialog systems represents a hurdle. To take a first step towards context-aware conversational agents, this thesis investigates the possibilities for automatically recognizing explicit and implicit meaning of utterances in dialogues using an explorative research approach. Based on theories of pragmatics as well as current developments in the fields of Natural Language Processing and Deep Learning, different models for the recognition of intentions in German dialogues were developed and tested under real-life conditions using a prototype of a dialog system. The results provide information about the most important design implications and challenges in the development of dialog systems with automatic intention recognition. Thus, this work not only contributes to the current research of context-aware conversational agents, but also serves as a starting point for future research and practice.

The Cardossier project was started at the University of Zurich to bridge the information asymmetry between a buyer and a seller in a used car transaction. The Information asymmetry is a well-known issue that the current used car market must deal with because it can reduce the quality of offered cars in the market. The Cardossier, which is based on a blockchain, should help to solve the transparency issue and increase trust through the usage of the blockchain technology. This work examines the trade practices in the Cardossier game, how these practices change through the introduction of the Cardossier and the impact of those practices. A total of three different versions and results of this game were analysed along with the interviews of those games to present a variety of results of those practices. The results of this work may be helpful for the future design of this game and future studies related to the Cardossier game.

As a growing number of organizations recognizes the potential of blockchain for new business models, inter organizational collaborations using blockchain are on the rise. Based on interviews with 53 members of 19 consortia applying blockchain toward a shared objective, this thesis studies motivations and dynamics within blockchain collaborations. Primary motivations for forming the consortia were efficiency and financial gains, innovation and gaining experience with the technology. Regarding dynamics in blockchain consortia, a not blockchain-specific factor (alignment on vision) is cited as the main hurdle, and the interviews suggest that some challenges emphasized in prior, non-blockchain research (e.g. power clashes) are less prevalent in the blockchain consortia.

The European Conference on Computer-Supported Cooperative Work, taking place in 2021, is organized by the University of Zurich. The aim of this thesis is to design and develop the optimal website to represent the academic conference and examine the hedonic quality aspects of such academic conference websites through a qualitative research. Furthermore, the findings of the qualitative research are presented and used in the implementation of the academic conference website and key guidelines are proposed for achieving a high hedonic quality of conference websites.

The purpose of this thesis is to design and develop the optimal ECSCW 2021 conference website. Moreover, in this thesis an integrated qualitative research is conducted in order to determine key guidelines for the pragmatic quality of conference websites and examine how stakeholder groups benefit from them. Thus, this thesis focuses on improving the effectiveness of information transfer and efficiency of stakeholder actions on conference websites and realizes the determined pragmatic quality guidelines in the implementation of the ECSCW 2021 conference website.

Structured entry of medical data offers many benefits, but it can also be cumbersome for medical staff. This thesis explores how to enable efficient structured data entry. From the reviewed literature, 122 recommendations that contribute to the success of an electronic health record (EHR) tool for structured documentation were extracted. Using EHR software reviews and demonstrations of EHR tools by their respective providers, five EHR tools were then analyzed based on these recommendations. This analysis revealed common gaps between the recommendations in the literature and the EHR tools currently on the market. In response, a solution is proposed on how to close these gaps. The main contribution of this solution is the intuitive way in which it shows connections between observations, diagnoses and interventions. This is facilitated by structured yet simple documentation of these concepts. The proposed solution may reduce cognitive workload, especially for inexperienced physicians. It may further contribute to facilitating patient empowerment and shared decision-making, with the downstream benefit of patients’ improved adherence to treatment.

The physicians of the pain clinic at the Center for Dentistry (ZZM) of Zurich University use an application containing images and options for the documentation of findings to educate patients. The aim of this work is to further develop the possibilities for personalization, especially in the pictorial representation of pain. For this purpose, patient surveys and online surveys were carried out, based on which new im-age material was designed and integrated into the app, which has been expanded in its functional scope. At the same time, a user interface for the presentation of 3D models was designed. After repeated evaluations and a focus group it can be stated that the aim of the future development of the app should be the integration of the WISE questionnaire and the creation of an integrated, user-friendly image processing interface, while the further development of the 3D features should initially be of minor concern.

Since the boom of blockchain when Bitcoin was initially released in 2009, many new areas of application have emerged for this new type of technology. One of the more popular application fields is the "Decentralized Autonomous Organization", or in short "DAO". Early research shows that this might be a new mode of organization, different from existing traditional organization forms. Therefore, more research on DAOs is needed to gain more insights into their governance structures. In this bachelor's thesis, a lens is developed to perform qualitative data analysis on 3 popular DAOs, Dash, MakerDAO and EOS. This is done by collecting and coding different forms of sources to each DAO and by applying a framework to analyze the governance characteristics to determine their mode of organization. The results for each DAO are further compared to each other to gain additional insights. The results show that the governance structures of the DAOs are similar to each other. As previous research has suggested, the results show that the mode of organization of each DAO is classified to be hanseatic, a mode of organization that is different from traditional ones. The findings further suggest that all DAOs are dealing with the same multitude of problems. The most critical problem is that they are struggling to keep themselves decentralized, raising the question if DAOs can even be considered as such. The research also suggests that DAOs need more time to mature to tackle the issues they have.

While the number of break-ins in Switzerland might be drastically declining, the people who do still become victims of burglars must face the consequences such an experience entails. The damage that needs to be repaired and the stolen property that needs to be replaced is only a small part of these consequences. The emotional stress that the victims experience is said to be far more severe than the material consequences. In order to prevent further break-ins from happening, many police departments offer free-of-charge burglary prevention advisory services. In these advisory services the police advisors personally visit the property in question and give their customers (= "advisees") recommendations on how to best secure their property. In close collaboration with several Swiss and German police departments the University of Zurich has developed a tablet application, the App EP, which police advisors can use during these advisory services to enhance the persuasiveness of their advisory efforts. On the basis of a survey and interviews with advisees and advisors, this thesis examines the implementation rates of the home security measures, which are recommended through such advisory services with the help of the App EP. Additionally, the underlying reasons for the decision to implement or not implement the given recommendations are explored. Five persuasion types of advisees and their respective advisory needs are presented. Also, the risk-effort calculation, which determines whether or not an advisee will implement the recommended security measures, is explained. Finally, recommendations for the further development of the App EP and such advisory services in general are made.

This thesis investigates the potential of a blockchain-based application to combat counterfeit watches in the Swiss watch industry. Four interviews with experts of the watch industry were conducted, which served as a basis for the development of requirements with the subsequent implementation of a prototype. This work provides an insight into the current state of the Swiss watch industry and information about the trust relationships of manufacturers, retailers, and private individuals in the Swiss watch market. It is shown that a blockchain-based solution, with the provision of certificates of authenticity, can offer added value between private persons. The benefit is limited to consumers seeking a guarantee of watch authenticity. The presented solution cannot solve the problem of the conscious purchase of cheap counterfeit watches.

These days many firms are collaboratively working on blockchain projects. However, this form of collaboration, commonly referred to as consortium, is influenced by a variety of external factors. To analyze these factors, the thesis uses existing literature to propose a new framework, specifically tailored towards blockchain consortia. The thesis then applies this newly introduced framework on interview transcripts, provided by the Information Management Research Group of the University of Zurich. The examination identifies which factors exist, how they impact the consortia and which approaches are used to deal with these impacts. The study then further examines the relation between the identified factors and determines which factors can be considered key and blockchain specific.

Despite the increasing influence of information technology systems on business processes, financial advisory services are still primarily based on traditional tools such as pen and paper. However, IT-systems can offer advisors several advantages through an improved knowledge transfer and simplified calculations. Research has shown that these tools can help mitigate some of the issues identified in financial advice to a certain extent. The goal of this thesis is to analyze how these digital tools influence the advisory process and practices in the setting of mortgage advisory sessions. Furthermore, the development of ThinkLets shall assist advisors in implementing these systems properly in the advisory environment to achieve the desired results. For this analysis, we encoded and examined video data gathered previously from both traditional and IT-supported advisory sessions. The analysis focuses on the LivePaper and BankNotes systems and illustrates how practices evolve compared to sessions in conventional settings.

Blockchain comes with the promise of being a disruptive technology with the potential for novel ways of interaction in a wide range of applications. Following broader application, scholarly interest in the technology is growing, though an extensive analysis of blockchain applications from a governance perspective is lacking to date. This research pays special attention to the governance of blockchain systems and illustrates decision problems in 14 blockchain systems from four application domains. Based on academic literature, semi-structured interviews with representatives from those organizations, and content analysis of grey literature, common problems in blockchain governance have been singled out and contextualized. Studying their enactment revealed their relevance to major organizational theories in what we labelled “Patrolling the borders,” “External Legitimation,” “Reduction of Discretionality,” and “Temporal Management.” The identification of these problems enriches the scarce body of knowledge on the governance of blockchain systems, resulting in a better understanding of how blockchain governance links to existing concepts and how it is enacted in practice.

Digitisation is having an increasing impact on everyday life. Companies are using these opportunities to drive digital innovations. Realising them successfully is perceived to be much more difficult than their non-digital counterparts. What are the associated risks with digital innovation? What makes promising projects successful, what makes them fail? Through qualitative interviews, the success factors and failure factors of digital innovation projects are examined in the Swiss industry. Although digital innovations are strongly dependent on the technology, the market and the company around the employees and their collaboration play a much higher role. This work contributes to a better understanding of the respective difficulties and differences that various companies face in digital innovation projects. Furthermore, a proposal is made on how a possible technology monitoring based on research data could look like.