Human-Computer Interaction – INTERACT 2021

During the COVID-19 pandemic, social distancing measures were employed to contain its spread. This paper describes the deployment and testing of a passive Wi-Fi scanning system to help people keep track of crowded spaces, hence comply with social distancing measures. The system is based on passive Wi-Fi sensing to detect human presence in 93 locations around a medium-sized European Touristic Island. This data is then used in website plugins and a mobile application to inform citizens and tourists about the locations’ crowdedness with real-time and historical data. To understand how people react to this type of information, we deployed online questionnaires in situ to collect user insights regarding the usefulness, safety, and privacy concerns. Results show that users considered the occupancy data reported by the system as positively related to their perception. Furthermore, the public display of this data made them feel safer while travelling and planning their commute.

In March 2020, the government of India ordered a nationwide lockdown to prevent the spread of Covid-19. This led to the shutdown of educational institutes throughout the country, restricting all activities to online mediums. The shift has affected how students engage with each other, where rather than in-person interaction, they meet through a variety of online tools. In this paper, we discuss how the normal everyday routine of ‘hanging out’ with friends has been transformed during a prolonged lockdown of over ten months and counting. We investigate the opportunities and challenges students encounter when socializing online through various online modes including video calls, communal movie watching and social media. We discuss how social interaction; in particular, hanging out with friends has been transformed through these technologies and its implications for facilitating spontaneous interaction, negotiating intimacy, mutual understanding, and accessibility to different social groups. Finally, we conclude with a discussion of how these factors impact the transition from in-person to online modes of casual social interaction.

This research investigates Italian citizens’ attitudes towards the contact tracing app Immuni, promoted by the Italian government to track the spread of Covid19 and prevent possible future out-breaks. More specifically, this paper tries to uncover the factors that have motivated individuals to uptake the app or fail to do so. We have used a variety of qualitative methods: firstly, a virtual ethnography has been conducted, analyzing thematically 3013 tweets with the hashtag #Immuni. To further investigate the motivation behind either the adoption or the refusal of Immuni, twenty semi-structured interviews have been carried out with potential and actual users. A doctor and a health official have been also interviewed as secondary users. The results show that multiple factors shape users’ attitudes towards Immuni and influence their willingness to adopt it. The most relevant have to do with the perceived benefits, perceived barriers, perceived efficacy of the app and the cues to action. The usability of the overall app is also discussed with a focus on some of the key issues identified by the users.

Social isolation is one of the negative consequences of a pandemic like COVID-19. Social isolation and loneliness are not only experienced by older adults, but also by younger people who live alone and cannot communicate with others or get involved in social situations as they used to. In such situations, social companion robots might have the potential to reduce social isolation and increase well-being. However, society’s perception of social robots has not always been positive. In this paper, we conducted two online experiments with 102 and 132 participants during the self isolation periods of COVID-19 (May-June 2020 and January 2021), to study how COVID-19 has affected people’s perception of the benefits of a social robot. Our results showed that a change caused by COVID-19, as well as having an older relative who lived alone or at a care center during the pandemic significantly and positively affected people’s perception of social robots, as companions, and that the feeling of loneliness can drive the purchase of a social robot. The second study replicated the results of the first study. We also discuss the effects of Big 5 personality traits on the likelihood to purchase a social robot, as well as on participants’ general attitude towards COVID-19 and adapting to the pandemic.

Increasing use of online conferencing systems, particularly over the past year, has highlighted problems in these systems, especially their poor support for small group interactions within larger meetings. These include clumsy small group formation (e.g., issues around joining and leaving existing groups), the difficulty of getting the correct level of audio isolation between groups, poor provision for shared editing of documents, as well as fatiguing aspects of video conferencing caused by presentation format and the necessity of remaining on camera view. This paper describes the motivation, design and implementation of a prototype online conferencing system, called BubbleVideo. Building on both virtual world and pure video paradigms, it implements an extensive 2D world with shared documents, in which users appear through real-time video, presented in “bubbles” that can be moved around. Users are given the possibility of deciding whether to join a group by viewing a conversation “leakage”, which group members can share with outsiders.

Evaluation of a novel user-performance model’s fitness requires comparison with baseline models, yet it is often time consuming and involves much effort by researchers to collect data from many participants. Crowdsourcing has recently been used for evaluating novel interaction techniques, but its potential for model comparison studies has not been investigated in detail. In this study, we evaluated four existing Fitts’ law models for rectangular targets, as though one of them was a proposed novel model. We recruited 210 crowd workers, who performed 94,080 clicks in total, and confirmed that the result for the best-fit model was consistent with previous studies. We also analyzed whether this conclusion would change depending on the sample size, but even when we randomly sampled data from five workers for 10,000 iterations, the best-fit model changed only once (0.01%). We have thus demonstrated a case in which crowdsourcing is beneficial for comparing performance models.

In this exploratory study we evaluated the engagement, performance and preferences of older adults who interacted with different citizen science tasks. Out of 40 projects recently active on the Zooniverse platform we selected top ones to be represented by 8 diverse, yet standardized, microtasks, 2 in each category of image, audio, text and pattern recognition. Next, 33 older adults performed these microtasks at home and evaluated each task right after its completion to, finally, share what could encourage them to engage with such tasks in their free time. Based on the results we draw preliminary conclusions regarding older adults’ motivations for engaging with such crowdsourcing tasks and suggest some guidelines for task design while discussing interesting avenues for further inquiry in the area of crowdsourcing for older adults.

A significant challenge for creating efficient and fair crowdsourcing platforms is in rapid assessment of the quality of crowdwork. If a crowdworker lacks the skill, motivation, or understanding to provide adequate quality task completion, this reduces the efficacy of a platform. While this would seem like only a problem for task providers, the reality is that the burden of this problem is increasingly leveraged on crowdworkers. For example, task providers may not pay crowdworkers for their work after the evaluation of the task results has been completed. In this paper, we propose methods for quickly evaluating the quality of crowdwork using eye gaze information by estimating the correct answer rate. We find that the method with features generated by self-supervised learning (SSL) provides the most efficient result with a mean absolute error of 0.09. The results exhibit the potential of using eye gaze information to facilitate adaptive personalized crowdsourcing platforms.

The driving experience has become one of the central decision factors when buying a car. In current manually driven cars, this experience is to a large extent influenced by driver-based infotainment functionalities. With the advent of rear-seat infotainment systems, manufacturers started to not only look at the driver’s perspective but also focus on passenger experiences. But passenger experiences can go beyond traditional aspects of user experience as they also include aspects of coziness and comfort in the context of riding, which we describe as passenger convenience. While insights about the design space and passenger’s needs are central when designing for an advanced level of passenger convenience, the body of knowledge in this area is limited. Therefore, we present the results from a repertory grid study (n = 32) where we investigated what makes a passenger ride in a manually driven car convenient. Based on three predefined and three participant-selected riding situations we accounted for common patterns and individual differences. The results confirm the importance of well-being, physical comfort, and safety. The interviews unveil that passengers strive for access to in-vehicle systems, the possibility to act as a co-driver, and the support for the integration of external technology, connectivity, and personalization. Based on our findings, we extracted a set of design recommendations to consider when designing automotive systems with passenger convenience and experience in mind.

We describe applications that implement interactions between the driver and their smart vehicle in a continuous interaction space characterized by the physical distance to the vehicle and by the smart devices that implement those interactions. Specifically, we demonstrate the principles of smart vehicle proxemics with smart rings, smartwatches, smartphones, and other devices employed to interact with the in-vehicle infotainment system while the driver traverses five distinctly identifiable zones, from inside the vehicle to the personal, proximal, distant, and covert zone outside the vehicle. We present engineering details of our applications that capitalize on standardized web technology (HTML, CSS, JavaScript), communication protocols (WebSocket), and data formats (JSON) and, thus, enable straightforward extension to accommodate other smart devices for new interactions with smart vehicles. We also point to future opportunities for designing interactions from a distance and function of the distance between the driver and their vehicle.

We introduce smart vehicle proxemics, a conceptual framework for interactive vehicular applications that operationalizes proxemics to outside-the-vehicle interactions. We identify four zones around the vehicle affording different kinds of interactions and discuss the corresponding conceptual space along three dimensions (physical distance, interaction paradigm, and goal). We study the dimensions of this framework and synthesize our findings regarding drivers’ preferences for (i) information to obtain from their vehicles at a distance, (ii) system functions of their vehicles to control remotely, and (iii) devices (e.g., smartphones, smartglasses, smart key fobs) for interactions outside the vehicle. We discuss the positioning of smart vehicle proxemics in the context of proxemic interactions more generally, and expand on the dichotomy and complementarity of outside-the-vehicle and inside-the-vehicle interactions for new applications enabled by smart vehicle proxemics.

Patients on haemodialysis face complex care pathways, a high treatment burden and lower quality-of-life. Working with multidisciplinary domain experts, we have conducted several iterative development cycles to design, develop and evaluate a portal for patients on haemodialysis that can help them better understand and navigate their care pathways. A key functionality of the portal is to improve data and information sharing with clinicians, including on key aspects of quality-of-life through Patients Reported Outcome Measures. A case study was conducted with multidisciplinary experts and patients in the NHS Greater Glasgow and Clyde health board (Scotland), using interviews combined with the System Usability Scale (n = 26). Patients’ feedback and system use observations were used to further refine the system design requirements and functionalities. Key lessons include: a wide preference for tablet-based input vs paper, identification of case-specific accessibility issues and situational impairment, benefits of self-completed digital data collection in overcoming such issues and promoting patient independence and privacy, with considerations for maintaining perceived value and engagement with such systems and when to offer alternatives.

Digital tools are an essential part of the work environment of many civil servants working at public authorities. They should therefore follow the prevalent work environment legislation. This paper describes a method to conduct digital work environment rounds to assess the usability of digitalized work to assess work environment risks and identify improvements. The methodology used is following a participatory action research project in collaboration between users, union representatives, managers and IT experts. The case presented in this paper describes the development of the digital work environment round in collaboration with a court in Sweden. It was based on an international standard in the making and collaboratively developed and adapted to the situation at the public authority. Based on this work the public authority has now decided to make this method a part of their annual systematic work environment management and other organizations are following in their tracks.

Studies indicate that the way structured participatory design activities are facilitated significantly effects participation, power relations, and generated output. Although facilitation is always adapted to the conditions of each design project, the way facilitation is enacted primarily depends on the individual facilitator. It is assumed that enactment on an individual level is influenced by how the facilitator conceptualizes both user involvement and their own role. To address these issues, semi-structured interviews were conducted with three facilitators in the same development unit of a large-scale health IT project. The primary focuses were the individual facilitator’s perspectives on the design process, their role, and their practice in the project. An in-depth qualitative content analysis of the transcribed interviews showed that the three facilitators held highly divergent perspectives on the facilitation of participatory design activities. Similarities were found between the facilitators’ perspectives and how user involvement is understood in different system development traditions. Based on the findings, a reflective approach towards facilitation is proposed along with examples of possible co-reflection activities among facilitators.

Google Design Sprint (GDS) is becoming a valued tool for interaction design practitioners today. Although GDS has some similarities to User-Centred Design (UCD), it does not study user needs before generating solutions. On the other hand, UCD provides little guidance on producing design solutions. We saw the two processes would nicely complement each other. This paper reports development of an intensive two-week interaction design course where UCD was combined with GDS. The feedback from 22 higher education students indicates how UCD helped them to keep the focus on the important things, and how the detailed structure of GDS process guided them fast forward. In the fast-paced Design Sprint, students need dedicated time for reflection. The contributions of this work include the course structure for teaching a User-Centred Design Sprint process, student feedback on the new process, and recommendations for teaching such a course.

The scope of Human-Computer Interaction (HCI) research is expanding with regard to the studied systems and stakeholders, and its impact areas. Service design has recently gained tractions in HCI as an approach to deal with these expansions. However, there has been confusion around the definitions and roles of service design in HCI, especially with its overlaps and differences with interaction design. To examine how HCI has adopted service design, this paper presents results from a systematic literature review on 52 papers from the most cited HCI publication venues. Our findings show that the adoption of service design concepts and methods in HCI has been sporadic over the past decade. The term service design has been interpreted as a variety of meanings. The most predominantly observed understandings include service design as a term for designing digital services instead of products, and as an approach providing a journey and system perspective to the design of social computing, Internet of Things, or other complex systems. Only a few studies adopted the fundamental logic of new value exchange or co-creation of systems from service design. We discuss the reasons behind the differing interpretations of service design by HCI and future opportunities for HCI to better benefit from service design.

It has long been known that our sense of smell is a powerful one that affects emotions and behaviors. Recently, interest in the sense of smell has been growing exponentially in HCI. However, the potential of smell to inspire design is still underexplored. In this paper, we first investigated crossmodal correspondences between scents and selected features relevant for design (clustered in sensory, bodily, and qualitative features). Then, we created a set of cards (EssCards) to visually summarize the key findings to inspire designers. We carried out two preliminary design exploration sessions using the EssCards. Based on our findings, we discuss how to inspire and challenge design opportunities around the sense of smell and reflect upon applications for smell as inspirational material for designing future interactions and experiences.

Sequential art in the form of comics is a powerful and effective vehicle for graphic storytelling and communication, rendering it a suitable means for design fiction. Few, dispersed examples introduce comic-based design fictions in HCI research, yet little is reported on the design and implementation process or the use of paper cut-outs. In this paper, we present our process of crafting and implementing comics and character cut-outs for facilitating co-designing fictions. Our utopian visions, “Tales from the Materialverse”, informed by our interest in future smart materials and their applications, were used as a provocative communication tool to mediate discussions and encourage critical thinking. Based on learnings derived from our process and an expert workshop, we propose suggestions for designing design fiction comics and cut-outs as an alternative co-design fiction approach.

Architecture has been taking new turns with rapidly developing digital design and fabrication technologies. Consequently, establishing a link between physical and virtual design methods remains an open area for investigation. This paper explores the contemporary idea generation methods and the role of physical and digital design techniques in the initial design processes of architecture. We report our findings from interviews conducted with 14 participants consisting of experts and practitioners from the architecture field. Then, we discuss potential application areas of the results in the context of HCI research.

Personas is a suggested method to extend IT professionals' understanding of users and users’ needs. A common advantage expressed is that personas extend the IT professionals’ empathy for the users, but a disadvantage is that personas are typically defined at the start of a software project and gradually are forgotten, since there is little reference to the personas through the software development project. In this paper we report experiences of coaching IT professionals in defining agile user stories based on personas, called: Persona User Stories (PUS). The aim of these workshops, was to extend the usage of personas and thereby extend the IT professionals’ understanding of their users. In a research project with three companies, we coached teams of IT professionals in three-hour workshops with 76 participants in total. The workshops were conducted at each company using personas already defined by the IT professionals. The persona descriptions were based on three types of information: (a) assumptions, (b) secondary research, and (c) data specific to a project. Our findings show that personas based on assumptions result in the participants questioning the description of the personas and having difficulties in understanding the personas. For making the persona user stories (PUS), the participants used themselves more often as a reference when working with the assumption based personas, than the participants using the other two types of personas.

We envision a unique social interaction system, ‘users-as-beacons’ built upon Bluetooth Low Energy (BLE) beacon technology, that could provide potential privacy benefits. It leverages BLE to employ the user devices to act as mobile beacons. Its potential applications include community-based social networking, localized advertising, and instant reviewing. To evaluate the potential for this system and inform design, we conducted an exploratory interview study of 27 participants of a hypothetical localized content-creating system. Using a design prototype and multiple scenarios as prompts, we asked questions regarding users’ perceptions of the potential benefits and challenges of a users-as-beacons system, focusing in particular on their privacy concerns and needs. Our results indicate that users do perceive the benefit of increased trustworthiness of user-beacons, but do not have expectations of greater location or behavioral tracking privacy. We highlight multiple design challenges of this system in supporting the trustworthy, relevant, and timely sharing of posts between people in a community.

In this paper, we present the design and the evaluation of an authoring tool for End-User Development, which supports the definition of Trigger-Actions rules that combines events and states in the triggers. The possibility of using either states or events in triggers has already been discussed in the literature. However, it is recognized that the state/event distinction is difficult to manage for users. In this paper, we propose an authoring tool that provides explicit support for managing this distinction. We compare it with a state-of-the-art authoring tool that implements the classical event-event paradigm.

This paper contributes to HCI with a foundation for exploring ways in which an Internet of Things (IoT) solution can support the maintenance of public automated defibrillators (AEDs) sustainably. AED is a critical lifesaving technology installed in non-medical environments. The technology places enormous demands on public engagement for its use and maintenance. Insights into how medical technology is maintained in non-medical environments can complement current innovations in lifesaving technology. This ethnographic study investigates the complexity around an IoT solution to remedy breakdowns in the maintenance of AEDs in a Danish context. We conceptualize maintenance as infrastructuring and identify how the diverse field of actors around an AED entails technical but primarily social dimensions revealing a controversy of responsibility and accountability. This article recommends that when developing an IoT solution for maintaining AEDs, technical, social, political, and ethical dimensions should be considered to address responsibility and accountability issues.

In response to an increasingly ageing world population, the World Health Organization has recently proposed a new framework – called Healthy Ageing – to better cater to older adults’ life needs and desires. This framework calls for transdisciplinary approaches to answer the challenges and opportunities posed by an ageing society. Following this framework, it is argued that the first step to accomplishing the goals of Healthy Ageing is to address the existing biases and stereotypes in the language used for addressing ageing issues. In this paper, we propose a series of linguistic guidelines that can help improve the lexicon used to talk about ageing in design research and practice, by countering many subconscious ageist biases and stereotypes. Ageism, especially in its implicit form, is a pervasive negative force that can affect not only design practitioners and researchers but also older adults, who often use ageist language to describe themselves. The proposed guidelines would also help the “false friends” of ageing people, who may be using apparently innocuous words, terminology, and remarks to refer to ageing, without perhaps realising the underlying ageist attitudes and stereotypes hidden in their used language. Overall, we aim to foster a change in design researchers’ and practitioners’ perspective on ageing, by paving the way for a transdisciplinary discourse on designing for ageing, which could then be further expanded and explored to eliminate ageism in design.

There has been a considerable amount of research aimed at automating the documentation of accessibility in the built environment. Yet so far, there has been no fully automatic system that has been shown to reliably document surface quality barriers in the built environment in real-time. This is a mixed problem of HCI and applied machine learning, requiring the careful use of applied machine learning to address the real-world concern of practical documentation. To address this challenge, we offer a framework for designing applied machine learning approaches aimed at documenting the (in)accessibility of the built environment. This framework is designed to take into account the real-world picture, recognizing that the design of any accessibility documentation system has to take into account a range of factors that are not usually considered in machine learning research. We then apply this framework in a case study, illustrating an approach which can obtain a f-ratio of 0.952 in the best-case scenario.

This research explores the relationship between social media and mortality, focusing on social media as a thanatosensitive technology that promotes death preparation and commemoration. Employing concepts from the Value-Sensitive Design approach, our research identifies three main stakeholders surrounding the online commemoration experience. As an exploratory study, we conducted in-depth interviews and observations (n = 9) to better understand key stakeholders’ experiences in online and offline commemorations. After analyzing the qualitative data, we extracted design requirements that were used to create a conceptual commemorative system and a functioning prototype. An informal user testing (n = 6) yielded positive feedbacks on the new system for it offers the unique opportunity to reflect existence, prepare for death, and create a community of commemoration that helps memorialize the deceased and recover from grief. The study suggests that commemoration via social media aids in maintaining user volition, induces active and interactive participation, and allows for a continuation of the digital legacy. Such findings may help deepen the understanding of humans’ perception of death and commemoration within the context of social media.

Social robots are autonomous entities able to engage humans at the emotional and social level. They are being used in several domains, especially in those where kids are the primary users (i.e., education, games, rehabilitation). The paper presents an experience in which the social robot Pepper is used as a storyteller. A storyteller robot should engage humans by combining its verbal and non-verbal behaviors and ‘immerse’ the user into the story. Therefore, to design an engaging and effective storytelling experience we started to address a first design issue: does a human voice have an advantage over a synthesized voice of the robot in this context? To this aim, two versions of the same story for kids from 8 to 9 y.o. have been developed. The social robot Pepper was used to tell the story in two modalities. In the first modality, Pepper storyteller was designed as a kind of audiobook in which the robot had just the role of a device, but the story was narrated by a human voice; in the second modality, Pepper was designed to tell the story using its own voice combined with non-verbal behaviors. The system has been tested in a real context and results show that Pepper’s voice affected more positively the children’s emotional experience, also by giving the children the perception that they learn more easily.

There exists discrepancy between the skills possessed by Human Computer Interaction (HCI) students and the expectations of user experience (UX) industry. This skill gap lowers HCI students’ successful transition from academia to the field of UX. To reduce the skill gap, HCI educators have started to integrate experiential learning techniques into their teaching, ranging from workshops to service learning and industrial internships. In joining their efforts, we have created an innovative usability training known as Collaborative Learning of Usability Experiences (CLUE) for HCI students of all levels. In this paper, we detail CLUE’s unique experiential training components—UX Internship, Short Courses, Workshops, and Knowledge Transfer—and we report the assessment of the program by evaluating our graduates’ job placement and their academic achievement and also by examining our UX industry partners’ evaluation of the program and student interns on five important UX skills.

The presence of mobile devices (e.g., smartphones, tablets and computers) in the classroom gives students the possibility of doing off-task activities during lectures. The purpose of this mixed-method field study was to learn more about students’ behaviors, reasons, and opinions regarding such activities and their consequences on learning. This study is one of few to take a holistic view on this topic by taking the use of all technical devices in class into account and assessing its consequences on learning objectively. This is important to gain a full picture concerning the consequences of off-task activities in class. Right after a lecture, bachelor students (N = 125) answered a survey containing questions on their usage of mobile devices during this last class. Furthermore, they took a test on the content of that lecture. Qualitative and quantitative analysis of data revealed that students spent an average of more than 19% of their time using a digital device for non-class purposes. Interestingly, this was not significantly linked with learning, although many students reported being aware of this behavior’s potential negative consequences. But there was a significant negative link between the number of received notifications and learning. These results suggest that external interruptions have a stronger negative effect than internal interruptions, allowing us to make better recommendations on how to use electronic devices in the classroom.

Listening to sound individually while in close proximity of other people is increasingly enabled by a range of technologies. One still in development is sound zone technology that aims to provide personal sound without headphones or other wearable speakers. User-oriented studies in the area of personal listening primarily emerge from the fields of acoustics and sound engineering but are gaining increasing interest within HCI research. In this paper, we present a study investigating the experience of personal sound in relation to different types of situations and personal sound technologies. Our findings show strategies for adjusting personal sound and social interaction, descriptions of sound quality in relation to sound and situation types, and insights into participants’ experiences of awareness using personal sound technology. The paper contributes with a thematic characterisation of this type of technology, serving as a foundation for further studies. This furthermore initiates a discussion on personal sound technology and soundscape composition in how situation types affect which sounds to include or exclude, and when.

As a technology that allows sound to be heard only in specific areas in public spaces, the directional sound of parametric speakers has been attracting attention. The parametric speaker is also expected to have a wide range of uses, in settings ranging from shopping malls to museums for presenting information sound. The purpose of this study is to measure evaluate the Speech Recognition Threshold (SRT) of parametric speakers and compare with that of loudspeakers at various noise levels through experiments. It can also reveal how much the parametric speaker can reduce the noise heard by non-target users, which is unavoidable with loudspeakers.Through experiments, the volume required to transmit an informative sound to a particular listener is about 12.61 dBA less than the level necessary for the loudspeaker. Moreover, the SRT of the surrounding people is about 2.48 dBA higher than that for the loudspeaker. Based on the results, the volume received by people in the surrounding area, including environmental noise, was calculated, and the effect of noise reduction is discussed. This study provides important insights for the flexibility in the design of the sound space.

Virtual Reality (VR) training has become increasingly important for police first responders in recent years. Improving the training experience in such complex contexts requires ecological validity of virtual training. To achieve this, VR systems need to be capable of simulating the complex experiences of police officers ‘in the field.’ One way to do this is to add stressors into training simulations to induce stress similar to the stress experienced in real-life situations, particularly in situations where this is difficult (e.g., dangerous or resource-intensive) to achieve with traditional training. To include stressors in VR, this paper thus presents the concept of so-called ‘stress cues’ for operationalizing stressors to augment training in VR simulations for the context of police work. Considering the level of complexity of police work and training, a co-creation process that allows for creative collaboration and mitigation of power imbalances was chosen to access the police officers’ knowledge and experience. We assert that stress cues can improve the training experience from the trainer’s perspective as they provide novel interaction design possibilities for trainers to control the training experience. E.g., by actively intervening in training and dynamically changing the interaction space for trainees which also improves the trainee’s experience. Stress cues can also improve the trainee’s experience by enabling personalizable and customizable training based on real-time stress measurements and supplementing information for improved training feedback.

Music streaming platforms are currently among the main sources of music consumption, and the embedded recommender systems significantly influence what the users consume. There is an increasing interest to ensure that those platforms and systems are fair. Yet, we first need to understand what fairness means in such a context. Although artists are the main content providers for music platforms, there is a research gap concerning the artists’ perspective. To fill this gap, we conducted interviews with music artists to understand how they are affected by current platforms and what improvements they deem necessary. Using a Qualitative Content Analysis, we identify the aspects that the artists consider relevant for fair platforms. In this paper, we discuss the following aspects derived from the interviews: fragmented presentation, reaching an audience, transparency, influencing users’ listening behavior, popularity bias, artists’ repertoire size, quotas for local music, gender balance, and new music. For some topics, our findings do not indicate a clear direction about the best way how music platforms should act and function; for other topics, though, there is a clear consensus among our interviewees: for these, the artists have a clear idea of the actions that should be taken so that music platforms will be fair also for the artists.

The recently proposed restorative environments have the potential to restore attention and help against fatigue, but how can these effects be verified? We present a novel measurement method which can analyze participants’ speech signals in a study before and after a relaxing experience. Compared to other measurements such as attention scales or response tests, speech signal analysis is both less obtrusive and more accessible. In our study, we found that certain time- and frequency- domain speech features such as short-time energy and Mel Frequency Cepstral Coefficients (MFCC) are correlated with the attentional capacity measured by traditional ratings. We thus argue that speech signal analysis can provide a valid measure for attention and its restoration. We describe a practically feasible method for such a speech signal analysis along with some preliminary results.

User reviews have become an important source for recommending and explaining products or services. Particularly, providing explanations based on user reviews may improve users’ perception of a recommender system (RS). However, little is known about how review-based explanations can be effectively and efficiently presented to users of RS. We investigate the potential of interactive explanations in review-based RS in the domain of hotels, and propose an explanation scheme inspired by dialogue models and formal argument structures. Additionally, we also address the combined effect of interactivity and different presentation styles (i.e. using only text, a bar chart or a table), as well as the influence that different user characteristics might have on users’ perception of the system and its explanations. To such effect, we implemented a review-based RS using a matrix factorization explanatory method, and conducted a user study. Our results show that providing more interactive explanations in review-based RS has a significant positive influence on the perception of explanation quality, effectiveness and trust in the system by users, and that user characteristics such as rational decision-making style and social awareness also have a significant influence on this perception.

The interdisciplinary field of explainable artificial intelligence (XAI) aims to foster human understanding of black-box machine learning models through explanation-generating methods. Although the social sciences suggest that explanation is a social and iterative process between an explainer and an explainee, explanation user interfaces and their user interactions have not been systematically explored in XAI research yet. Therefore, we review prior XAI research containing explanation user interfaces for ML-based intelligent systems and describe different concepts of interaction. Further, we present observed design principles for interactive explanation user interfaces. With our work, we inform designers of XAI systems about human-centric ways to tailor their explanation user interfaces to different target audiences and use cases.

The interdisciplinary field of explainable artificial intelligence (XAI) aims to foster human understanding of black-box machine learning models through explanation-generating methods. In practice, Shapley explanations are widely used. However, they are often presented as visualizations and thus leave their interpretation to the user. As such, even ML experts have difficulties interpreting them appropriately. On the other hand, combining visual cues with textual rationales has been shown to facilitate understanding and communicative effectiveness. Further, the social sciences suggest that explanations are a social and iterative process between the explainer and the explainee. Thus, interactivity should be a guiding principle in the design of explanation facilities. Therefore, we (i) briefly review prior research on interactivity and naturalness in XAI, (ii) designed and implemented the interactive explanation interface SHAPRap that provides local and global Shapley explanations in an accessible format, and (iii) evaluated our prototype in a formative user study with 16 participants in a loan application scenario. We believe that interactive explanation facilities that provide multiple levels of explanations offer a promising approach for empowering humans to better understand a model’s behavior and its limitations on a local as well as global level. With our work, we inform designers of XAI systems about human-centric ways to tailor explanation interfaces to end users.

Software that uses Artificial Intelligence technology like Machine Learning is becoming ubiquitous with even more applications ahead. Yet, the very nature of these systems has made it very hard to understand how they operate, creating a demand for explanations. While many approaches have been and are being developed, it remains unclear how strong this demand is for different domains, application types, and user groups. To assess this, we introduce a novel survey scale to quantify the demand for explainability. We also apply this scale to an exemplary set of applications, novel and traditional, in surveys with 212 participants, showing that interest in explainability is high in general for intelligent systems but also traditional software. While this validates the heightened interest in explainability, it also reveals further questions, e.g. where we can find synergies or how intelligent systems require different explanations compare to traditional but equally complex software.

Nowadays, search engines, social media or news aggregators are the preferred services for news access. Aggregation is mostly based on artificial intelligence technologies raising a new challenge: Trust has been ranked as the most important factor for media business. This paper reports findings of a study evaluating the influence of manipulations of interface design and information provided in the context of eXplainable Artificial Intelligence (XAI) on user perception and in the context of news content aggregators. In an experimental online study, various layouts and scenarios have been developed, implemented and tested with 266 participants. Measures of trust, understanding and preference were recorded. Results showed no influence of the factors on trust. However, data indicates that the influence of the layout, for example implicit integration of media source through layout structuration has a significant effect on perceived importance to cite the source of a media. Moreover, the amount of information presented to explain the AI showed a negative influence on user understanding. This highlights the importance and difficulty of making XAI understandable for its users.