The following projects have been accepted for the Academy Assistants 2016.
Virtual Bad: A Study on Virtual Agents that Physically Threaten Human Beings
Intelligent Virtual Agents (IVAs), interactive human-like characters, become widely used for numerous applications, varying from healthcare decision support2 to communication training. In such applications, IVAs play various roles in which they interact with users, for instance as an instructor or teammate4. Interestingly, in the vast majority of these cases, IVAs are friendly and supportive. Instead, the area of IVAs with a ‘negative’ attitude towards users (i.e., ‘virtual bad guys’) has been heavily under-researched.
However, ‘virtual bad guys’ are a highly interesting topic of study for at least two reasons:
1) Several prominent people recently expressed their concern that autonomous systems might evolve to a point where they threaten human beings. Controlled studies can provide a better understanding of how humans would react to such threatening AI systems.
2) The concept of virtual bad guys opens up a range of useful applications, including virtual training of aggression de-escalation skills (e.g., for security personnel), Virtual Reality exposure therapy, and anti-bullying education.
Constructive and Creative Co-Creation Campaigns
In web-based health campaigns, co-creation is a new persuasive strategy, enabling audience members to become active contributors to the campaign. Consider a recent campaign by the Dutch Cancer Foundation, which aimed to persuade its audience that non-smoking should be the social norm. It did so by asking audience members through social media to complete the slogan “Smoking is sóóó…” with something old-fashioned. While many audience members followed the campaign’s directive, some deviated and wrote a positive message about smoking (e.g., …nice) or a negative reflection on the campaign (e.g., …the ‘smoking is sóóó campaign’).
Such ambivalent responses are found for many co-creation campaigns, leading to the question when and how co-creation campaigns are successful (or not). Our project takes on this question and breaks new ground in two important ways. First, we take an interdisciplinary perspective and explore the role of both cues of the online environment (social sciences)
and linguistic cues (linguistics) in establishing the ways in which audience members co-create campaign slogans. Second, many co-creation studies are observations of cases of failed or successful co-creation. We propose to supplement such studies with a production experiment in which audience members co-create slogans in a controlled research environment.
Specifically, participants will be randomly exposed to one of several (fictitious) co-creation campaigns that differ in cues offered by the online environment and in linguistic cues used to trigger co-creation. This enables us to establish how these cues work (together) in generating audience responses, and help us to further entangle which elements constitute successful web-based co-creation campaigns.
|Christian Burgers||Romy van den Heerik|
Computer-led speech audiometric screening in toddlers
For young children access to speech is essential for the development of oral language. Reduced speech intelligibility due to congenital deafness or severe hearing loss increases the risk for permanent language impairment. In adults, such aspects of functional hearing are often measured through a ‘listen-and-repeat’-task based on sentences. However, speech test materials are not readily available for children under six. In this project, we want to address this issue by developing a new linguistically controlled speech audiometric screening instrument that is suitable for children from 2 to 6 years old and to gather normative data of hearing children in this age group.
During the last year, a pilot study was set up leading to a bèta-version of such an audiometric screening test. It was developed as a ‘picturepointing’ task using words and short sentences that are adapted to the developmental language level of the target group and includes diagnostic features. In this project, we would like to fine-tune and finalize our prototype so that it can be used in specialized audiological centers. In order to do so, a number of scientific and technical challenges need to be addressed, namely to determine (i) the exact contribution of non-auditory (linguistic) factors in speech understanding at this young age; (ii) the effect of picturepointing vs. verbal repetition on the reliability of test outcomes; and (iii) the beneficial effect of computer-led procedures on test compliance in very young children.
|Martine Coene||Michel Klein|
Strengthening aid accountability: Impact assessment of accountability initiatives following the Nepal earthquakes
The 2015 earthquakes in Nepal caused massive devastation, but also mobilized major aid efforts. Innovative GIS-mapping tools using mobile technologies (such as QuakeMap1) were
adopted to identify community needs. An important drawback of these tools is that they remain inaccessible to communities where connectivity is weak or absent. In response, the Nepalese social enterprise Accountability Lab2 established ‘mobile citizens helpdesks’, whereby volunteers map where aid efforts are needed, where these have been delivered, and which blank spots remain, combining handheld devices, community radio and community centers. However, a year on, it remains unclear to what extent these efforts actually impacted affected communities and actual accountability of aid to vulnerable communities remains unclear.
This project seeks to respond to this need, toward demarcating reliable impact indicators, implementing these, and developing possible improvements to ensure optimal accountability of aid efforts that reach the most vulnerable affected communities. Drawing on a social science perspective, the project will demarcate what we need to know in order to conduct a reliable impact study, studying existing data in terms of where aid efforts have been targeted and how/why these have or have not realized their aims, and ensuring sufficient information is generated during data collection. We aim at developing innovative and relevant interfaces, for example by developing voicebased and mobile interfaces.
|Julie Ferguson||Victor de Boer|
The vast scale and impact of disaster relief calls for joint efforts among the international community. This applies to disasters emerging from natural events (such as earthquakes) as well as from political conflicts (such as the stream of refugees fleeing war-torn areas). International disaster response is increasingly organized in the form of civil-military interventions, sanctioned through international law United Nations resolutions. Civil-military responses are often organized in a top-down, ‘command and control’ manner. However, at the same time, international disaster response often emerges in the form of bottom-up efforts, self-organized by affected citizens and their local and international networks, for instance through online, crowdsourced initiatives. However, the question as to how legal frameworks oriented toward civil-military interventions can accommodate these bottom-up, online initiatives remains unanswered. The lack of understanding of how legal parameters can better represent changing modes of disaster relief and vice versa lead to uncertainty among those responsible for international response work, inhibiting the effectiveness of response efforts. This proposed project seeks to address this deficit in understanding, toward the development of an interdisciplinary, legal-organization sciences approach toward international disaster response.
|Arno Lodder||Peter Groenewegen|
Generating expert and patient versions of medical guidelines from a single source
Evidence based clinical guidelines are systematically developed statements to assist practitioners and patient decisions about appropriate health care for specific circumstances. They provide clinicians with healthcare recommendations based on valid and up-to-date empirical evidence, they improve healthcare outcomes and reduce healthcare costs up to a 25%. Increasingly, such guidelines are not only developed by experts but also include the perspectives of patient organisations. Patient versions of clinical guidelines are increasingly developed to enhance shared decision making and to promote patient self-management.
The clinical guideline and its patient version are currently developed in isolation and the alignment of those versions is difficult both when developing and when updating the guideline.
We propose a method that is based on a generic computer model for guidelines from which we are able to generate a guideline version specific for different users (e.g. patients, experts, GPs and family members). We will investigate this method on a small scale by (1) modelling selected guideline fragments from expert and patient versions; (2) comparing the modelled fragments and (3) unifying these models in a single coremodel (4) generate different version depending on the needs and wishes of the accompanying stakeholder groups.
|Annette ten Teije||Violet Petit-Steeghs|
Effects of social network structures on the quality of knowledge in online communities
The Internet has fundamentally changed how work gets done in the 21st century. For example, people increasingly spend time on the Internet where they share and develop knowledge in online communities. Yet, little is known about how high-quality knowledge comes about in these communities. This is surprising, because stakeholders such as organizations, policy makers, or activist groups can profit from high-quality knowledge shared and produced in online communities. In this study, our goal is to test how social network structures influence the quality of knowledge developed in online communities. Our project will contribute to the literature on online communities and knowledge management, because we will conceptually develop and empirically test which social network structures are conducive for sharing and producing high-quality knowledge. As opposed to prior studies in this area, we will leverage automated techniques for our analysis. In particular, we will develop Natural Language Processing based techniques to recognize the substantial core and value of social media contributions. Besides the scientific contribution, our project will also deliver valuable insights for practitioners, which can aid in better managing online communities that are geared toward sharing and producing knowledge. We will use large data sets of three online communities to provide a ‘big data picture’. Our analysis will be based on a theory-based conceptualization of social network structures and knowledge quality. As a result of the automated analysis using the algorithms developed in the context of this project, we can automatically recognize high-quality knowledge based on a number of variables that are partly based on theory and partly developed using human coders. What is more, we can automatically extract the social network structures in the data. Based on these results, we will then test which social network structures are conducive for sharing and producing high-quality knowledge across the three data sets.
|Christine Moser||Henrik Leopold|
Shaping the image of stem cell research: Dynamic discourse analysis of Dutch TV media
Stem cell research is a rapidly developing field of science, investigating new possibilities for treating a wide range of serious health conditions. However, it is also highly controversial from the ethical, legal and social point of view. The role of media and press in the agenda building and framing of stem cell research is undeniable and has been discussed in numerous papers (but none regarding The Netherlands). However, media representation of stem cell research isn’t static but appears to change over time. The aim of proposed research project is to investigate changes in how stem cell research and its results have been presented in television media in The Netherlands. Drawing on data gathered from the museum Beeld en Geluid, we will examine how public discourse on stem cell has changed in the last 10 years, and which factors have played a role in this development. The dynamics of the discourse will be analyzed from two perspectives. First, a health-science perspective, focusing on how scientific developments affect the presentation of stem cell research in the media. Second, a philosophical perspective, focusing on the interaction between epistemic, social and moral values in public discourse. During the data collection stage, we will identify television programs from the major broadcasting agencies in The Netherlands that feature extensive discussion of stem cell research. The data will be analyzed using the discourse analysis approach. The project will result in a dynamical analysis of expert-to-public communication of stem cell research, that sheds light on how scientific developments interact with social and moral values in shaping the public image of controversial science.
|Elena Syurina||Henk de Regt|