Listen
Neueste Veröffentlichungen
- ItemDocumenting Computer Vision Datasets: An Invitation to Reflexive Data Practices(2021) Miceli, Milagros; Yang, Tianling; Naudts, Laurens; Schüßler, Martin; Serbanescu, Diana; Hanna, AlexIn industrial computer vision, discretionary decisions surrounding the production of image training data remain widely undocumented. Recent research taking issue with such opacity has proposed standardized processes for dataset documentation. In this paper, we expand this space of inquiry through fieldwork at two data processing companies and thirty interviews with data workers and computer vision practitioners. We identify four key issues that hinder the documentation of image datasets and the effective retrieval of production contexts. Finally, we propose reflexivity, understood as a collective consideration of social and intellectual factors that lead to praxis, as a necessary precondition for documentation. Reflexive documentation can help to expose the contexts, relations, routines, and power structures that shape data.
- ItemSurfing in sound: Sonification of hidden web tracking(Georgia Institute of Technology, 2019) Lutz, Otto Hans-Martin; Kröger, Jacob; Schneiderbauer, Manuel; Hauswirth, ManfredWeb tracking is found on 90% of common websites. It allows online behavioral analysis which can reveal insights to sensitive personal data of an individual. Most users are not aware of the amout of web tracking happening in the background. This paper contributes a sonification-based approach to raise user awareness by conveying information on web tracking through sound while the user is browsing the web. We present a framework for live web tracking analysis, conversion to Open Sound Control events and sonification. The amount of web tracking is disclosed by sound each time data is exchanged with a web tracking host. When a connection to one of the most prevalent tracking companies is established, this is additionally indicated by a voice whispering the company name. Compared to existing approaches on web tracking sonification, we add the capability to monitor any network connection, including all browsers, applications and devices. An initial user study with 12 participants showed empirical support for our main hypothesis: exposure to our sonification significantly raises web tracking awareness.
- ItemPower, roles and adding value. Reflecting on the challenges of bridging across research and action on an international community networking project(2020) Gaved, Mark; Calderón Lüning, Elizabeth; Unteidig, Andreas; Davies, Gareth; Stevens, JamesThe three year EU-funded MAZI research project (www.mazizone.eu) brought together universities, civil society organizations, and neighbourhood groups to design, develop and trial a digital toolkit for supporting local sustainability in four European countries. Funder constraints, partner ambitions and community needs had to be balanced to both adhere to academic research protocols while making a difference in the neighbourhoods where research and action took place. These sometimes conflicting ambitions caused partners to continuously question whose agendas were best being served by the project activities. They had to confront asymmetries of power, capacity, and credibility both within the consortium and within the community settings. Local circumstances changed; partners had to negotiate new, unfamiliar, and changing roles; and guises had to be adopted to progress sometimes conflicting ambitions. In this paper, we report on the challenges encountered in two of the pilot locations, Berlin and London. These two pilots were similar as they consisted at the outset of a university partner previously unconnected to the locality, working with a civil society partner that was deeply embedded in the setting though long-term engagement. In both cases, the pairings sought to work closely together both on the ground and in research tasks. Finding acceptable compromises stimulatedconsiderable self-reflection and required ongoing negotiation. We offer insights on the potentials and pitfalls of civil society activistsand academic researchers collaborating within a research framework from the perspectives of both, with the goal of building a bridge of understanding between these two viewpoints
- ItemFrom Data to Discourse. How Communicating Civic Data Can Provide a Participatory Structure for Sustainable Cities and Communities(Mid Sweden University, 2021) Shibuya, Yuya; Hamm, Andrea; Raetzsch, ChristophThis study explores how Civil Society Organizations (CSOs) have leveraged civic data to facilitate democratic participatory structure for sustainability transitions around the case of bicycle counters in three US cities over a ten-year period (Seattle, San Francisco, Portland). We identified that CSOs have played crucial roles in public discourse by (1) sustaining long-term public issues through shaping affective as well as analytical discourses and (2) fostering citizens’ sense of ownership and contributions toward sensor devices and the data they generate by contextualizing them through local civic life as well as connecting issues to actors in other cities.
- ItemUnexpected Inferences from Sensor Data. A Hidden Privacy Threat in the Internet of Things(Springer International Publishing, 2019) Kröger, Jacob; Strous, Leon; Cerf, Vinton G.A growing number of sensors, embedded in wearables, smart electric meters and other connected devices, is surrounding us and reaching ever deeper into our private lives. While some sensors are commonly regarded as privacy-sensitive and always require user permission to be activated, others are less protected and less worried about. However, experimental research findings indicate that many seemingly innocuous sensors can be exploited to infer highly sensitive information about people in their vicinity. This paper reviews existing evidence from the literature and discusses potential implications for consumer privacy. Specifically, the analysis reveals that certain insufficiently protected sensors in smart devices allow inferences about users’ locations, activities and real identities, as well as about their keyboard and touchscreen inputs. The presented findings call into question the adequacy of current sensor access policies. It is argued that most data captured by smart consumer devices should be classified as highly sensitive by default. An introductory overview of sensors commonly found in these devices is also provided, along with a proposed classification scheme.