Workshop on Biases in Social Computing Data and Technology
Co-located with the 2nd European Symposium on Societal Challenges in Computational Social Science (EuroCSS): Bias and Discrimination, Cologne, Germany
Workshop: December 5th, 2018, 2:30pm – 6pm, URL: http://jdiesnerlab.ischool.illinois.edu/calls/biasescss2018.html
Conference: December 5-7, 2018, URL: http://symposium.computationalsocialscience.eu/2018/
Collecting, preparing, and analyzing digital traces and other human-centered data involves or even requires researchers to make a plethora of choices that can impact and bias research outcomes and conclusions. These choices refer to sampling, representation and provenance of data, experimental design, and the selection and configuration of algorithms, methods, and tools. While these decisions are increasingly embedded in datasets and technologies, in many cases, there is an insufficient understanding of their impact on research outcomes, and a lack of best practices and norms for documenting and communicating these choices. Additionally, artificial intelligence, machine learning, and data mining systems might unintentionally pick up on and further disseminate or reinforce biases in society and data. As a result, researchers using datasets and tools often do not have the means to determine whether the data and tools they are using are representative and reliable enough to provide a solid foundation for your research. These possible sources of bias highlight the importance of validation, transparency, and replicability/reproducibility.
Topics of interest -- as they relate to biases
- Quality and Reliability of Big and/or Social Data
- Data Provenance
- Data Governance
- Education and Information (at all levels, including K-12, higher education, educating/informing the public)
- Social Implications
- Validation (including possible biases in validation outcomes)
- FATE (Fairness, Accountability, Transparency, Equity/Ethics)
WORKSHOP FORMAT
The workshop will consist of three one-hour panels/discussion rounds on current issues related to biases: 1) technical/computational challenges and solutions, 2) social implications and ethical concerns, and 3) policy and regulations. Speakers/discussants will be selected based on their abstracts (submission details below). In each panel, speakers will briefly introduce their work or point of view (about 3 speakers per panel and 10 minutes per speaker), and we will then discuss the topic of each panel in detail, including taking questions from the audience.
WORKSHOP AGENDA
2:30-3:30pm | Technical/computational challenges and solutions
- Chair: Wouter van Atteveldt
- Hyunjin Song, Petro Tolochko, Jakob-Moritz Eberl and Hajo Boomgaarden: When Does Garbage Start to Stink? Imperfect Gold Standards and the Validation of Automated Content Analysis [pdf]
- Rob Heyman, Cora Van Leeuwen and Myriam Sillevis Smitt: From methods to increase internal transparency to algorithm package leaflets for improving accountability towards (re-)users [pdf]
- Discussants: Antske Fokkens
3:45-4:45pm | Social implications and ethical concerns
- Chair: Jana Diesner
- Arpita Biswas, Marta Kolczynska, Saana Rantanen and Polina Rozenshtein: Algorithms, fairness, and race: Comparing human recidivism risk assessment with the COMPAS algorithm [pdf]
- Merja Mahrt: Biases in platform data and their implications for communication research and other social sciences [pdf]
- Discussants: Madhu, Wouter van Atteveldt
5:00-6:00pm | Governance and Policy
- Chair: Antske Fokkens
- Kevin Donnelly: Deviance and Bias in Nineteenth-Century Social Physics and Anthropometry [pdf]
- Fariba Karimi: Structure of social networks can induce biases in algorithmic rankings [pdf]
- Discussants: Alessandro Provetti, Jana Diesner
WORKSHOP CHAIRS
- Wouter van Atteveldt, VU University Amsterdam
- Jana Diesner, University of lllinois at Urbana-Champaign
- Antske Fokkens, VU University Amsterdam
PROGRAM COMMITTEE
- Carlos Edmundo Arcila, University of Salamanca
- Nigel Bosch, University of Illinois at Urbana-Champaign
- Chieh-Li (Julian) Chin, University of Illinois at Urbana-Champaign
- Jinseok Kim, University of Michigan Marijn Koolen, Huygens Institute for the History of the Netherlands
- Michaela Maier, University of Koblenz · Landau
- Malvina Nissim, University of Groningen
- Alexandra Olteanu, IBM Research
- Katrin Weller, GESIS – Leibniz-Institut für Sozialwissenschaften
For any other question, feel free to contact us at biasescss2018@easychair.org
CALL FOR EXTENDED ABSTRACTS -- Closed
- Submission deadline: November 18, 2018, 11:59pm (Anywhere on Earth)
- Notification: November 25, 2018, 11:59pm (Anywhere on Earth)
Registration details are available at https://www.eventbrite.com/e/2nd-european-symposium-on-societal-challenges-in-computational-social-science-eurocss-tickets-48695663039. Authors of accepted submissions will also be notified about how to register for the symposium even after the official registration has closed.
Researchers are welcome to submit abstracts of work in progress or work that has already been published. We will use a single blind review process.
Submit your abstract (in English) through EasyChair: https://easychair.org/conferences/?conf=biasescss2018. Submissions will mainly be evaluated based on their relevance for the workshop and their potential to stimulate interesting discussions. Submissions should be abstracts of 500 to 1000 words, plus references and figures (if applicable, not counting towards word limit) summarizing the work (for empirical work) or point of view (for vision and position papers) to be presented and discussed. Each abstract will be reviewed by a multidisciplinary Program Committee.
Accepted submissions will be non-archival, i.e., there are no proceedings. We will publish the names of authors and titles on the workshop website.
Submit your abstract (in English) through EasyChair (https://easychair.org/conferences/?conf=biasescss2018) by November 18, 2018, 11:59pm (AOE)
If you have any questions, please contact biasescss2018@easychair.org