Faculty Candidate Presentation: Peter Organisciak
- SLIS Classroom 332
- Presentation Abstract: Crowdsourcing Metadata: Understanding Human Biases of Crowds in Information Systems
Crowdsourcing is increasingly used in the sciences and humanities for collecting hard-to-automate data at large scales. People can provide latent information about documents that would not be possible to ascertain computationally, and are adept at many abstract, qualitative, or critical actions. Paid online crowds are valuable in information retrieval for building critical training or evaluation sets, while engaged volunteer crowds are helping citizen science or cultural heritage systems in transcription, encoding, or document quality judging. Crowdsourcing considers these types of human contribution at newly accessible scales.
However, people have predictable and unpredictable biases that call for extra care when modeled in an information system. As we increasingly turn to online crowds, we need a better understanding of the nuances of what is collected. This talk discusses data reliability concerns around crowdsourcing, measuring the extent to which latent biases are magnified by the style of collection. This research is important for designing valid, reproducible crowd tasks. Additionally, personalized crowdsourcing is introduced, a mode of crowdsourcing that capitalizes on the subjective quirks of crowds rather than correcting for them.
Peter Organisciak is a doctoral candidate in library and information science at the University of Illinois. He also holds a master's in humanities computing/library and information studies from the University of Alberta and a bachelor's (honours) in communications studies and multimedia from McMaster University. He has served as a teaching assistant at the University of Illinois, completed an internship at Microsoft Research in Cambridge, Mass., and worked as a freelance developer. For more information, visit http://www.porganized.com/.