Human Computation @ XI
Overview
BigData is often the result of the continuous collection and processing of human generated data (e.g. Twitter, Facebook). Thanks to this massive influx of information, we can generate movie recommendations, create precise spell checkers, tag images at no cost, and even save lifes. But what if the crowd is not only generating data, but also helping in the processing of tasks where computers fail? What if this is achievable at scale?
One of the interests of our laboratory is to explore ways to make reliable, scalable and timely crowdsourcing systems. Our previous endeavours resulted in the following achievements:
- Designed a crowd reputation and aggregation mechanism, based on probabilistic networks (ZenCrowd).
- Introduced the concept of push-crowdsoucing and built a task recommender system for workers based on their social profile (Openturk)
- Designed several hybrid Human-Machine architectures to solve large scale data integration and retrieval problems. Namely: entity linking, instance matching and query understanding.
Research papers:
- Pick-A-Crowd: Tell Me What You Like, and I’ll Tell You What to Do, in WWW, 2013
- CrowdQ: Crowdsourced Query Understanding, in CIDR, 2013
- Large-scale linked data integration using probabilistic reasoning and crowdsourcing, The VLDB Journal, 2013
- Mechanical Cheat: Spamming Schemes and Adversarial Techniques on Crowdsourcing Platforms, CrowdSearch 2012 workshop at WWW
- ZenCrowd: leveraging probabilistic reasoning and crowdsourcing techniques for large-scale entity linking,, WWW 2012
Systems:
- The Openturk Chrome Extension for worker task recommendation. Available on the Chrome Webstore .