New Media Research Institute

Hot Topic: How do we encourage helpful and thorough tagging and metadata contributions? Why will people (scientists and public) use DO? Distributed vetting and assignment of credibility.

Cathy Boggs, Matt McClintock, Ron Rice, Peter Berliner, Will McClintock

Automation: Detailed profiles can provide the content for automated tagging of some objects. Credibility can partially be attributed by profiles (i.e., a scientists is automatically, at first, more credible than a high school student). Requires an entity management system. Make sure that system can take advantage of existing metadata tags (e.g., time stamps, geotags in photos). Automated extraction of information from the object that will tell you what it is (e.g., color info can tell you about the content).

Content level evaluation: Can also use the Amazon.com mechanical turk model. Many people look at the same data, repeatedly tag it, and you use the aggregate (or average) value to assign. Product reviews (e.g., was this helpful?).

System level evaluation: Was the search mechanism that you used in G.O. a good one? Was the tool you used to get here a useful one?

Have a thesaurus?

Link objects to Google Scholar or Web of Science.

Track object use as a proxy for "citations".

Must be centralized tools for governance.

Additional question: "How do you prevent malicious tagging?"