Human-Centric Challenges in Building and Using Structured Web Databases
             AnHai Doan, University of Wisconsin / Kosmix

Large structured Web databases (e.g., Freebase, Google Scholar, YAGO,
and concept taxonomies at many startups) are often built using a
combination of algorithm and human effort, over a variety of data
sources, such as relational databases, Web pages, blogs, and
tweets. Once built, the databases can be used to develop many
applications that let human users query, explore, mine, and monitor
the data, among others.

In this talk I will discuss human-centric challenges that arise from
building and using such databases, drawing from my work in the
Cimple/DBLife project at Wisconsin, and in the Social Genome project
at Kosmix. Example challenges include how to crowdsource the building
process, how to allow human users to contribute structured data as
well as text, how to grandma and grandpa query such databases, and how
to effectively use scarce human resources in detecting and monitoring
Twitter events. I conclude by discussing why humans will become an
integral part of data management processes and why it is important to
find principles that uncover how they behave as a part of such
processes.