Resolve As You Go With Your Friends:
Allowing Communities to Collaboratively Resolve Data Conflicts
==============================================================

Yannis Katsis (UCSD)
(joint work with Alin Deutsch, Yannis Papakonstantinou and Vasilis Vassalos)


A quick look at real-world data unveils that data are often conflicting:
In contrast to what our database education has led us to believe, data
do not always satisfy the intended integrity constraints. Instead they
may violate them leading to database instances that are "inconsistent"
w.r.t. those constraints. For instance when integrating person data from
different sources, it is almost certain that we end up with conflicting
information about the same person (i.e. two tuples that share the same
primary key value).

On the other hand, most applications nowadays treat conflicting
data as bugs that have to be avoided at all costs. They usually go to
great lengths to ensure that the data store only contains
non-conflicting data.

In this talk we take a different approach with Resolve As You Go With Your
Friends; a system that allows user communities to manage and resolve data
conflicts. Conflicting data are are not a bug anymore; they are a feature
and they have their own lifecycle: a) They can be introduced into the
system, b) they can be queried and c) they can be resolved by the community
members either collaboratively or in isolation (e.g. whenever there is a
disagreement of opinions between different users).

Have you ever felt the need for a different wikipedia, where you will
be able to see not only a single "blessed" version of the article, but all
alternative opinions that people have about it and where you can help resolve
the conflicts? If the answer is yes, you should not miss the talk.

PS: Neither should you if the answer is no...