Conflict Resolution in Online Databases*

Shared online databases, such as Google Fusion Tables or Quickbase, allow non-programmer community members to collaboratively maintain and browse data. While community members may believe in conflicting facts (due to conflicting sources, measurements or opinions), current online databases do not yet offer support for the management of data conflicts. Ricolla is a novel online database that treats data conflicts as first-class citizens. Unlike prior work in uncertain databases, which was made to provide a database back-end to application logic, Ricolla is tuned to the requirements of the online database paradigm, allowing intuitive visualization of conflicts and collaborative data editing/conflict resolution. The proposed end-to-end system makes the following contributions: a) an online database paradigm that captures conflicts, allowing data query and update, while enabling personalized, “as-you-go” conflict resolution, b) a data model and corresponding generic user interface for explaining the conflicts to the users in a simple and compact form that is aligned with the requirements of online databases; we formalize the “simplicity” and “compactness” requirements and show that Ricolla strikes a novel point in the tradeoff between simplicity, compactness and expressive power, c) personalized conflict resolution via resolution actions that allow each user to resolve individual data conflicts, and via a resolution policy language for performing a set of resolution actions based on high-level criteria (e.g., give priority to data based on their value, provenance or timestamps), and d) a set of algorithms for implementing the system on top of an RDBMS.