III: Small: Personalized Inconsistency Resolution in Online Databases

Supported by NSF award IIS:1117527

PI: Alin Deutsch

Co-PI: Yannis Papakonstantinou

Students & postdocs:

Yannis Katsis (postdoc)

Abstract:

The RICOLLA project manages inconsistencies in structured databases maintained collaboratively by an online community. RICOLLA remains fully functional in the presence of inconsistencies, enabling "resolve-as-you-go" consistency. RICOLLA allows users to collaboratively resolve certain conflicts while disagreeing on others. Building Ricolla involves the following technical contributions: a) a novel architecture that tolerates inconsistency, allows data query and update, while aiding inconsistency resolution by community members; b) a data model and interface for explaining the inconsistencies to the users; c) a set of resolution actions that allow each user to resolve individual data inconsistencies; d) a resolution policy language for summarizing a set of resolution actions based on high level criteria; and e) a set of algorithms for implementing the system on top of a relational database management system.

The resulting techniques and prototype contribute to the infrastructure for the next generation of online databases. This benefits a variety of online communities who need to collaboratively edit structured data, ranging from the scientific domain to digital government and social networks. RICOLLA's evaluation includes as use cases two scientific communities (biologists and geoscientists) and UCSD students taking database classes. Direct deployment in teaching serves to both improve students' online collaboration and collect their feedback for RICOLLA's evaluation and tuning purposes. Publications, technical reports, software and experimental data resulting from this research are available at the project web site http://db.ucsd.edu/ricolla.

Findings

Publications:

[TR2012] Yannis Katsis, Alin Deutsch, Yannis Papakonstantinou, Vasilis Vassalos “Conflict Resolution in Online Databases  UCSD technical report.

[BC2013] Alin Deutsch and Richard Hull. Provenance-Directed Chase&Backchase. In Search of Elegance in the Theory and Practice of Computation LNCS Volume 8000, pp 227-236                                                                                                                       

 

[C2014]    Ioana Ileana, Bogdan Cautis, Alin Deutsch, Yannis Katsis: Complete yet practical search for minimal query reformulations under constraints. SIGMOD Conference 2014: 1015-1026

 

This project web page: http://db.ucsd.edu/ricolla