09 Oct 2015

Orchestra: Facilitating Collaborative Data Sharing

[3 page] Orchestra: Facilitating Collaborative Data Sharing

http://repository.upenn.edu/cgi/viewcontent.cgi?article=1012&context=db_research

transaction — the basic unit of operation

a network of collaborators (participants), each of which has a local database …

each site spends the majority of time operating in a locally autonomous mode, with users posing queries and making modifications directly over a local …

upon a user’s request, … performs update exchange; …

two basic operations of update change are publication and reconciliation

reconciliation combines candidate … with … at any later point in time, the administrator can manually resolve the conflict between …

[Penn Ive, site] Orchestra: Managing the collaborative sharing of evolving data

map data from one database to another — where the database may potentially have different schemas and interfaces … once data is mapped, it is frequently modified in multiple places at once, and the challenge lies in “synchronizing” or “reconciling the modifications” …

project overview project focuses on the challenges of such data sharing scenarios in the sciences — …., similar but not identical data, differing levels of data quality / confidence, and a variety of different target audiences. … each database owner would like to store a “live” view of all relevant knowledge in its domain — however each site is being independently extended, corrected, and analyzed … unfortunately, there is often no consensus on what the best data is — certain data item will always be disputed or revised. our focus is on how to support reconciliation across different schemas, with disagreeing users. in general, each participant in the system specifies whom it trusts, and this is used to locally resolve conflict.

the orchestra collaborative data sharing system

http://repository.upenn.edu/cgi/viewcontent.cgi?article=1691&context=cis_papers

a single queryable (queriable) "mediated" data instance

fetch / publish update; transform (map) to → commit; update base table

  • stage 3:
  • stage 2: filter by local constraints; apply local curation / modification; reconcile conflicts; update made by … A get translated into B’s schema and applied …