ROMIP'2008 Tracks

A track in ROMIP is a separate section of the seminar devoted to one concrete task with strictly defined rules of evaluation of the participating systems. Participants don't have to take part in all the tracks at once - they choose one or several tracks which are most interesting for them, so they can concentrate on solving one of the proposed tasks.

According one of the ROMIP principles rules for each track rules are defined by the participants - the organizing committee only coordinates this process and the evaluation. You can track the tasks and rules definition discussion and of course participate in it, subscribing to the ROMIP mailing list (the discussion is currently led mostly in Russian).

ROMIP'08 tracks:

• Ad hoc search
  • in a collection of Web pages
  • in a collection of legal documents
  • in a mixed collection
• Classification
  • of legal documents
  • of Web sites
  • of Web pages
• News clustering
• Query-biased summarization
• Similar documents search (by a sample document or a text fragment)
• Fact extraction from news
• Image tracks
  • Content-based image retrieval
  • Near duplicates detection