Probing a Collection to Discover Its Language Model
Abstract
Most solutions to distributed IR rely on access to a language model for each text collection, but it has been unclear how the model can be obtained reliably in real-world distributed environments. This paper proposes a solution based upon probing the collection and demonstrates its effectiveness on four databases.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1998
- Accession Number
- ADA478103
Entities
People
- Aiqun Du
- Jamie Callan
Organizations
- University of Massachusetts Amherst