Uploaded image for project: 'Terrier Core'
  1. Terrier Core
  2. TR-71

Allow BitInputStream structures to be split across multiple files

    Details

    • Type: Improvement
    • Status: Resolved
    • Priority: Minor
    • Resolution: Fixed
    • Affects Version/s: None
    • Fix Version/s: 3.0
    • Component/s: .indexing, .structures
    • Labels:
      None

      Description

      Inverted/direct files are traditionally contained in one single file. There is no reason for this to be the case. BitIndexPointer has two fields: long (for byte offset), and byte (for bit offset).

      A byte has range 0-255, but we only need 0-7. This means we could use the remainder of the byte to indicate which file a given pointers points to.

      Why? Well:
       * for MR indexing, we create one entire index per reducer. We could instead create one file of the inverted index per reducer.
       * for Inverted2DirectMultiReducer, we create one file of the direct index per reducer, but then have to join them later. This join is unnecessary.

        Attachments

          Activity

            People

            • Assignee:
              craigm Craig Macdonald
              Reporter:
              craigm Craig Macdonald
            • Watchers:
              2 Start watching this issue

              Dates

              • Created:
                Updated:
                Resolved: