File Format Corpus File Format Benchmark

1
File Format Corpus /File Format Benchmark
2
Background I

• Content complete
• "A set of files, which cover between them the
  widest variety of possible types of content."
• Purpose
• Test organisational procedures

3
Background I

• Content complete
• Feature complete
• "A set of files, which cover between them the
  widest variety of possible features of a given
  format."
• Example Baseline TIFF as well as tiled files
• Purpose
• Test completeness of implementation of a format

4
Background I

• Content complete
• Feature complete
• Performance defining
• "A set of files, which is sufficiently large, so
  two or more programs processing them can
  meaningfully be compared in their performance."

5
Background II

• Role of Cologne in Planets "Preservation
  Characterisation"
• XCEL "A formal language, which allows the
  expression of all existing file formats in such a
  way, that an interpreter of a file format
  expressed in this language can parse any file of
  that format."

6
(No Transcript)
7
(No Transcript)
8
(No Transcript)
9
(No Transcript)
10
(No Transcript)
11
Background II

• Therefore
• UzK will create library of file format documents
  as part of its work in PC.

12
Proposal

• (1) Augment this library of format documents by a
  feature complete / performance defining benchmark
  / corpus.

13
Proposal

• Start with / concentrate on
• (2) File types which form PLANETS PC - PA
  baseline.
• (TIFF v. PNG PDF v. RTF (LATeX) WAV v. MPG)

14
Proposal

• Start with / concentrate on
• (3) File types which are recommended by "widely
  known preservation recommendations".
• http//www.fcla.edu/digitalArchive/pdfs/recFormats
  .pdf
• (Florida Centre for Library Automation)
• http//www.digitalpreservation.gov/formats
• (Library of Congress)