Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.
Comment: Corrected links that should have been relative instead of absolute.

See Quick Reference Version for a user-friendly guide to file naming.

We have borrowed extensively from the IU Digital Library Program http://wiki.dlib.indiana.edu/confluence/display/INF/Filename+Requirements+for+Digital+Objects.

This best practice will only apply to files that the Libraries process and therefore does not include self-submission file names. This could be used to consult on best practices but will not be mandatory for self-submission.

MIT Libraries File Naming Scheme

As we create new collections, it is useful if there is consistency in the filenames assigned to digital objects. Each collection will impose its own restrictions on filenames, but following these requirements will ensure basic consistency across collections and make later processing of the files much easier.

In general, we won't change the file names on born-digital material unless the names violate the Requirements and Best Practices below.

The need for standardized filenames

The needs to be fulfilled by enforcing requirements on filenames are (in order of importance):

  1. Ease of identification. During the life of a digital file, it moves through various locations. These moves may be due to manual processing or automatic processing. A file may stay in a given location for an arbitrary length of time, during which human memory of its origin fades. At all points in the life of the file, humans and automatic processes must be able to easily identify which digital object the file belongs to and the position it occupies within that object. As a side effect of this, the filename will facilitate the process of locating metadata about the file.
  2. Automatic processing systems must be able to make basic assumptions about the filenames they will process. Developers should not be concerned with complex processing to handle special characters. (Although as a security matter, it is good practice to verify that a filename conforms to these requirements before initializing an automatic process.)

Requirements for filenames

Absolute requirements

All files must conform to the following requirements:

  1. Each filename must contain an identifier that uniquely specifies a single digital object within the parent collection.
  2. If a digital object consists of multiple files, each filename must contain the object's identifier, along with a unique sequence number.
  3. Each filename must be fully specified. It cannot just be a sequence number that is dependent on location within a directory structure for context.
    Rationale: Files are often moved between locations for processing or testing. It is not always feasible to move an entire directory structure with the file, so all necessary information must be in the filename itself.
  4. Filenames must not include spaces.
    Rationale: There are many instances where using a space in a filename can cause programs to misbehave. Automatic processing as well as human access to the file becomes more difficult when spaces are involved.
  5. The first character of the filename must be an ASCII letter ('a' through 'z' or 'A' through 'Z').
    Rationale: Many programming and metadata languages place this restriction on their identifiers. Filenames should be usable as identifiers in these languages (e.g., section ID's in a METS document).
  6. The "base" filename may include only ASCII letters ('a' through 'z' and 'A' through 'Z'), ASCII digits ('0' through '9'), hyphens, underscores, and periods. No other characters are permitted.  See "Best Practices" regarding the use of periods and uppercase letters.
    Rationale: Characters from other character sets can be difficult to read, depending on program support and available fonts. Many operating systems and programs are unable to correctly process non-ASCII characters. Punctuation and other ASCII characters not listed here may have special meanings, depending on the context; files using these characters may cause unexpected problems.
  7. The "base" filename must be followed by a single period and a suitable extension to specify the type of file. The extension should consist of three letters (e.g., jpg, txt, xml, tif), but longer extensions are permissible if they are widely used (e.g., html, tiff, djvu, aiff).
    Rationale: Whenever possible, the extension should make sense to a human. On systems where the file extension dictates automatic behaviors, the file should exhibit the expected behavior.
  8. A derivative file must have the same name as the master file, except the "base" filename should have an indication of the derivative's type appended (e.g., "full" or "screen" for images, an indication of the bitrate for audio files). Derivative files will typically have a different file type, and therefore a different extension, than the master file.
    Rationale: It should always be easy to identify files with master-derivative relationships.
  9. Directory (folder) names should not include periods.
  10. Document the file naming scheme for each project and explain the decisions that were made. (U South Carolina)

Best practices

The following "best practices" should be followed whenever possible. If one of these practices is not followed, the change should be well documented, with a description of the reasons for not following the practice.

  1. When creating filename standards for a new collection, the standards should be based on existing collections/objects with similar characteristics.
    Rationale: Minimizing the variability in filename standards eases both automatic and manual processing.
  2. Portions of the filename should indicate more specific detail as they are read from left to right. That is, the far left portion of the filename should indicate the collection name or home library of the item, the next portion should indicate the subcollection or aggregation, followed by a piece (page/section) number, and ending with the indication of derivative size. (Any of these portions that do not apply to the current file may be omitted.)  What if there is more than one home library? Or more than one home organization, e.g. some Sloan Working Papers are also part of the SCISR publication series?
    Rationale: Alphabetical listings of files make more sense with this organization.
  3. Whenever possible, the digital object's "primary" identifier (the identifier appearing in the filenames) should correspond to an identifier in use for the original (physical) object, such as the official or unofficial collection name or Archives collection number. If the format of the primary identifier conflicts with the absolute filename requirements, appropriate changes should be made. If the format of the primary identifier conforms to the absolute filename requirements but violates best practices, it may be left intact. 
    Rationale: It should be easy to determine the relationship between digital files and physical objects. This is easier if the identifier in the filename is as similar as possible to the identifier associated with the physical object.
  4. Page numbers should be padded with leading zeros so that all filenames in a collection have the same number of characters for the page number portion. In most cases, this will be two or three digits.  When determining the number of characters, consider how the collection might grow and the number of loose pieces, foldouts, or other physical elements that may amount to more than one image per page.
    Rationale: This forces pages to display in the correct order when listed alphabetically, and provides more visual consistency when scanning a long list of files.
  5. While periods are permissible in "base" filenames, it is highly recommended that they be avoided.
    Rationale: Some programs assume that there is only a single period in a filename, and will behave strangely if multiple periods are present.
  6. It is preferable that all letters in a filename be lowercase. If a filename includes consecutive human-readable words, they may be denoted by CamelCase (e.g., wnp-04-RoyalSociety-ncn-t123.tif). This is expected to be relatively rare, though.
    Rationale: Lowercase letters aid human readability and make it easier to type the filename. In collections where filenames contain many human-readable words, CamelCase aids readability.
  7. Distinct portions of the filename should be separated by underscores.
    Rationale: Separating the portions makes the filename both easier to read and easier to process automatically. Note that it is reasonable for the "identifier" portion of the filename to retain hyphens in identifiers from external sources, as in ihs-SHMU_01_13-01-05.tif. This reduces confusion when locating items provided by other institutions.
  8. File names should be limited to 31 characters or fewer (including the period and file extension).  Total path length (directories + file name) should not exceed 256 characters. (BCR-CDP; controlled vocabulary.com)
  9. While it is permissible for two different collections to contain files with identical names, this should be avoided.
    Rationale: It will not be possible to know of all filenames in use. Nonetheless, identical names can be confusing, and care should be taken to reduce the probability of identical names.
  10. For derivative files intended primarily for Web display, one consideration for naming is that images may need to be cited by users in order to retrieve other higher-quality versions.  If so, the derivative file name should contain enough descriptive or numerical meaning to allow for easy retrieval of the original or other digital versions. (NARA)
  11. Derivative names are based on Stellar: cp=class projection (ideal for Flickr), sv="screen view" (viewing and printing), tm="thumbnail"

Sample filenames

Collection

Filename

Notes

Edgerton Collection

MC025_nb41_017.tif

MC025 is the collection number for the Edgerton Collection in the Institute Archives.  nb stands for notebook; nb41 stands for notebook #41.  017 stands for the 17th sequential image of notebook #41 (which may or may not be exactly page #17).  In this case, the source for the digital image is the notebook itself.

Edgerton Collection

MC025_nb41-mf_017.tif

As above, except that the source for the digital image is the microfilm of the notebook.  nb41-mf stands for the microfilm (mf) of notebook #41.  017 stands for the image of the 17th whole frame on the microfilm.

Edgerton Collection

MC025_nb41-mf-split_017.tif

As above, except that the digital images have been split and cropped so that they no longer represent a whole frame from the microfilm.  nb41-mf-split stands for split images made from the microfilm of notebook #41.  017 is the 17th sequential image.

Edgerton Collection

MC025_nb41-mf-split_017-tm.jpg

As above, except this is a derivative file.  017-tm stands for the thumbnail (tm) of the 17th image made from splitting digital images of the microfilm frames. 

Edgerton Collection

MC025_nb41_017-tm.jgp

A derivative file based on Example 1, at top of chart.  017-tm stands for the thumbnail (tm) of the 17th image.  The source for the digital image is the original notebook.

Archives collections

MC####_123456_001.tif

MC#### stands for the collection number. 123456 stands for a random 6 digit number generated by the archives.  See Institute Archives and Special Collections page for more information.

Multi-volume item from general collection

barker_TrAmSocStTr_v001_0001.tif

barker is the home library collection for this item.  TrAmSocStTr is an abbreviation of the title: Transactions of the American Society for Steel Treating.  v001 stands for volume 1, and 0001 stands for the first image of this volume.  The Aleph system number (00291693) would be in the metadata. 

Book (Off Campus Collection)

science_PriOfRel_0064.tif

science is (sort-of) the home library collection for this item stored in LSA.  PriOfRel represents the title: The Principle of Relativity.  0064 stands for the 64th image of the book.  The source for the digital image is the book itself. The Aleph system number (001020855) would be in the metadata. 

Book (Off Campus Collection)

science_PriOfRel_0064-tm.jpg

As above, except this is a derivative file.  0064-tm stands for the thumbnail (tm) of image 0064.

What is Engineering? Freshman Lecture Series

WIE77_n02_pt01.mj2
WIE77_n02_pt02.mj2

WIE77 represents the home library collection: the What is Engineering freshman lecture series from 1977. n02 stands for the lecture number.  pt01 and pt02 stand for two Motion JPEG 2000 files, one for each part of the lecture.  (The original lecture was recorded on two tapes, parts 1 and 2, and this division has been maintained.)  [LIBCMS:Hypthetical example; these mj2 files do not exist]

What is Engineering? Freshman Lecture Series

WIE78_n02_pt01.mj2
WIE78_n02_pt02.mj2

WIE78 represents the home library collection for the What is Engineering? Freshman Lecture Series from 1978.  n02 stands for the lecture number.  pt01 and pt02 stand for two Motion JPEG 2000 files, one for each part of the lecture. [LIBCMS:Hypthetical example; these mj2 files do not exist]

What is Engineering? Freshman Lecture Series

WIE78_n03_pt01.mj2
WIE78_n03_pt02.mj2

WIE78 represents the home library collection for the What is Engineering? Freshman Lecture Series from 1978.  n03 stands for the lecture number. pt01 and pt02 stand for two Motion JPEG 2000 files, one for each part of the lecture. [LIBCMS:Hypthetical example; these mj2 files do not exist]

Perceptual Form of the City - actual file name used

KL_000123_02_sv.tif

KL stands for Kepes-Lynch, the unofficial name of this collection.  000123 is a sequential id number for the piece.  02 represents the second image from this item.  sv stands for "screen view," the type of derivative.

  • sv = screen view
  • tr =  transcript
  • tm = thumbnail
  • cp = classroom projection

Perceptual Form of the City - proposed file name if we could do it over

PFC_Boston_123456.tif
(items from archives: MC0208_IRIS image number.tif)

PFC stands for Perceptual Form of the City (collection name).  Boston differentiates this subcollection from images from New York City and other locations.  123456 is the IRIS image number.

Project Whirlwind

MC1234_reel#_report#.pdf

MC1234 represents the Archives' collection number.  The smallest unit is pdf because the images are already bundled into pdfs by report number.

Sloan Working Papers

SWP_WP#_report#_page#
SWP_oclc#_report#_page#

Sloan_WP#_report#+version_page#
Sloan_year_report#+version_page#

Prefer "Sloan" to SWP because "Sloan" is the name of the DOME community in which the pdfs would reside?
Using OCLC number does not allow for chronological sorting. Four-digit calendar year would work.

CEEPR

CEEPR_0090_paper#

 

RLE

RLE_report#_section#_page#

 




CSAIL

CSAIL_wps_WP#_page#

wps = working paper series?
CSAIL is a community in DOME.

Barker books

BarkerBks_title_page#
Barker_fund/collection_title_page#

No DOME community yet for general library collections.  Use LIBRARIES?  Use library name?  (But what if BARKER folds into SCIENCE?)

 

 

 

Directory Structure

When vendor is sending files on hard disk:

Shipment number (MIT generated)
    Collection number or Aleph system number, with volume number if applicable
             Image files, with derivatives interfiled
             checksum.md5
    Collection number or Aleph system number, with volume number if applicable
             Image files, with derivatives interfiled
             checksum.md5