MARC standard

From MARC must die!

Jump to: navigation, search
  • Obscurity - why use non meaningful "names" like 100 and 245?
  • Not used outside libraries
  • The storage of data is not separated from the presentation of the data

MARC format

MARC is the dominant data format used in libraries so that those who deal with library data will sooner or later encounter it. The abbreviation MARC stands for Machine Readable Cataloguing which points out at its original purpose to serve as a machine-readable representation of a record describing an information resource. Now, it is roughly 40 years old and it approaches or exceeds the end of its life-cycle. There are different flavours of MARC, the most widespread being MARC 21 or UNIMARC.

Even though the MARC is usually referred to as data format, it is rather a mark-up language for text. It was designed with printing its contents to a traditional catalogue card which means it contains some rudimentary presentation markup, such as punctuation signs to visually delimit different data elements.

When using MARC the data are segmented into records. For example, the library datasets often consist of bibliographic or authority records. MARC records structure data into fields and subfields which means that the data elements in MARC can be identified by field's tag, subfield's code, and sometimes by one or two indicators. The values of data elements do not have specified a datatype so their value must be always treated as unstructured text. The non-explicit semantics of MARC is the cause of multiple issues.

The lack of explicit structure that can be easily manipulated results in a situation in which the records must be considered as black-boxes from the viewpoint of a machine. In case of MARC records, the structured data that can be processed by an automated means are often hidden in between text. MARC does not have a schema describing the format and for that reason it cannot be charge for a self-describing format. By contrast, MARC needs for its interpretation an external data which can be read only by humans (e.g., handbooks). It turns out that MARC records work only in software that is specifically designed with MARC in mind. Even though the MARC offers a range of serializations, such as binary or XML, its semantics is always the same, based on the model of a paper catalogue card. MARC records do not work on the Web.

Personal tools