1. Introduction

After more than one year of practical work with the version 1.1, the structuring rules have been reviewed. Thus the version 2.0 has been created and then modified into the version 2.1. The structure proposal is divided into two parts: a general one described in [1] and its concrete application for various types of digital copies as manuscripts, old printed books, sheet music, poster collections, newspapers, sound recordings, etc. This is the concrete application for manuscripts and old printed books. 

The goal of this work remains the same: to create a standardized description of digital copies for use in the UNESCO Memory of the World programme. As the preceding version, also this one has been written on the basis of recommendations and remarks of the Memory of the World Sub-Committee on Technology. The proposal is designed to enable the reading of the digital copy in a standard WWW browser; therefore, its structure is based on the HTML version 2 [2] and it is enlarged with special tags marking important data which can be used for further work with the digital copy. This may be e.g. the use of special software for users or the automated building of a catalogue database of digital copies. The description of enlarging tags is contained in [1]. The formal documents are contained in the special DTD (DOBM_dec.txt, dobm_dtd.txt, dobmndtd.txt) written for this purpose [3] . 
 

Each digital copy of an old printed book or of a manuscript  must be presented in a certain structure which should correspond intuitively to the component parts of the original. In the same way as a book can be divided into chapters or a chapter into paragraphs, the description of a manuscript or of an old printed book can be divided into a bibliographic record, description of a book as a whole, and descriptions of individual pages. On the same level on which there are pages of manuscripts, there are also front and back covers, spine, reinforcing strips, enclosed sheets, and edges. 

Fig. 1 Component parts of the digital copy of a manuscript 

The bibliographic record of the original is stored in a separate file. Its name is arbitrary, but we shall call it DESCR.HTM. The record gives the basic information about the digitized document. The record statements are based on the definition of the first level of AACR2 [2]. Other statements are added to complete it with practical information about e.g. the character (type) of the document, its location, its owner, etc. The content of this file is described in the chapter 12. BIBLDESCR. 
 

The further unit in the metadata structure is the description of the book as a whole. This category called BOOK contains the data not included in the BIBLDESCR category. It contains also references to all the pages of the original and to related parts. 
 

Each page from the original corresponds to a file from the PAGE category. Here is only data  related to the page of the original. The mandatory elements are the categories PAGINATION or FOLIATION and references to the data - digital photographs of the original. The other elements as e.g. transcription of the text, its translation, description of the illumination, etc., can be freely added. It is also useful to place a preview image into the description of the page. For this purpose, the PREVIEWQ category is defined (see the chapter 10 Image data). 
 

The entire digital copy of a manuscript or of an old printed book consists of three types of descriptive parts. The first one contains information about the carrier on which the copy is written. It is an easily descridable part which will be necessarily changed only in case of migration of the copy onto another type of recordable media. At present, its content is defined for the CD-ROM. The second part in the SGML format defines the map of the structure of METADATA and that of the stored  DATA. This part will have to be modified in the case of restructuring of digital copies (e.g. adding of new statements to the description) or when the descriptive structure is merged into larger units. (For example, when integrating the digital copy into the structure describing a library of digital copies.) The third part consists of files in the enlarged HTML. Here, in a predefined skeleton, there is stored information concerning the digital copy itself. This part is not usually modified and neither the migration of copies onto other media nor the merging of copies into higher structures have any impact on it. The modification is needed, however, if new statements are added into the description. 

Fig.2  Structure of the digital copy   

Even if in principle it is not difficult to store copies on any data carriers, the actual production of digital copies concerns the storage of digital data on CD-ROM. It is for this type of media that the descriptive file is prepared. 
In the main directory of the CD-ROM there is the MNSXDEF.INF file. The content of the file consists of sections. In the DISC section there is the description of the carrier on which the copy is placed. It contains these keys: VERSION, NOOFDOCUMENTS. 
NOOFDOCUMENTS stands for the number of documents placed altogether on one CD. 
VERSION stands for the version of the description rules. It is mandatory for this version and its value is 2.10. The default value is 1.00. 

The following sections contain information about one or more digital copies. The sections are marked DOCUMENT_1, DOCUMENT_2, etc. The sections contain these keys: VERSION, TYPE, NOOFDISCS, CURRDISC, DOCID, ENTRYPOINT. 

The file is a simple text in the ASCII format, the diacritics are not permitted, and there is no distinction between low and upper case. 

Signification of the parameters: 
 

 

NOOFDISCS	-	number of discs on which the digital copy is placed. The default value is 1.
CURRDISC	-	if the copy is placed on several CD-ROM's, this key defines the order of the concrete CD-ROM.
DOCID	-	una mbiguous identification of the digital copy.
ENTRYPOINT	-	reference to the file containing the map of the digital copy
NOOFDOBMFILES	-	number of DOBM files belonging to the document on this CD-ROM. Mandatory.

For illustration here is an example: 

   [DISC]
   NOOFDOCUMENTS=1
   VERSION=2.1
   [DOCUMENT_1]
   TYPE=MANUSCRIPT
   DOCID=cz2Manuscriptnkp/nkmsxiiia6
   NOOFDISCS=1
   CURRDISC=1
   ENTRYPOINT=Manuscr.sgm
   NOOFDOBMFILES =322

If the copy is placed on more CDs, then there will be a copy of the descriptive structure on each CD. 
 

Each digital copy is defined unambiguously by the key DOCID. The format of the identification string is: the first two characters stand for the country following the ISO 3166 standard, then 1 to 5 letters follow which are composed from the regional telephone code of the town of the author of the digital copy, the following 3 to 8 letters mark the author of the digital copy (probably the abbreviation of the institution); the first part identifying the author is separated by a slash. The slash is followed by the identification string of the document (8 to 32 letters). It is recommended to use for example the shelf-number of the original document. In the whole string, it is permitted to use only these characters: A-Z, a-z, 0-9 a /. 
 

This SGML file contains the definition of the structure of the stored copy; it can be called a map. There are all the categories which are significant from the point of view of the digital copy. For each category, significant statements and their types can be predefined. The format of this file and the meaning of individual tags are described in [1]. The concrete application for manuscripts defines the categories and the statements corresponding to the character of the manuscripts and the needs of users. Some statements are mandatory, they cannot be omitted in the digital copy. There are also recommended statements which have a predefined format (category, name, and type). It is recommended to include them into the description if it makes sense for the concrete work. There are also other arbitrary statements which the author of the description of the copy can add following the below indicated rules. The map also defines the tree structure of the digital copy (see Component parts of the digital copy of a manuscript or of an old printed book). On the highest level in the hierarchy in this specification there is the DOBM category BIBLDESR. In this category there is metadata concerning the bibliographic record. From this category there are references to the description of the book as a whole (DOBM category BOOK) and DOBM category TECHDESCR describing the technological tools and technical methods related to the manufacturing of the digital copy. The BOOK category contains references to the descriptions of all the pages of the manuscript. The description of a page of the original is written in files of the PAGE category. This PAGE category should contain at least the pagination or foliation and one reference to the digital photograph of the page of the original. It is also possible to enlarge this category with other metadata and data, a thus to create a larger and more complete description of the individual pages. 

A typical example of such an SGML file is in the Supplement A. 

 Fig. 3 A scheme of the DOBM categories of pages 
 
 

Fig. 3 A scheme of the DOBM categories of pages   

The digital photographs can be presented in the structure of the description on various quality levels and in various contexts. Some of them serve as previews for the quick orientation of the user, the other ones are data meant for the display of  pages of the original in the quality depending on the used hardware or networks. 
 

 

GALLERYQ	-	very small thumbnail preview images enabling quick orientation in the gallery of images (category GALLERY). The recommended format is GIF, max.size 10 KB
PREVIEWQ	-	preview images of  pages of the original used in the description of individual pages (category PAGE). The recommended format is GIF, max. size 50 KB
INTERNETQ	-	this quality is defined for the on-line viewing of the digital copy on Internet. The limits of the file size are proposed to enable acceptable work with the digital copy during the normal traffic on Internet. The recommended format is JPEG, max. size 150 KB
NORMALQ	-	the quality for current needs of a greater part of users. The recommended formatis JPEG, resolution 180 dpi, the size of compressed images should be around 1 MB.
EXCELLENTQ	-	the highest quality enabled by the digital camera or  scanning device. The recommended format is JPEG, the resolution depends on that of the digital camera or scanning device.

 

Besides the images of the entire manuscript pages, it is also possible to include into the description the references to various types of interesting parts of  pages. For this purpose, there is a category called DETAIL. 

A special group is the category WATERMARK. It is the data containing previews of watermarks designs and colour attributes of the paper types. The recommended format is GIF, max. size 50 KB. This category is an alternative to the gallery type of images and it is used for the quick orientation within the manuscript from the point of view of material. 
 

The references to the images of individual pages of the original are marked by special tags. For each quality level a special tag is defined in the head of the file containing the description of the page (DOBM category PAGE): 
  

GALLERYQ	-	< DOBM.DATA CTGLABEL="GALLERYQ"
		NAME="Gallery Picture" TYPE=IMAGE>
PREVIEWQ	-	<DOBM.DATA CTGLABEL="PREVIEWQ"
		NAME="Preview Picture" TYPE=IMAGE>
INTERNETQ	-	<DOBM.DATA CTGLABEL="INTERNETQ"
		NAME="Internet Quality" TYPE=IMAGE>
NORMALQ	-	<DOBM.DATA CTGLABEL="NORMALQ"
		NAME="Normal Quality" TYPE=IMAGE>
EXCELLENTQ	-	<DOBM.DATA CTGLABEL="EXCELLENTQ"
		NAME="Excellent Quality" TYPE=IMAGE>
DETAIL	-	<DOBM.DATA CTGLABEL="DETAIL"
		NAME="Detail" TYPE=IMAGE>
WATERMARK	-	<DOBM.DATA CTGLABEL="WATERMARK"
		NAME="Watermark"TYPE=IMAGE>

For the example see the chapter PAGE category. 
 

This category is the root of the tree structure and it covers the digital copy of a manuscript or of an old printed book. The BIBLDESCR category contains a bibliographic record of the original document. The cataloguing language used in the digital copy should be English, while the language of the original is used to express those values where it is recommended by the cataloguing rules. Below there is a list of  categories prepared for encoding the statements of the bibliographic record: 
   

AACR2	- marks the statements based on the AACR2 standard [7]
orig	- the contents (values) of statements are in the language of the original
eng	- the contents are in the cataloguing language (mandatory in English, parallel national version possible)

  

MAINTTL	main title of the book, mandatory	AACR2	orig
FSTOFRESP	first statement of responsibility	AACR2	orig
EDITST	edition statement (1st, 2nd, ... edition, etc.)	AACR2	orig
GMD	general material designation, type of the document e.g. Manuscript, mandatory	AACR2	eng
PBLSHER	publisher	AACR2	orig
PLACEPBL	place of publication	AACR2	orig
PRINTER	printer	AACR2	orig
PLACEPRT	place of printing	AACR2	orig
DATOFPUBL	date of publication, mandatory	AACR2	eng
PHYSDESCR	physical description; it can contain the below indicated items	AACR2	eng
	MATERIAL	material from which the document is manufactured	eng
	SIZE	size	eng
	EXTENT	extent, number of pages or folios	eng
	ILLUM	illuminations (yes/no), possibly their very brief  characteristics	eng
	NOTATION	notes for music (yes/no, possibly their very brief characteristics	eng
NOTES	notes	AACR2	eng
ANNOT	annotation, obligatory		eng
It can contain the below indicated items:
	BASINF	basic information	eng
	CONTENTS	contents	orig/eng
	INDEX	index, list of the most important places in the document	orig/eng
	LIT	literature, bibliographic references	orig/eng
SHELFNO	shelf-number		orig
LIBRARY	library; place where the original is deposited		eng
OWNER	owner of the original if different from LIBRARY		eng

 

   Example of a typical BIBLDESCR file is in the supplement B. 

It is recommended to respect the English as a cataloguing language (shown above). In national versions, it is possible to replace English by the national language. The items in the language of the original are not translated in any national or English versions. 
     

The technical and technological information about the digitization of the original concerns the description of the technological devices, initial resolution, the fact whether the original was digitized directly or through colour slides, what kind of lights were used, etc. In the TECHDESCR category, this information is marked by the DX tag of the CAPTURE category. 

   <!DOCTYPE DOBM PUBLIC "-//AIP//DTD DOBM 2.1//EN">
   <DOBM CTGLABEL="TECHDESCR" NAME="Technical
     description"  SPEC="NKP//MANUSCRIPT 2.1">
   <HTML>
        .
        .
        .
  <DOBM.DX CTGLABEL="CAPTURE" NAME="Capture">Digitized
     by the Kodak DCS 460 camera.</DOBM.DX>
        .
        .
        .
   </HTML>
 

 

The BOOK category covers all the units of the PAGE category. There are references to all the files describing the individual pages of the manuscripts: 

    <!DOCTYPE DOBM PUBLIC "-//AIP//DTD DOBM 2.1//EN">
    <DOBM CTGLABEL="BOOK" NAME="Book"
      SPEC="NKP//MANUSCRIPT 2.1">
    <REFERENCE HREF="001r.htm" CTGLABEL="PAGE"
      NAME="Page">
    <REFERENCE HREF="001v.htm" CTGLABEL="PAGE" 
      NAME="Page">
    <REFERENCE HREF="002r.htm" CTGLABEL="PAGE"
      NAME="Page">
    <REFERENCE HREF="002v.htm" CTGLABEL="PAGE"
      NAME="Page">
    <REFERENCE HREF="003r.htm" CTGLABEL="PAGE"
      NAME="Page">
        .
        .
        .
    <HTML>
        .
        .
        .
    <!-- Body of the  HTML-->
        .
        .
        .

    </HTML>
 
 

Each page of the original is described in a file of the PAGE category. It is possible to place references to image data there, to previews, or descriptions of individual pages. The minimal requirement for this category is that it should contain the identification of the page of the original (pagination or foliation) and references to image data with photographs of the page of the original on selected quality levels. An example of the content of the file of the PAGE category is 21 Supplement D - Example of the file of the PAGE. 
 

The concrete application  2.1 simplifies and speeds up the automated processing of data describing the digital copy. This new construction of the structure enables an easy integration of digital copies into higher units and the migration of the copy onto various media. A more careful classification into units corresponding to HTML files improves access to digital data in a WWW browser as well as by the special software. The programs for generation of this structure have been prepared. They enable to produce correct digital copies even without detailed knowledge of these structuring rules. This enables the departments specialized especially in the work with manuscripts to concentrate mainly on the contents of the description without the necessity to solve problems of informatics.