This paper discusses the development of Comic Book Markup Language, a XML vocabulary, with both DTD and schema (W3C and RELAX NG) representations, for encoding comics, comic books, and graphic novels. For further introductory and background information, please see the abstract above. In the sections that follow, I do not give a detailed explanation of the DTD and all its elements and attributes. Instead, I provide an overview of the data I am seeking to encode, the motivations behind the development of CBML, the importance of metadata to the project, and the basic structure of comics. I also provide a number of illustrations and examples. The DTD and schema and other detailed documentation about the project will be available by the time of the conference at http://www.cbml.org/.

The motivations behind the creation of CBML are the same as those shared by many other scholarly digitization projects. One motive is preservation. Comic books, especially older ones, are typically printed on cheap, high-acid newsprint. They are in constant danger of decay and irreparable loss. By digitizing these important artworks we preserve them for our own and future generations. A second motivation is access. Many comics are rare, expensive, highly-prized collector's items. While many rare and other comic books are often available in affordable reprints, the reprints, more often than not, fail to reproduce the entire comic book with the all-important context of the original advertisements, indicia, and so on. By digitizing these comics they are made accessible to a far greater number of scholars and students. In many cases, copyright restrictions prohibit one from making these digitized artifacts available for free over the Internet. But even when access is controlled in accordance with applicable laws and agreements, the digitized copy provides for greater access than is provided by a single (or a few) print copies of the original. Multiple students, for instance, may access the same digitized work simultaneously. And scholars and students and other researchers may, with the proper authentication, access a work from their office, dorm room, or home. A third motivation is increased functionality. The digitized version offers functionality not available in the original print version of the comic book. This added functionality includes the ability to perform on the data full-text searches and complex queries facilitated by added and regularized metadata and the rich and detailed encoding added to the original document. If, for instance, one wants to find every appearance of the character Galactus in the first one hundred issues of the [ Fantastic Four ] comic book, the only way one can do this with the print comic books is first to acquire access to the first one hundred issues of the [ Fantastic Four ] and then to read the text and study the artwork on each page of each issue. On the other hand, with the digitized collection containing metadata elements such as thorough character lists, one could quickly find this information using a simple query.

As mentioned in the previous section, one motivation behind the creation of CBML is to provide added value and functionality to the print originals. Much of this added value and functionality is dependant upon the metadata elements and attributes that are included in CBML. Metadata are, of course, data about the data--data that describe the data. In CBML, the data are the original comics, or comic books, or graphic novels. The metadata is, most obviously, the markup that is added to the original comic. The metadata also include added data elements and attributes that are not explicitly available in the original comic book. For example, the splash page of a comic may list Adam Austin as the artist. Nowhere in the comic book does the name Gene Colan appear. But the trained encoder knows that Adam Austin is a pseudonym for the famous comic artist Gene Colan. So the encoder may encode the name as follows:

            <name type="person" regularized="Colan, Gene">Adam Austin</name> 
        

Here, the added metadata, which are not explicit in the original, would allow a scholar researching the work of Gene Colan to find comics drawn not just by Gene Colan but also by his pseudonymous identity Adam Austin. This resolution of "true" or legal identities with pseudonymous identities is a function of metadata.

Character lists are another case in which added metadata may be extremely useful. I previously mentioned the case of a researcher searching for all appearances of the character Galactus in selected issues of the [ Fantastic Four ] . But the character may appear in the comic, even though his name never appears in the text. He may be referred to by a nickname; or his image may appear, without any textual reference, as a shadow in the final panel of the comic book, foreshadowing his appearance in the next issue. These appearances would not be uncovered by a full-text search of the original textual data of the comic. However, if such an appearance is noted in a metadata element such as a characterList or in the characters attribute of a panelDescription element, this metadata may be searched to retrieve more accurate information about a character's appearances. And of course super heroes are a common subject in comics, but super heroes go by many names. Spider-Man, for instance, is also know as Spidey, Web-Head, Wallcrawler, and other nicknames. And Spider-Man also has a secret identity and may be referred to as Peter Parker, or his Aunt May may call him Petey. If one is searching for all appearances of or references to Spider-Man or Peter Parker, one may have to search for dozens of different names and epithets. However, if all references to the character Spider-Man, regardless of the variation on his name or identity, are encoded with an IDREF attribute that points to a single authoritative and regularized name with a description of the character and a list of the multiple name and identity variations, one can easily find all the appearances of and references to that character without having to conduct multiple or complex searches--one simply searches for elements containing an IDREF matching the character in question.

The development of CBML began as an effort to encode a single representative instance of a comic book. I chose a fairly standard or traditional comic book from the 1970s by Jack Kirby, one of the towering giants of twentieth-century comics. By encoding this single representative instance of a comic, I have encountered many of the metadata, structural, and vocabulary issues one is likely to encounter in a majority of comics. Then moving past this single instance, I have looked at some more experimental examples in which the structure and layout of panels and text differs somehow from standard practice. These exceptions to the norm have forced me to consider many other possibilities and opened up the DTD/schema to deal with a wider variety of approaches to comics and comics art. As the development of CBML continues, I and others who join the development process will continue to look for challenging examples and modify CBML accordingly so that it will provide an appropriate vocabulary for encoding most, if not all, comics.

In developing CBML I have relied on the many excellent examples provided by other XML vocabularies, particularly the Text Encoding Initiative ( Text Encoding Initiative (TEI) ) Standard Generalized Markup Language (SGML) and XML markup language. The TEI is designed to encode existing texts, include novels and other prose, poems, dramas, and dictionaries. The TEI provides an extensive vocabulary and a documented and standard method for extending and modifying the stock vocabulary. It is widely used in the academic and digital library communities to encode existing scholarly and other texts. The text of a comic book, captions and speeches or thoughts by characters, shares much in common with other literary forms, such as the novel and the drama. For encoding these features of the text, the TEI provides a useful model. Other features of comics, such as the often complex arrangement and sequence of panels, sound effects (e.g. POW!, BAM!, KRASH!), and advertisements are less readily described by the stock TEI vocabulary. Rather than extend the TEI by adding and renaming elements, redefining content models, and so on, which would result in a document type difficult to recognize as having originated from the stock TEI, I have found it preferable to build a new XML vocabulary specific to comics, comic books, and graphic novels, borrowing where appropriate from the TEI and other vocabularies. XHTML forms, for instance, provide a useful model for encoding the many cut-out mail order forms that are found in the advertisements in comic books.

In this paper, I have just touched, in an introductory way, on the initial stages of development of CBML. I hope to have shown that CBML is an ambitious, interesting, and valuable project. The issues encountered and lessons learned in the development of CBML can be of great use to others developing XML vocabularies for describing complex documents, especially documents that describe sequential artwork and combinations of visual art and text. CBML remains in the early stages of development, and the hope is that after XML 2002, I will interest others in this project and develop a community of interested parties who will participate in the further development of CBML. The groundwork I have laid should serve as a useful starting point for development by a larger community of interested parties. I will also pursue placing CBML with an appropriate home in the form of a standards body such as Organization for the Advancement of Structured Information Standards (OASIS)http://www.oasis-open.org/.

Along with the development of the XML vocabulary for CBML, I am simultaneously working on a web-based prototype for searching the resulting CBML-encoded comic books and displaying the textual and image data and metadata. My plan is to have this prototype far enough along so I may demo it at XML 2002.

Given the early stages of development, I hope interested parties will regularly visit http://www.cbml.org/ and the conference web site http://www.xmlconference.org/ for updated versions of this paper, the CBML DTD and schema, the search and display prototype, and other news regarding the ongoing development of CBML and related projects.