Callimachus—avoiding The Pitfalls Of Xml For Collaborative Text Analysis
Annotation structures vs exploratory structures
Putting it another way, annotative projects are those in which the text is linked to external information, be it commentary, historical context, biographical details or any other class of supportive information. The primary research of such projects tends to be aimed at uncovering such additional annotative materials and incorporating them into the growing network of information attached to the core text(s). Users of these editions can then explore the hyperlinked network of information to gain a deeper understanding of the primary work. These projects are externally focused—looking to the world outside the text.
In contrast, our hypertext edition has evolved to become exploratory and inwardly focused. Rather than making connections to external data, such projects are about discovering relationships and connections that exist within the text itself. Unfortunately, as our hopes for the evolving hypertext edition grew, we found that we were dealing with XML-editing software that was rather technical and tended to exaggerate the coded view of the text. In other words, our inward focus was becoming so encumbered by the technical requirements of code that we were losing the plain textual intimacy with which humanists are traditionally more comfortable. Looking through a text to its metadata (or its coded tags, in this case) is revealing, but eventually the complexity of the tags obscures the text itself and, as scholars, we find ourselves at a greater distance, one which hampered our exploratory work. The technology that was beginning to cause problems was the same that was so helpful when our project was less exploratory and more explicatory or annotative.
[...]
Exploratory projects, apparently, are a different class of problem altogether. In these projects, the nature of the relationships that will be discovered is simply not known at the outset. The relationships may be semantic, linguistic, structural or something entirely different. They may link two concepts together or they might involve thirty concepts. They may be entirely internal relationships, or they might relate to external information. The relationships may be objectively verifiable or subjective interpretations.
Furthermore, due to the often incremental nature of the exploration, a relationship discovered today may be a simplification of a greater truth uncovered tomorrow or it might be a red herring, only awaiting tomorrow’s discovery to disprove it. Consequently, even if one could presuppose a structure for the information to be encoded, that structure is regularly subject to revision and change, often drastic change. Using an annotative model to construct an exploratory project is a mistake that we have come to call the fallacy of prescience.