I am beginning to work on a project digitizing a series of notebooks, and each page in the notebook will contain annotations created by the project archivist. I am going to be using Annotorious (http://annotorious.github.io/) for the annotation engine, following the model of using OpenLayers to annotate zoomable images.
Much to my surprise, TEI can handle the storage of annotations natively within the schema with facsimile and surface elements. Some use cases are well-documented at http://tei.oucs.ox.ac.uk/Talks/2009-07-dublin/talk7_facsimile.xml. However, the attributes designated for storing x and y coordinates of the annotations only accept literal pixel coordinates, i. e., they must be positive integers. For annotations made in OpenLayers, for example, the x and y coordinates are represented relatively within the bounds of the image, which are dynamically calculated based on the height and width of the source image file. Therefore, the surface element may appear as follows:
1 | <surface ulx="-0.86269991934014" uly="0.34903242955874003" lrx="-0.053827751" lry="-0.216358041"/> |
The values of these attributes are invalid according to the schema. Ideally, I would like to use an official schema. Are there other semantically acceptable ways to encode these coordinates? Or should I edit the schema and validate on a local, modified copy? Furthermore, should I send an email out to a TEI list with this use case? It seems plausible that other projects may want to annotate images in this way, and perhaps the attributes should be relaxed to accept decimal numbers rather than integers.
