Lessons learned from using SGML in the Text Encoding Initiative

In April of 1994 the ACH-ALLC-ACL Text Encoding Initiative published Guidelines for Electronic Text Encoding and Interchange (Document TEI P3). SGML was used as the basis for the encoding scheme that was developed. Several innovative approaches to the use of SGML were devised during the course of the project. Three aspects of this innovation are documented in the paper. First, all of the tags are organized into sets that can be included easily into the project DTD, which allows the corresponding features to be used in documents only when required. Second, mechanisms were developed to relate parts of documents in non-hierarchic ways. Third, a mechanism was developed to allow extension of the DTD in a disciplined manner. We comment on the effectiveness with which SGML could be used in these ways and the shortcomings we perceive.