Reference and annotation in digital texts. From citation to “Watson”

The lack of segmentation by pages and lines creates severe problems with scholarly citation of digital documents - in spite of all advantages of digital texts Not only citation, but the whole editorial work of identifying, comparing, manipulating and annotating text make use of reference to distinctive parts of text - and also the programs that support such tasks. The paper propagates to base reference in digital texts entirely on a structure of meaningful entities like sentences or words. Based on such a structure most annotational markup can be handled “stand-off”. I refer to an implementation of my own and to “Watson”, a very ambitious project of annotation. Using it a computer won in Jeopardy, a prominent US Quiz-Show.