In ParlaMint there was only one element that was supposed to contain text that was linguistically analysed, namely <seg>, and I changed that to <p> in PressMint and documented it.
However, PressMint allows other elements that can directly contain text and which can be sensibly analysed (as we saw in the NL sample). In particular, children of body and div that contain text are: head label note p byline docDate.
Currently pressmint-tei2text.xsl outputs only <p> while pressmint2xmlvert.xsl only <p>, <head> and <note>.
The question is how to extend these two scripts:
- extend to all elements listed above; however, if some partners decide the won't annotate some of them, then we have a problem
- dynamically output only those that satisfy
./s[w | pc]; however, this means that we can produce the .txt files only for .ana
@matyaskopp, do you have any comment on the above?
Note also (this should probably be another issue) that .txt files now can be produced either from plain text TEI or from .ana, which might not be a good idea, as in ParlaMint we had problems because some partners had differences between the two, either .ana contained less text, or, in one case, more text, which caused problems for downstream processing. Maybe we should make two versions, X.txt and X.ana.txt? And then possibly compare them in validation and complain if they are different.
In ParlaMint there was only one element that was supposed to contain text that was linguistically analysed, namely
<seg>, and I changed that to<p>in PressMint and documented it.However, PressMint allows other elements that can directly contain text and which can be sensibly analysed (as we saw in the NL sample). In particular, children of body and div that contain text are: head label note p byline docDate.
Currently pressmint-tei2text.xsl outputs only
<p>while pressmint2xmlvert.xsl only<p>,<head>and<note>.The question is how to extend these two scripts:
./s[w | pc]; however, this means that we can produce the .txt files only for .ana@matyaskopp, do you have any comment on the above?
Note also (this should probably be another issue) that .txt files now can be produced either from plain text TEI or from .ana, which might not be a good idea, as in ParlaMint we had problems because some partners had differences between the two, either .ana contained less text, or, in one case, more text, which caused problems for downstream processing. Maybe we should make two versions, X.txt and X.ana.txt? And then possibly compare them in validation and complain if they are different.