[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: The mrk element
Hi everyone, Two questions on the topic of the mrk element: === 1) type='generic': As I was trying to get something stable for <mrk> I realized that we left the question about <mrk translate='yes'> half unanswered (we got side tracked in talking about schema flexibility). It seems that if we use type='generic' we should make translate mandatory in this case, and forbid the other attributes. That would eliminate un-desirable cases. So to summarize, is the following looking fine? A) If type='generic' (default value for type) - translate is required - value is forbidden (because unused) - ref is forbidden (because unused) - id is optional B) If type='term' - translate is allowed (if not present its value is the same as the parent) - value is forbidden (because unused) - ref is an optional pointer to additional information (what format this additional information is in?) - id is optional Example: <mrk translate='no'>text with <mrk type='term'>TERM</mrk></mrk> C) If type='comment' - translate is allowed (if not present its value is the same as the parent) - value is required if ref is not set and contains the text of the comment - ref is required if value is not set and points to an external comment (what format this comment structure has?) - id is optional D) If type='<something-else>' - translate is allowed (if not present its value is the same as the parent) - value and ref and id have a behavior defined by the user-defined annotation === 2) multi-spans annotations: We touched on the cases where we would have a single <mrk> split into two by a segment break for example. We said we would keep them as two separate mrk (vs introducing some <sa>/<ea> element) and allow for joining them when the split is removed. This led to the question of an extra attribute to indicate that two adjacent <mrk> should not be re-joined. I think it might be easier to use the id attribute: same id means the mrk elements are re-joinable: Before: <segment><source>Some <mrk translate='no'>text: stuff</mrk></source></segment> After the split: <segment><source>Some <mrk id='1' translate='no'>text: </mrk></source></segment> <segment><source><mrk id='1' translate='no'>stuff</mrk></source></segment> Note that if the split include ignorable we are for some funky markup: <segment><source>Some <mrk id='1' translate='no'>text:</mrk></source></segment> <ignorable><source><mrk id='1' translate='no'> </mrk></source></ignorable> <segment><source><mrk id='1' translate='no'>stuff</mrk></source></segment> Any thought? -yves
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]