search

mmcclenahan84 / ww1-project

Final project from our Digital Humanities class analyzing letters from WWI soldiers to see how the war impacted their mental and behavioral health
http://letters.obdurodon.org/index.xhtml
3 stars 0 forks source link

Reference list #2

Open djbpitt opened 1 year ago

djbpitt commented 1 year ago

The schema and initial reference list look great! Just a few comments:

  1. You've used camel case in a few places (e.g., metaData) and snake case in others (e.g., country_stationed). You don’t get a say in terms that are predefined in Relax NG, like minInclusive (good use of that, by the way!), but for everything else it’s best to standardize on one strategy for all of your own elements and attributes—pick whatever you prefer.
  2. Your comment says that you allowed multiple countries of origin, but I think you meant multiple countries where the person was stationed (which is what your schema models and what we find in the XML—at least so far). If there are also some instances of multiple countries of origin, you can adjust the schema accordingly.
  3. Perhaps once you've encountered a few countries you might want to use an or-group of strings instead of text as the content model. That would protect you from accidentally writing "UK" for one and "United Kingdom" for another and "England" for another. If you want to use the same list of countries for both origin and stationed, you could, using the named pattern we mentioned earlier in class (feel free to peek at the bottom of http://dh.obdurodon.org/relaxng_notes.xhtml for a reminder), but we can think of reasons that you might want those lists to be different. You can defer that decision until you have more records if you think that best.
  4. You may want to make month an or-group of strings, consisting of the twelve months, so that you won’t accidentally write "January" in one place and "Jan." in another. On the other hand, if you have month references in other letters like "May-June" or something similar, modeling the value as text is probably best.
  5. You allow "both" as a value for both positive and negative sentiment, but would you prefer "mixed"? They mean the same thing, but "mixed" might feel more idiomatic.

At the moment you have one set of metadata for all letters, which is sensible because they're all from the same source, and you make a note of this in your comments in the XML and the Relax NG. If the letters have individual URLs you might want to include those in the per-letter records, though, which will make it easier to look them up later should you want to. You might also, in your eventual project site, want to link to them, and having the URLs in the context of the individual letters will make that easier.

amcheckeye commented 1 year ago

Thanks for all of the useful insights. I decided to stick with snake case, as that is my natural inclination and what I usually choose for file names anyways. I updated my comment to say "countries_stationed" because that is indeed what there are multiples instances of. Item #3 is good advice, we'll have to discuss that in the future. The months and "mixed" sentiments have also been changed because I agree with this reasoning.