Annotation as a New Paradigm in Research Archiving

What it is about

The paper argues that the results of digital scholarship (query results, feature values, topic assignments) should be kept as annotations, and that these annotations should be portable, so that they still make sense after the software that made them has changed. The two test cases are the CKCC project (about 20,000 seventeenth-century Dutch letters, with topic models such as LDA, LSA and Random Indexing) and the DTHB project (a linguistically tagged Hebrew Bible in the Emdros text database, queried with MQL). The authors built two applications: QFA, which shows queries and word-level features (70 key=value pairs) as annotations on the text, and TKA, which shows topic and keyword assignments. Annotations are modelled as a body (a query, a feature, a keyword) that targets a set of passages. The examples show why this matters: the feature gender=masculine alone has 101,335 targets, and an old query can stop running when the database system changes. The authors conclude that annotations can carry queries and features, but that the open questions are how to address fragments of resources and how to attach metadata.

Ideas

• I1. Record every published or repeated query as an annotation whose body is the query text, the engine version and the run date, and whose targets are the matched node or token ids. A later run of the same query can then be compared with the stored result, and a changed tree store does not silently change the old answer.
• I2. Turn each UD feature value of the trees (for example Tense=Imp on verbs, Gender on nouns) into an annotation with the token ids as targets, kept outside the tree files and linked only by symbolic ids, with no foreign-key dependency. A debug view can then list every imperfect verb in a passage, and replacing the trees does not break the index.
• I3. Use the reverse direction the paper calls serendipity: from a result back to the passages that produced it. For a grammar rule in hbo, the rule firing on a verse is an annotation (body = rule id, targets = the tokens), so a reader can see where a rule applies in the text, not only what it says.

For Mova

I1 belongs in world and in the tree store (the record of a world run and its version). I2 is a debugging index over treebanks/hbo/ptnk for The verb. I3 is a rule-explanation view for 13. Biblical Hebrew in Mova: the tools. No ids are taken from the paper; only its model (body plus symbolic targets) is used.

Cite

The paper itself, not this card. BibTeX from the library's record of the work. DOI 10.1002/meet.14504901084 · arXiv 1412.6069

@misc{roorda2014annotation,
  title = {{Annotation as a New Paradigm in Research Archiving}},
  author = {Roorda, Dirk and Heuvel, Charles van den},
  year = {2014},
  doi = {10.1002/meet.14504901084},
  eprint = {1412.6069},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.1002/meet.14504901084}
}

References

5 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. ACLS (2006). Our Cultural Commonwealth: The Report of the American Council of Learned Societies’ Commission tinyurl.com/intered-lvn. find
2. Sanderson, R., van de Sompel, H. (Eds.). (2011). Open Annotation: Beta Data Model Guide. Web document. Retrieved 2012-04-28 from openannotation.org. find
3. Wittek, P., Ravenek, W. (2011). Supporting the exploration of a corpus of 17th century scholarly correspondences by topic modeling. In B. Maegaard (Ed.), Proceedings of Supporting Digital Humanities 2011: Answering the unaskable. Copenhagen. find
4. Talstra, E., Sikkel, C., Glanz, O., Oosting, R., Dyk, J.W. (2012). Text Database of the Hebrew Bible. Dataset available from Data Archiving and Networked services after permission of the depositor through Retrieved 2012-04-28 from tinyurl.com/dans-wivu. find
5. Zerby, C. (2003). The Devil's Details: A History of Footnotes. New York: Touchstone. find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/langstream-2014-roorda-annotation-new-paradigm-research-archiving/