A Crowd Sourcing / Search v.s Browse groups
B Localization of IS Standards to the humanistic discipline—networked ontologies
C Groups—relationships/ biographical data
D Crosswalks across time/ culture
E Scholarly and reusable resources
Crosswalk Breakout Group
Crosswalk or merge? Merging requires deduplication.
Factors that determine type and possibility of crosswalk: granularity, hierarchy, cultural- or dataset-specific features; changing meaning over time. Inevitably “dumb down” from more to less granular. Definition: overarching structure layered on top of project, does not define scope of project.
Types (do these in order):
Translation/transcoding: concept goes by different names, e.g., “I”, “uno”, “1”. “aristocracy” ?= “nobility”; “elite” = “gentry”. (Granularity, e.g., gardener, botanist, farmer, agriculturalist, etc., isn’t a cross-walk issue; it’s broader ~ narrower term.) Same term may have different meanings in different times, e.g., printer in Samuel Johnson’s time is a publisher and bookseller. This is about equality: find as many correspondences as possible.
Merging: Or perhaps granularity is a type of merging, e.g., merging granular terms for agricultural workers into a single farmer/gardener umbrella topic. Add together the stuff you couldn’t merge. E.g., “there’s no English counterpart to Huguenot, so add the category or merge into ‘religious minority’.” New umbrella category.
Union vs intersection: (A, B, C) + (B, C, D) might yield (A, B, C, D) or (B, C). Coexisting systems: date systems?
Crosswalk: relationships or implementation? How about: best practices for creating reusable ontology?
No Ur-ontology, only the ability to translate from one to another. OR (rephrased): Is there no ur-onotology (through plurality/collection of specific project ontologies)?
Elisa’s incomplete notes on group collective discussions:
Objects: How can we interlace layers of descriptive structures for objects?
If you make a category infinitely reusable, it’s not practically helpful for specific things. Example of a book in thing theory (many distinct manifestations)
Chris Warren: Guiding principles: LAYERING and EXTENSIBILITY
Aaron adds; MODULARITY — suggests that FOAF is used in a modular way: use the structure of it…
RDF: subject-predicate-object structure
VIAF ids can be scraped from their API (Is it this? http://api-portal.anypoint.mulesoft.com/oclc/api/oclc-viaf-api)
How to fine-tune/check group definitions:
D-Dupe: to help identify duplicate (or potentially duplicate) nodes http://linqs.cs.umd.edu/projects/ddupe/
Poisson Graphical lasso: network stats to infer relationships within a group—This has “tweakable coefficients”