Twenty-first century technologies and techniquesdemand expanded and extended documentation efforts tocontinue human progress. Documentation challengesinclude managing records of manufacturing and supplychain events in complex, global production (e.g., Apple'smicro etching of quick codes on iPhone screens); manag-ing records of 3D printed objects and organisms(e.g., StanfordBASE Lab's additive printing of layers ofcells to generate a human heart); and documentation sys-tems for the inevitable development of bit-level water-marks.

Just as importantly, we must work todevelop understanding of the nuances of different algo-rithms so that training sets may be well-suited to theproblem domains to which the generative tools areapplied

To do so, we must build upontraditional collection development strategies to make thestructure and provenance of data sets transparent or atleast documentable

The most used basis for organizing in universal systemshas been division (or collocation) by scholarly disciplines.

ICONCLASS

Provenance is a fundamental principle of archivalscience, referring to the individuals, groups, or or-ganizations that originally created or received theitems in a collection, and to the items’ subsequentchain of custody. According to archival theory andthe principle of provenance, records which originatefrom a common source (or fonds) should be kepttogether—either physically, or, where that is notpracticable, intellectually in the way in which they arecatalogued and arranged in finding aids

LCC is a model of an enumerative sys-tem, in which all the classes are listed (and the systemstherefore is comprehensive, LCC fills up about 41 vol-umes). Faceted systems, on the other hand, do not list alltheir classes, but provide building blocks from which spe-cific classes for each document may be formed (Rangana-than was inspired by the Meccano toy).

Universal systems are less important for spe-cial libraries and for scholarly subject retrieval comparedwith special systems designed for subject bibliographiessuch as MEDLINE or PsycINFO

Knowledge organization processes (KOP) are, for exam-ple, the processes of cataloging, subject analysis, indexingand classification by humans or computers

Knowledge or-ganization systems (KOS) are selection of concepts withindication of selected semantic relations

As Bazerman (2012) reminds us, while recognizingthe social importance of effective search engines andother systems of structuring knowledge and inscrib-ing writing, we still need to understand the activitycontexts of those producing and using knowledgeand information because no matter how fragmen-tary, how automatic, and how fast informationcomes to a user, the very user (herself/himselfplaced in an activity contexts [sic]) must ultimatelymake sense of the information found and that sensecannot be made without understanding the variousof activity (and the practices ) producing that infor

core principle of the domain-analytic approach is:“The starting point for understanding classification is onethat any object, any document and any domain could beclassified from multiple equal correct perspectives”

KO is first and foremost institutionalized in professorshipsat universities around the world, in teaching and researchprograms at research institutions and schools of highereducation, in scholarly journals