In a Karlsruhe, DE-based group, young researchers are asking an interesting question:
“What data structures do humans think in? What data types should be supportet by a cognitively adequate PKM tool?”
If you think of what relations can occur in sentences / text snippets that we write down in paper notes, the following (lower level) relations need to be accommodated:
- Partonymy (Part-Whole-Relation);
- Hyp[er]onymy (Sub/superordination), including (informal) Tags, (formal) taxonomy, and facet classification / feature semantics;
- Order, including temporal and causal sequence
- Synonymy — where I don’t know how to translate our German words of the two different types, sinnverwandt and sachverwandt: ~ related in sense vs. related matters;
- General kinship/relatedness, in a broader sense, such as camps, coalitions, scholarly schools of thought (especially in references: co-authorship, conference proceedings, edited volumes);
- Antonymy, e. g. enmity;
- Markedness, e. g. Central/Prototypical, or (more processual than structural): Todo/Done, Question/Answered, Remarkable, Criticism-provoking;
- Interrelation that cannot yet be specified more precisely;
- Interrelation that is not yet wrote down more precisely;
- Association that remains vague;
- Genuine cross-reference (“see also”).
These basic relations could perhaps be aggregated to a higher level, on the basis of core processes and core structures. Core processes of thinking include the following:
- Join (create relations of type 1-5 and 8-11);
- Distinguish/differentiate (relations of type 6-7);
- Zoom, in a broader sense, including
- Expand <-> Collapse,
- Title/Abbreviation <-> Quicktip
- Link to detail <-> Overview page
This list can be further reduced to core structures:
- hierarchical relations (1-7 and zoom),
- non-hierarchical relations and cross-references (8-11)
if we create a tag (type 2) for all the marked items (7), and if we put two antonyms (6) under their common hyperonym (2).
I think the above two elementary relation types, hierarchical and non-hierarchical, can be intuitively assigned and noted down without distraction from the task at hand. Later on, they could be refined to more sophisticated subtypes, if the notes are so valuable that this seems desirable.
If you have ideas about the conceptual data structures, please tell me, or write directly to the group’s wiki.