Frank Mitchell

Human language interface

Tags as concepts, ideas, spaces

Merging tags

  1. Drop the plural morphemes: cats == cat
  2. Acronym expansion: ny == “new york”

Both can be combined in the same storage space; both end up doing the same thing. Morphemes map to each other. Bidirectional mapping. Unidirection: needed, wanted? Probably not.

Self-organizing tag spaces

Don’t overwhelm the user. Cover 90% of the information space. That’s good enough. Maybe a fuzzy number filter. Evaluate how often a tag is used. Is this often enough? Language as a basis for this selection.

“While I was learning Russian, I put check marks in the dictionary. If I had four checks next to a word I knew it was one I should probably memorize.”


Rare tags don’t need to be shown. Overlapping of tags will ensure those nodes with rare tags are eventually shown.

Fuzzy number filter can be controlled via contributing percentage of a given tag. Tag must be associated with at least n% of total nodes.

Browsing tag spaces

Let users define intersections. Pick from a list. No need for URLs (command line) interface. Select / deselect -> same operation, click the tag.

Do we need to keep a separate viewable list of what’s been selected, or can we simply change colors in our list of all tags?

Combining tag spaces

Can we merge two people’s tag spaces? What tags are common to both of them?