tagging_workshop_2006_sep_kees

<< back

"Tags are a short-hand for documentation. People really want documentation."


Issues Discussed

  • language for the tags and search capacity on those tags
    • dealing with slang, alternate meanings
    • standardization of language
  • enticing/rewarding people to tag ("Bootstrapping"?)
    • Solution? Seed tags with automated material, make it easy to see
      • change logs
      • document contexts
    • Solution? Wizard as a tree-selection ("is this a horse?" "how is it NOT a horse?")
    • Solution? Wizard asking direct, thought-provoking questions
      • what's cool?
      • why do people want to use it?
    • Solution? reward by searched tag success (best searched tags are worth more to the author of that tag)
  • tagging old source
    • Write sentences and extract terms from those instead tags directly.
    • attach tags tightly to some specific source (source, blog, doc)
  • automation of tag extraction
    • automatable
    • scalable
    • many languages
    • fingerprint used APIs
  • "gaming" the tagging system, evil taggers, etc
    • Solutions? "good" tag ratings vs "bad" tag ratings (reputation?)
  • DOAP adoption is slow (Debian rule? like manpages?)
  • Is there a license for APIs? (Use, but not take advantage of.)
  • Gather comments about functional relationships between two areas of code.
  • Getting to block and line addressability for commenting.
  • What makes sense for filtering or guiding relevance?
    • Solution? recency? activity?
  • Copyright on the tags themselves, or the data exchange?
  • Possible new Aspect tags
    • environment
    • standards
  • Automatic tag similarity discovery, tag context clustering
  • Best-Practice for manual tagging
    • What is it?
  • Wizard for just describing a project's capabilities.
  • Find short piece of code, see what could say about it
    • Problem: usually comments describe "doing", not the "idea" of the code
  • Documenting software patterns/ideas
    • Ward's code from his presentation
  • Time-stamping is a well-solved process. Could use more date-associations being made. (ha ha: http://nethash.org/)


Areas to Examine

<< back

Groups: