Tuesday, May 12, 2009

Organization tools (a lazyweb request)

this may read as incomplete / incoherent / stream of consciousness...  it makes sense to me at the moment, I'll explain it later (or sooner, if you write, I'll clarify..).

a description of (certain aspects of) a tool that I think would be useful.

A tool to organize information.
Actually, different organizations:  temporal, alphabetic, linked, conceptual, mindmap, spatial, geographical, geospatial.

but, also allow these organizations to be changed.
incremental refinement:  I maybe tag a bunch of stuff as t01; later, i decide these can be either t02a or t02b.  but, maybe i don't want to tag it all at once, i'd like to save my workflow (after being able to define it) to refine it...  and visualize it, in various ways, various metrics.
info i create, info automatically crawled (searches, feeds, etc)
occasionally cull ("forgetting") this.

other large tasks in organization require some kind of workflow.
example:  organize my digital pictures.  
this includes:
 - make a complete list (also: note any large gaps in numbering which may indicate an incompleteness; search dvd/cd backups for these
 - i want to tag all my pictures, just for completeness.  things like people, things, bugs, flowers, plants, fungi, etc.   eventually i expect the computer to be able to do this to a large extent, similar to the face recognition that goole does wiht picasa(web)
 - i want to geotag all my pictures.  oh, assemble my gps .gpx files and sync those up (gpicsynch is good for this)


I should do this incrementally; eg, as I import each set of pictures, do tagging then...  some of this would be easier if so.  However, I also see how large-batch processing can be more consistent because you have in mind what the rules are (the tag dog v canine, the 'which one did i use before rule').  this implies some sort of meta-tools too that can organize your tags, give some historical perspective ("the tag 'canine' was used 42 times from mar-jun .."), renaming rules (everything tagged t03, ask whether it should be t03a or t03b).

you might tag your tags, create your own ontology or folksonomy for a task.
where are standard lists of tags?  this is where taxonomical rules would be ideal.

data extraction, template filling for generation of facts.
so you're not just doing a google search and looking at results and clicking on documents,
but creating a knowledge base that you can come back to later.
instead of closing the tab or 'going back', rate in some way what interest it was
 - worthless / unrelated
 - good stuff
 - not related, but interesting (sever / de-weigh from current interest, add to something else)
 - worth further looking
 - watch for changes


I think there is a set of tasks that can be done as part of the incremental workflow, and another set that may be better suited to large batch style works.  This may vary by person, or by mood, and may change over time.  You process a large batch and decidee "it would have been easier if I added that info at the time I created it" and at other times you may feel "I'm not going to do this now; I'll wait until I get a big batch and do everything later."

ways to select items of interest--search via any of the above organizations; 
selected items are now 'in the workspace' associated with my particular 'task focus'.
tomorrow i may reuse this; a few weeks now i wouldn't mind being able to come back to this particular view.  items in the particular view can be operated on, made more important, weighted, throw them together in a page with other attributes and put it out somewhere so google can find it...


links as first order types?
also while working, weighting/strength of the link.  
spreading activation, associative memory--what is the query language?
these weights have a temporality, if i renew an interest from the past, then probably related items should also have a fairly high ranking.
what kind of learning can take place?
while i'm learning the computer can note and learn something about what i'm doing,
ask me some questions ("is this link similar?  which is more X, A or B?").
maybe learn some rules about how to help me (the tripod links are pretty dumb, don't show me parked domain link farm pages even though it's in the top results, etc).

now, 
some of these tools exist in some rudimentary fashion.
there are simple mindmaps.
there are the more advanced "semantic desktop" tools, for either viewing or authoring semantic web 'pages'.
i am interested in tool-building, but not necessarily tool-building for it's own use.
again, this needs to be incrementally easy so i can build a piece at a time,
or somewhat self-learning / taught / trained.


i am tired of being frustrated by machines that slow me down.
whatever happened to the mathematica-style notebooks that have dynamic info,
where you can have active formulae in them?
what happened to the memex?

I guess I will go off and look at what tools exist:  
 - JENA
 - mozilla ubiquity
 - piggybank and other firefox plug-ins for the semantic web
 - stuff at w3c
 -  OWL and protege-related tools; kqml ?
 -  dublin core and related tools

I will try to post some links to these, if I find good summary pages, 
or a new post.  If you have something, please link.

I am posting this to the lazyweb.
lazyweb is the concept that if you wait long enough (or do enough searching)
someone will have built the thing that you need.
maybe someone will read this post and be inspired.