Dripmail

"It's sort of like being mentioned on Days of Our Lives"

-- Gregory Mathes, on what being Slashdotted means, in terms my mother
might be able to understand.

... I think the next "goal" is to devise a system that is similar to
Josh's Tangent [ http://tangent.cx ] but that works based on tags across
systems, and has simple social networking / friends stuff.

1. Stop using Furl and start using Delicious (or explore Furl's keywords
capabilities)

2. Get in the habit of adding tags to Flickr photos / roll it into my
autogal program.

3. Get off lazy ass and finish trenchant daily tag stuff (really, it's
almost ready, I think)

4. Integrate tagged material from other feeds (delicious, flickr) into
trenchant daily tag archive.

5. Figure out how to do "and see material tagged similarly from my
friends" on the tag archives pages.

So, really, it'll be easy!

XFML has done some work in this area and is designed to allow people to
publish a taxonomy, and what pages on their site are occurrences of topics
in their taxonomy (and it's faceted!) It's cool for information
architecture nerds, but I drove myself crazy fucking around with that
stuff, and really, I think Delicious and Flickr and the fact that I spent
all that time researching Folksonomies show I really should just abandon
hierarchy for a flat tag namespace. (It also makes processing things
easier when you don't have arbitrarily deep levels of things -- you can
avoid thinking in XSLT.)

I think maybe the plan should be to develop a simplified version similar
to XFML, like, TSML (Tag Sharing Markup Language) maybe, and then develop
a suite of scripts that processes TSML from you and your friends and
generates archives by tags / links to your friends stuff. Of course, it
would help if I had friends. But if I make some Delicious/Flickr API calls
that translate into this vocabulary, it'll be like I had friends who do
this. (Also, it may make some sense to try and develop an RSS extension
that will have this functionality, because the kids *love* RSS.)


-adam

Hi Mike,

I'm getting a lot of interesting feedback on the web and via email about
the paper and folksonomies, but yeah, it's a bit overwhelming --
especially since I never really thought anyone would read it. But I'm
going to try to get to it all at some point!

Perhaps I'm too cynical, but reading Rosenfeld and Shirky's posts, I was
reminded of the distinction (in my mind) between people who write for love
or interest ("amateur" in the purest sense) and those that have a vested
economic interest in making loud, bold, pronouncements because they need
to be high visibility "experts" who can charge for speaking gigs, get book
deals, etc. Not that they don't have valid points, but sometimes it gets
annoying to hear Jakob Nielsen-esque overexaggerations all the time.

Rosenfeld's point that:

"Folksonomies aren't likely to organically arrive at preferred terms for
concepts, or even evolve synonymous clusters. They're highly unlikely to
develop beyond flat lists and accrue the broader and narrower term
relationships that we see in thesauri."

(His larger point about folksonomies being inherently inferior to
professional metadata seems to ignore the large body of research saying
library users ignore those ugly red LCC volumes and aren't at all happy
with categorization and classification schemes for the most part.)

I think this understandably drives Shirky crazy: one of the fundamental
strengths of a folksonomy -- or even just free tagging without the social
context -- is that rather than having explicitly defined relations between
terms (which makes applying metadata exponentially more difficult) we just
have the terms, and then tease out latent relations through other means.
They aren't supposed to "develop" complex term relationships, because
folksonomies are a reaction against that entire paradigm.

My hunch is that there's probably something in cognitive science to back
this up -- the sense that adding relations/hierarchy isn't a linear
increase in complexity, but some massive jump. One of my colleagues in the
department was a brain and cog sci major at MIT and was interested after
my short presentation on folksonomies in class; I keep meaning to ask him
about this.

You're right in one sense that:

"Folksonomies are unlikely to evolve synonyms for the simple reason that
people will usually choose just one of the many synonyms available."

This is true from a "tag" centric viewpoint but less so in a "document"
centric view. (Hopefully this will make some sense.) The assumption is
that synonyms or similar tag relations are inferred by examining a tag,
then observing what other tags tend to be applied along with that tag.

In Flickr, there is almost always a one-to-one relationship between a
photo and a user, since people are uploading and tagging their
own photos. There's no overlap in the documents (photos) -- each photo is
unique.

In the context of Delicious (and you can certainly envision other systems
where this would be true) there is a many to one relation between users
and documents -- i.e. somehow over 300 people tagged my folksonomies
paper with various terms.

If we look at it in the context of a document, all the terms various
users applied to that document are possibly "related," and the more
documents those terms share, the more likely a relation, and the more
likely a "synonym" -- although in the thesauri sense it's just as likely
a broader/narrower relation, which is again more reason to just not deal
with that complexity.

There's also the possibility (I think Furl does this) of using traditional
statistical text analysis approaches to create document similarity indexes
-- and then correlate that to tags. So if documents are "textually"
similar, there may be a possibility the tags applied to them are also
similar even in the absence of direct tag overlap. Image similarity is not
as well developed or understood, but seems at least theoretically to hold
some promise for things like Flickr. Or it could be disastrous. My only
experiment with image similarity was a small, fun project on video game
content based retrieval:

http://adammathes.com/academic/search-engines/video-game-images/ which
turned out to be a lot more effective than I thought it would be.

But yeah, it's obviously a serious challenge and issue.
Another possibility is to think about explicit user generated
meta-metadata -- letting users explicitly create relations between terms
for themselves. This was brought up by my professor after my presentation.
I'm a little squeamish about it because of my point about relations adding
a huge jump in complexity, but maybe if you limited it to a single
ambiguous type of relation "related to" it might be worth a shot.


-adam


ps - I CC'ed this to dripmail [ http://trenchant.org/dripmail], my
languishing experiment in email sharing.


On Fri, 7 Jan 2005, migurski wrote:

> You've been sent a Flickr Mail from migurski:
>
> ------------------------------------------------------------
>
> :: Folksonomies (surprise!)
>
>
> I know you're just drowning in chatter about your famous
> article, but here's some more I just banged out:
> http://mike.teczno.com/notes/folksonomy_synonyms.html
>
> Sorry for mailing you via Flickr, I happened to be logged
> in and it was easier than looking for your real address. :D
>
>
> Cheers!
>
> ------------------------------------------------------------
>
> To reply to this message, click here:
> http://www.flickr.com/messages_read.gne?id=629408
>
> To update your email settings, click here:
> http://www.flickr.com/profile_email.gne
>
> The Flickr Team
>
> www.flickr.com
>

http://trenchant.org/tags/

I haven't done a lot of 2003 and haven't even started 2004 yet, so a lot
of stuff is missing / not there yet. And I'm still working on the scripts.

And because I tagged most of it originally with the intention of doing a
complex XFML based hierarchy faceted doodad thingee, I'll probably end up
retagging some of it next semester when I'm bored and have nothing to do.

AN interesting thing about this is that it was generated
from an RSS 2.0 (with the category element being used for a tag) dump of
trenchant.org. Since all the software today pretty much can speak RSS,
this is a pretty re-usable solution (if you're interested.)

The ultimate goal is probably to switch from an XML parsing method to a
database-backed solution -- which will allow me to monitor friends RSS
feeds and then mix them into my tags pages when they overlap. I can throw
in my friends Flickr photos, Delicious links, web posts, whatever.

If you're interested in playing with this (and have tags / "spices") on your
entries, let me know. Which... you probably don't, so, yeah.


-adam

And I'll probably add the tags to the individual daily entries too (the
code is already in the templates, but I'm not going to regenerate the site
until I'm happy with /tags/)

On Spiceplay, I use the Lingua::EN::Inflect to make plural versions of
tags/spices synonymous within the system (yes, even though the site is
PHP). Could there not be a Lingua::EN::Thesaurus module to do the same
thing with more disparate tags?

I'm still waiting for the ultimate language parsing tool that will let
me link posts about "sky" to posts about "blue", but surely a SOAP hook
into a central thesaurus could solve some tagging problems.

-josh

Hmm... the Getty's Art and Architecture Thesaurus

http://www.getty.edu/research/conducting_research/vocabularies/aat/

has always been of interest to me. But as an LIS nerd, I should
probably know of something more general -- the Library of Congress
Subject Headings is the first to come to mind, I wonder if it /
anything similar has some reasonable API open somewhere or some open
electronic version I could parse through.


-adam

On Sat, 8 Jan 2005 20:13:00 -0800, Josh Santangelo <josh@[xxx] wrote:
> On Spiceplay, I use the Lingua::EN::Inflect to make plural versions of
> tags/spices synonymous within the system (yes, even though the site is
> PHP). Could there not be a Lingua::EN::Thesaurus module to do the same
> thing with more disparate tags?
>
> I'm still waiting for the ultimate language parsing tool that will let
> me link posts about "sky" to posts about "blue", but surely a SOAP hook
> into a central thesaurus could solve some tagging problems.
>
> -josh
>
>

Oh god. Please don't use the Flickr style of
the-font-size-equals-the-frequency. It's cool for them because it's
consumerish, but Trenchant needs a regular all tag vs posts sortable
table. Please.

If you're interested at all in talking to Spiceplay stuff, it would
probably not be that hard to add RSS categories to our feeds, though
I'm not familiar with that bit of the spec.

But wait, I thought you hated RSS?

-josh

My main issue with RSS was more the culture of dehumanizing
aggregators -- why read material in an environment devoid of context
where everything looks the same? -- but I've since given in and use
Bloglines. I don't actually *read* things in the aggregator and can't
stand full text feeds, but, whatever. It's actually probably better
than Spyonit was, and I loved the hell out of that.

RSS as a *syndication* technology is brilliant! I mean, you can put
headlines on other people's sites! That was sarcasm, since nobody ever
really did this -- but actually it's sort of what I'm thinking about
doing here.

RSS is really the only metadata format people on the web (with content
that is not totally boring government crap) so at some point I have to
say, well, at least it's usable -- even if the politics and crap
behind it is repulsive.

I'll put up a public RSS 2.0 feed for trenchant at some point with
examples but basically:

<item>
<link>http://example.com/whatever/</link>;
<title>whatever</title>
<category domain="http://spiceplay.com/">glasses</category>;
<category>palm pilot</category>
<category>porn</category>
</item>

Domain is optional, and in this case we probably don't care at all.


-adam


On Sat, 8 Jan 2005 20:31:30 -0800, Josh Santangelo <josh@[xxx] wrote:
> Oh god. Please don't use the Flickr style of
> the-font-size-equals-the-frequency. It's cool for them because it's
> consumerish, but Trenchant needs a regular all tag vs posts sortable
> table. Please.
>
> If you're interested at all in talking to Spiceplay stuff, it would
> probably not be that hard to add RSS categories to our feeds, though
> I'm not familiar with that bit of the spec.
>
> But wait, I thought you hated RSS?
>
> -josh
>
>

On Sat, 8 Jan 2005, Josh Santangelo wrote:

> Oh god. Please don't use the Flickr style of
> the-font-size-equals-the-frequency. It's cool for them because it's
> consumerish, but Trenchant needs a regular all tag vs posts sortable table.
> Please.

Heh, Klara said the font-size thing was the only interesting thing about
this whole thing.

I will do a boring table version too. Any suggestions on design, Mr. UI
wizard?


-adam

A girl would. Freakin' girls, man.

<table>
<tr>
<th><a href="sort tags alphabetically">tag</a></th>
<th><a href="sort tags by frequency">frequency</a></th>
<th>&nbsp;</th>
</tr>
<? foreach($tags as $tag) { ?>
<tr>
<td><a href="/tags/<?= $tag->name ?>.html"><?= $tag->name ?></a></td>
<td><?= $tag->count ?></td>
<td>
<? foreach($tags->entries as $entry) { ?>
<a href="/daily/whatever/"><?= $entry->subject ?></a><br />
<? } ?>
</td>
</tr>
<? } ?>
</table>

A first attempt, anyway.

I think the whole font-size thing works to generate traffic only to the
biggest (most popular) items, which are not necessarily the most
interesting ones.

It would work better if the size were tempered by date. If you have 100
items about x but that were posted two years ago, and 50 items about y
that just got posted, the latter should be bigger.

Visual size tempered by time. Can I patent that?

-josh

On Sat, 8 Jan 2005 21:48:52 -0800, Josh Santangelo <josh@[xxx] wrote:
> A girl would. Freakin' girls, man.

Such misogyny!


(And besides, I like the font size thing too!)

-adam

boys are gross.

also, the visual representation of knowledge is hottt, etc.

http://blog.benbrown.com

that's got a mysql backend thing with all sorts of tag fun. i'll be
building in a bunch of rss play stuff shortly. I want to play with you,
adam.

An interesting thing about Flickr that I just noticed the other day is
that you can add tags to *other* people's photos, if they list you as a
contact. That's kind of neat and would be interesting if implemented on
a web-wide level.

-josh

Yeah!

Actually, I was working on a thing like that called Taggr -- the idea
was to let anybody tag Daily entries, and then release a software
suite so anybody could add that functionality to their sites.

Then I figured this would probably just be a gateway to
spam/troll/craptastico. So it really needs to be tied into a
friends/trusted person network. Piggybacking on top of Typekey (bleh)
or Flickr API (maybe) would be a good idea, but I'm just not as into
the "let people add things to my writing." Hell, I don't even allow
comments, so I lost interest.

(Also, it feels like just sort of reinventing Delicious, but less centralized.)


-adam


On Sun, 9 Jan 2005 14:47:46 -0800, Josh Santangelo <josh@[xxx] wrote:
> An interesting thing about Flickr that I just noticed the other day is
> that you can add tags to *other* people's photos, if they list you as a
> contact. That's kind of neat and would be interesting if implemented on
> a web-wide level.
>
> -josh
>
>

a more interesting app for something called taggr would be to pull common
tags off of flickr and off of delicious, then offer alternate posting
interfaces to both of those (api, baby) that autocomplete tags, ala
google's thing.

This could be neat:

"I'm getting well underway with coding a real, grown-up version of the
framework under this site, which I must do in order for the new job.
Thus, I'll give away more code for building weird semantic webbish sites
soon, under an open-source license, if anyone wants it. For real this
time. I swear. I really want to. I won't screw it up. It will be a
combination of Apache2 mod_rewrite rules, PHP5, XSLT, MySQL, and the
Sesame RDF engine, and you'll need to understand all five technologies
to use it. It won't have a content management interface aside from weird
XML files, and it won't work at all when you start it up. You'll love that."

http://www.ftrain.com/InstallingUpdates.html

If it's the guts of http://harpers.org/, which he built, I definitely
want that shit.

-josh

Maybe someone just saved you a bunch of effort....

http://www.niallkennedy.com/blog/archives/2005/01/technorati_laun.html

-josh

I'm sort of meh on the technorati thing...

I haven't looked at the API yet, so I'm not sure what/if you can get
tag/category information out. Even if I can, I don't see any reason to
build on top of their API because Technorati sort of creeps me out -- and
I think just by monitoring RSS feeds myself, I can get the info I need,
and then I'm not dependent on some for-profit company's API that may or
may not disappear or change or charge me.

Also, again, seeing what *everyone* has tagged with a certain tag is
interesting at first, but I like the idea of being able to filter this and
see things just from friends and sources I trust (or one degree out from
that.)


-adam

trenchant.org - entwined - all tags

Adam has tagged all of his daily entries over the last few years One
day, he will grow up to be the senile old man wandering around with the a
handheld labeler, tagging everything in sight.

to adammathes folksonomies tagging socialnetworking projects trenchant
crazy by kathrynyu ... on 2005-01-17 ... copy this item

http://blog.benbrown.com/

"This is a blog by Ben Brown.
He was an internet rockstar."



-adam

I'm certainly not logging the web. I think its closest to a bullshit-log.

------------------------------------------------
Ben Brown,
Internet Rockstar
----> http://blog.benbrown.com
------------------------------------------------

What, no XML feed?

I really ought to get one of those blog things too. Maybe once my life
gets actually interesting, or maybe when I have some original thoughts.

-josh

I'm not sure why I always think it's hilarious when people accidentally
respond to the entire list...

And I think the eavesdropping was just about to get good, but then 30 other
people started sending angry complaints to the entire list, followed by the
inevitable "UNSUBSCRIBE ME!!!" emails to the entire list.



-adam



-----Original Message-----
From: students-bounces@[xxx]
[mailto:students-bounces@[xxx] On Behalf Of mmille23@[xxx]
Sent: Tuesday, January 18, 2005 9:41 AM
To: students@[xxx]
Subject: Re: [Students] New LRL Policies

What can I say...I love the ass. Especially if there's a
nice juicy thicket directly above it. Speaking of thickets,
is our grumpy lover boy still around this semester? I guess
I should check his blog like a good little stalker.

Happy Tuesday, MM

---- Original message ----
>Date: Tue, 18 Jan 2005 09:20:12 -0600
>From: Cheryl Costello <cmcostel@[xxx]
>Subject: Re: [Students] New LRL Policies
>To: students@[xxx]
>
>oh, mary, you're such an ass-kisser ;)
>
>man, i'm totally hyped about this access, too! now i don't
>even have to volunteer to get in! cool! that's the way it
>should be.
>
>happy new semester.
>
>cheers ;)
>
>cheryl
>
>---- Original message ----
>>Date: Tue, 18 Jan 2005 09:14:35 -0600
>>From: <mmille23@[xxx]
>>Subject: Re: [Students] New LRL Policies
>>To: students@[xxx]
>>
>>Thank you! I appreciate the extended access and look
>forward
>>to attending the overview session.
>>
>>Mary Miller
>>Graduate Student
>>
>>---- Original message ----
>>>Date: Sat, 15 Jan 2005 23:42:41 -0600
>>>From: Ken Spelke <spelke@[xxx]
>>>Subject: [Students] New LRL Policies
>>>To: students@[xxx]
>>faculty_staff@[xxx]
>>>
>>> Dear Members of the GSLIS Community,
>>>
>>> Over the past several months, a number of
>>> conversations have taken place between students and
>>> Dean Unsworth, between students and myself, and
>>> among the staff of the Office of Information
>>> Technology and Research (OITR). These conversations
>>> have specifically focused on concerns over the
>>> management of our Learning Resources Lab (LRL)
>>> computing facility and access policies which some
>>> felt impeded students' ability to complete
>>> assignments and conduct necessary research. These
>>> concerns included perceptions that the LRL was
>>> unavailable when needed, and that lab users did not
>>> have adequate access to help desk staff within the
>>> LRL.
>>>
>>> The Dean and I agreed that it would be appropriate
>>> to conduct a complete review of the LRL policies and
>>> procedures. This review was initiated last Fall and
>>> included discussions with students, OITR and GSLIS
>>> staff, and a survey on LRL services which many of
>>> you completed toward the end of last semester.
>>> Based on responses from that survey, informal
>>> follow-up interviews with a range of students and
>>> faculty, and on-site visits to a number of computer
>>> labs on the UIUC campus, I am pleased to announce
>>> the following changes which have been made to LRL
>>> management procedures and which will take effect on
>>> January 18th, 2005:
>>>
>>> 1. The LRL doors will be open during the same days
>>> and hours as the User Services office (currently 8
>>> AM - 6 PM Monday through Friday).
>>>
>>> 2. Masters and Ph.D. students, as well as all
>>> faculty, will be granted 24x7x365 access to the LRL
>>> after attending an overview session about LRL rules,
>>> policies and procedures. The volunteer requirement
>>> for Masters students in order to receive 24 hour
>>> access will no longer apply.
>>>
>>> 3. User Services will staff a Help Desk "Annex" in
>>> the LRL on a regularly scheduled basis, with a
>>> target of around 4 to 6 hours per day, including
>>> some evening and weekend hours. Each semester,
>>> feedback from students will be used to determine the
>>> exact support hours for the coming semester. Fewer
>>> workshops will be offered to students in lieu of
>>> more direct help with technologies as needed while
>>> working in the LRL.
>>>
>>> 4. Students who found volunteering in the LRL to be
>>> a valuable way to broaden their experiences while in
>>> GSLIS are encouraged to consider an assistantship or
>>> practicum with OITR User Services as a way to
>>> provide a more formal opportunity for such
>>> experience.
>>>
>>>
>>> Given the breadth of the LRL policy changes
>>> described above, we are asking everyone, Masters,
>>> Ph.D. students, and faculty alike, to attend one of
>>> our upcoming policy overview sessions regardless of
>>> their current access status. Students who were
>>> volunteers last semester, and who had 24 hour
>>> access, will continue to have that access through
>>> the end of January. Please note that all users who
>>> wish to have the new 24x7x365 access will only be
>>> permitted to do so if they have attended one of the
>>> overview sessions, as described in number 2, above.
>>>
>>> I am very grateful to Martin Wolske and the User
>>> Services staff for all of their constructive
>>> thinking, flexibility, and willingness to try to
>>> help make the LRL a more user-friendly and valuable
>>> GSLIS resource.
>>>
>>> With best wishes to each of you for a successful and
>>> enjoyable Spring Semester,
>>>
>>> Ken Spelke
>>> Associate Dean for Information Technology and
>>> Research
>>>
>>>
>>> --------------------------------------------------------
-
>-
>>-------
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>________________
>>>_______________________________________________
>>>Students mailing list
>>>Students@[xxx]
>>>https://mail.isrl.uiuc.edu/mailman/listinfo/students
>>________________
>>_______________________________________________
>>Students mailing list
>>Students@[xxx]
>>https://mail.isrl.uiuc.edu/mailman/listinfo/students
>________________
>_______________________________________________
>Students mailing list
>Students@[xxx]
>https://mail.isrl.uiuc.edu/mailman/listinfo/students

Yeah, where the hell is the XML feed?!

XML IS THE FUTURE AND THE FUTURE IS NOW.

Hey, remember when Ben and I were doing that company thing about loosely
coupled web services and XML and contextual web stuff? I feel like maybe
now we're just getting to the point where we could explain that to people
and they'd realize we were really smart.

I mean, we were outputing XML versions of bookmarks AND wishlists and
notes! AND IT WAS IN 1993!

Goddamn. Was that really that long ago? If you'd done it five years
later, Google would've bought you by now.

I totally applied to work there, too.

-josh

hey, thanks for reminding us that we suck and could have been rich if only
we had timed things better.

jerk.

ps I like how dripmail has basically become an XML nerdout.


------------------------------------------------
Ben Brown,
Internet Rockstar
----> http://blog.benbrown.com
------------------------------------------------

I had never gone to Technorati til today, but then I got sort of a guided
tour, and I've got to say that their tagging stuff is pretty nifty. For
my own ego stroking, I get these options:


Things tagged with my name from 4 major web services:
http://www.technorati.com/tag/benbrown

And people who talk about or link me:
http://www.technorati.com/cosmos/search.html?rank=&url=blog.benbrown.com

That's pretty damned nifty -- nifty enough that I added their little
technorati rel="tag" links and started pinging their shit when I post.

now, if they'll give me a list of all 100,000+ tags they have in their db,
maybe we can do something interesting.


------------------------------------------------
Ben Brown,
Internet Rockstar
----> http://blog.benbrown.com
------------------------------------------------

So now that Google officially fixed the internet and saved my life and =
still doesn't charge for picasso, can we finally get rid of dripmail and =
go back to comments? Your website was so much better when there was the =
discussion board and you could post comment about daily entires and =
engage in light-hearted banter with fellow surfers from across the =
planet and when it was still on the viagra.org domain!