Information Flow #7 ~~ November 2002 ~~ Advanced Queries through Interaction

~~~ Ramana Rao's INFORMATION FLOW ~~~ Issue #7 ~~ Nov 2002 ~~~~~

"I find that a great part of the information I have was
acquired by looking up something and finding something else
on the way."

-- Franklin P. Adams

~~~ IN THIS ISSUE ~~~ November 2002 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~

* Announcements
* Advanced Queries through Interaction
* Return to the London Underground
* Links and Resources

~~~ Announcements ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Web SITE REDESIGN Launched, earlier this month. Comments?

FORWARD to a FRIEND: If you have find this newsletter useful
or interesting, please forward it as a holiday treat or an act
of good karma to somebody else. Note that all issues are
archived on my Web site.

~~~ Advanced Queries through Interaction ~~~~~~~~~~~~~~~~~~~~~~~

It is well-known that most users don't use advanced queries.
They hardly even use multiple word or phrase queries. I have
long believed that end users can perform what amounts to quite
sophisticated queries. *If* they are allowed to do so in the
right way.

Through interaction. By sequences of simple actions with
well-arranged opportunities for bits of human sensemaking.

And by blending searching and browsing. The two basic
information access approaches can be distilled down to the
following two unit interactions:

Human: query (question)
Computer: result (answer)

Computer: display menu (display index or view)
Human: select (navigate)

The first interaction underlies traditional search, and in fact
database query systems in general. Easy enough to see when/how
this interaction might break. You must have a question. You
must master the artificial language of a particular system. And
the answer must be available or computable.

The second interaction is the essence of browsing, navigating,
exploring. The system displays a menu, listing, or page; or a
view of a structure or space. The user digests the display to
decide what to do next. Here too the difficulties and breakdowns
are apparent. Making organized and effective displays takes a
great deal of effort. As can understanding and using them.

Sequencing chains of these interactions, you end up with the
familiar sequences of say submit-submit-submit (note who is
submitting to what) or click-click-click (lost in hyperspace).

However, nothing stops us from mixing the two interactions. In
fact, most Internet users commonly search to get to a promising
site and then browse from there, sometimes returning to search
with a better sense of what they want or how to say it.

Looking again at the interaction units, notice that the human and
computer switch order in a call and response pattern. So if a
response can be shaped into a call of the other interaction, you
can get into a nice rhythm between the user and the system.
Starting with a query, you get the following:

Human: query
Computer: display result
Human: select

Indeed, that's what's happening when a user clicks on a result
item. But here we can introduce the "well-arranged opportunity
for a bit of human sensemaking."

The result could show indexes over matching items, arranged by
various attributes including subject headings, contained named
entities like people, organizations, and places, and clusters
over concepts nearby the query term. The interaction sequence
could be:

Human: query
Computer: display result index
Human: select query
Computer: display result index

In this sequence, the user is *advancing* the query through
selection. For example, the query could be a single word, and
the select could add a restrictions e.g. a category or a named
entity. This interaction converts the advanced query form into a
series of clicks, when and if they make a difference.

So far we started with a query, but we could easily start without
a query in a more browse-style interaction:

Computer: display collection index
Human: select query
Computer: display result index

An example of this sequence is starting at the top of a directory
and clicking down the category tree.

These various dialogues showing how specifying and selecting,
search and browsing can be combined in a way that leverages their
strengths and avoids their weaknesses. Such sequences can
generate quite sophisticated queries, but in a manner that allows
users to build them incrementally as part of the process of
exploring and better understanding their needs and available
content.

In a nutshell, these rich sequences are about expanding the
information access experience from a pure command line paradigm
to a full workspace paradigm. This transformation is analagous
to what the graphical user interface did for the personal
computer. Unix hackers are able to do quite amazing things by
composing operators at the command line, as are Dialog or
Lexis-Nexis supersearchers with advanced search. Interactive
environments can provide the same power to everybody else.

By now, on the Internet, users do in fact rely on a variety of
systems that offer rich interaction sequences, as well as, a
variety of well-organized sites that allow efficient browsing
through indices and other tools.

Intranets and extranets still lag in offering these options. The
portals and content access systems that have been deployed tend
toward search only systems. And the private content collections
are generally not well-organized, for lack of resources and
attention. This will change over the next few years as software
and new practices for organizing and interacting with content are
rolled out.

I'd love to hear your stories. What techniques do you use for
searching the Internet? And do you have available to you tools
on your intranet, extranet, or private systems that start to
offer richer interaction?

~~~ Return to the London Underground ~~~~~~~~~~~~~~~~~~~~~~~~~~

Last month, I made another trip to London. I didn't expect to
write a sequel to last month's article, but something happened on
the way to the Tube.

A fire fighters strike!

You ask, what does that have to do with the Tube? Well, it
closed all the Tube stops without escalator or stairway access.
Not just a few stations.

And my JPG file wasn't updating automatically. So for my first
meeting, I took a cab to Picadilly Circus.

After the meeting, my colleague Russell and I needed to get back
out to North London for dinner. We needed to take the Victoria
line from Picadilly Circus. Straightforward enough, but the
Picadilly line was closed and so there was massive overflow onto
the Victoria.

On the platform, along with us was a *crush* of people jammed
right up to the yellow band at the edge of the platform.

"Stand clear, a train is arriving." Now that's a gap worth
minding!

We watched several jammed trains arrive, and the interchange of
say a dozen people getting off at each door, and I do believe
more getting on. We watched a couple of arrivals and departures,
before we got our turn right near the yellow line.

Inside the car, we couldn't have fallen over if we wanted to,
really we were one with the beast humanity. As we had come down
into the Tube, Russell had said, "I couldn't do the daily commute
into London, any more." I had said, "Well I might prefer say a
20 minute commute by Tube to say the 40 minutes I have at best
early in the morning in the Bay Area." Now as we jostled toward
our connections, a couple of stops down the line, Russell caught
my eyes. Smiling, he said, "Re-evaluating the tradeoff?"

At the Warren Street station, the flow was greater away then
toward our destination of the northbound Northern line platform.
And in fact, the platform was less jammed, and the people there
didn't have the same degree of urgency about the next train. Two
observations, one based on a simple mathematical model and the
other on social cues. That should work. I commented, "Looks
like this line should be better."

"We'll see," said Russell.

The next train came in, and it was as jammed as the train we'd
been on. Foolish me, my local observations didn't tell me about
the bigger picture.

I'm the last person to get onto the carriage. In fact, I'm
standing right at the door edge, my head poking out across the
curve of the door. I look down the line and see a man standing
tall like me still. A bell rings, I hear the door start to
close, I see the man pivot his head in, and I do the same to
clear the door. My eyes catch a red sign on the door that says,
"Obstructing the Door Can Be Dangerous."

On the train, an announcement tells us that we can't get off at
the station we want. Or the station before. So we get off two
stations ahead of our destination.

Emerging from the station, Russell asks at a newstand right in
front of us, "Which way toward Belsize?" I waited, then suddenly
I had the realization that a right turn would continue us in the
direction the train was going. The exit had been simply a turn
right onto the escalator and a turn right out the exit, so it was
quite easy to "feel" the right direction.

But there was also a feeling of unease. A dying of the
expectation of being able to rely on spatial skills and on maps
and models.

The next couple of days, with the strike and closures for
maintenance, I asked people more often for information, and
several times this altered my course. Rather than planning and
predicting with incomplete maps and models, I waited to see how
things would turn out. And navigated onward to my destination.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

The Access Guides and the Contradictions of Design
Philip E. Agre
~> http://dlis.gseis.ucla.edu/people/pagre/access.html

I have never met or interacted with Phil Agre despite 20 years
of near crossings and perhaps ten strong connections. He was
at the MIT AI lab when I was an undergrad at MIT. I remember
reading one of his papers and appreciating its broad view of
intelligent behavior. This essay, which I found recently,
looks critically at a recent edition of the Access Guides, and
speaks broadly about design.

The British Library
~> http://www.bl.uk/about/didyou.html

In my recent London trip, I made a pilgrimage to the new
British Library. I heartily recommend visiting. One detail
from a Library pamphlet met my fancy: "The Victoria Line runs
below the library. At its closest, it approaches within 6
meters of the basement walls. When a train passes it can be
clearly heard in the basement"

"Mind the Gap"
~> http://www.bbc.co.uk/dna/h2g2/alabaster/A676956

The Life and Times of the London Underground Map
~> http://www.bbc.co.uk/dna/h2g2/alabaster/A673517

Going Underground
~> http://www.goingunderground.net

The Way Out Tube map
~> http://www.goingunderground.net/#wayout

The Great Bear
~> http://www.dareonline.org/artwork/patterson/patterson3.html

Triggered by comments from readers, I've googled to some very
London Underground arcana on the web. I had no idea that the
"Mind the Gap" was so culturally imbued (t-shirts, gameshows,
etc.), but it isn't surprising now (he says looking out the
rear view mirror). The articles and links by the Mole, an
avid observer of Life Underground, are fascinating.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Ramana Rao is Founder and CTO of Inxight Software, Inc.
Copyright (c) 2002 Ramana Rao. All rights reserved.
You may forward this issue in its entirety.

See: http://www.ramanarao.com
Send: [email protected]
Archive: http://www.ramanarao.com/informationflow/archive/
Subscribe: mailto:[email protected]
Unsubscribe: mailto:[email protected]