Types of Faceted Search

27 Jul 2009 - 8:17am
5 years ago
8 replies
1768 reads
Vishal Subraman...
2005

There seems to be two major methods of faceted search
design- for the sake of my own vocabulary- I call them 'static' and 'dynamic'
(I suppose there would be some formal terminology, but I couldn't find
any).
The Static model (Eg: Search for laptops in Yahoo Shopping-
http://shopping.yahoo.com/search;_ylt=AitSGJz_qz4V.Woy9WLIzSQbFt0A?p=laptop&did=)
typically involves clicking through one facet element at a time (multiple
select being present occasionally) causing the search results to filter
accordingly

The Dynamic model (Eg: Flights in Kayak- http://www.kayak.com/r/K0nJrU)
enables the user to quickly navigate different options to find the needed
product/ service.

In my mind, the dynamic model is a much better user experience because it
allows for faster modification of facets. But there seems to be a dichotomy
of implementation based on the product category. Travel products generally
use the dynamic model, while Shopping uses static. The obvious difference
between the two is in the temporal nature of the data involved. Travel data
is transient. Shopping, not as much- the static model hence allows the
creation of 'pages' for significant facet elements.

I can however see the dynamic model being tweaked in such a way that it
supports data with a longer shelf life and hence SEO friendly etc. If that
is indeed the case, is there any other reason to use the static model? Am I
missing something here?

-Vishal

Comments

27 Jul 2009 - 10:29am
John Lucas
2009

IMO, The choice largely depends on your implementation options. From
an IxD perspective I think an Ajax based implementation is definitely
superior as far as speed of refresh. But it might limit your browser
choices. Speaking for the static example, the same widgets in the
Kayak example can be implemented server side as well so you don't
necessarily lose capabilities, it's just slower.

>From a designer's perspective i'd work out the types of facets,
views and sorting methods you'd like to use and then work with your
implementation team to see what your options are.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Posted from the new ixda.org
http://www.ixda.org/discuss?post=44132

27 Jul 2009 - 11:31am
morville
2010

Dynamic beats static from a UX perspective. I agree it comes down to a
question of implementation (cost, performance, accessibility). I love how
results fade *before* you release the sliders at VW UK...

http://www.volkswagen.co.uk/#/new/

...but I don't love waiting for the application to load prior to
interaction.

Peter Morville
President, Semantic Studios
http://semanticstudios.com/
http://findability.org/

-----Original Message-----
From: discuss-bounces at lists.interactiondesigners.com
[mailto:discuss-bounces at lists.interactiondesigners.com] On Behalf Of Vishal
Iyer
Sent: Monday, July 27, 2009 9:17 AM
To: discuss at ixda.org
Subject: [IxDA Discuss] Types of Faceted Search

There seems to be two major methods of faceted search
design- for the sake of my own vocabulary- I call them 'static' and
'dynamic'
(I suppose there would be some formal terminology, but I couldn't find any).
The Static model (Eg: Search for laptops in Yahoo Shopping-
http://shopping.yahoo.com/search;_ylt=AitSGJz_qz4V.Woy9WLIzSQbFt0A?p=laptop&
did=)
typically involves clicking through one facet element at a time (multiple
select being present occasionally) causing the search results to filter
accordingly

The Dynamic model (Eg: Flights in Kayak- http://www.kayak.com/r/K0nJrU)
enables the user to quickly navigate different options to find the needed
product/ service.

In my mind, the dynamic model is a much better user experience because it
allows for faster modification of facets. But there seems to be a dichotomy
of implementation based on the product category. Travel products generally
use the dynamic model, while Shopping uses static. The obvious difference
between the two is in the temporal nature of the data involved. Travel data
is transient. Shopping, not as much- the static model hence allows the
creation of 'pages' for significant facet elements.

I can however see the dynamic model being tweaked in such a way that it
supports data with a longer shelf life and hence SEO friendly etc. If that
is indeed the case, is there any other reason to use the static model? Am I
missing something here?

-Vishal
________________________________________________________________
Welcome to the Interaction Design Association (IxDA)!
To post to this list ....... discuss at ixda.org Unsubscribe ................
http://www.ixda.org/unsubscribe List Guidelines ............
http://www.ixda.org/guidelines List Help ..................
http://www.ixda.org/help

27 Jul 2009 - 11:48am
Vishal Subraman...
2005

The VW example leads another question- search v/s filtering. Its
always 'faceted filtering' not
search, isn't it? For example, in http://autos.yahoo.com/carfinder (interact
with any one facet, say price $15k- $25k to get this started).
Although none of the makes are shown selected, the results are for all
makes selected. The
act of selected a particular make filters the results. Should there have
been better indication that all makes are selected in the initial state or
is this a case of good inconsistency from a IxD perspective?
-Vishal

On Mon, Jul 27, 2009 at 12:31 PM, Peter Morville <
morville at semanticstudios.com> wrote:

> Dynamic beats static from a UX perspective. I agree it comes down to a
> question of implementation (cost, performance, accessibility). I love how
> results fade *before* you release the sliders at VW UK...
>
> http://www.volkswagen.co.uk/#/new/
>
> ...but I don't love waiting for the application to load prior to
> interaction.
>

27 Jul 2009 - 1:31pm
ambroselittle
2008

On Mon, Jul 27, 2009 at 12:48 PM, Vishal Iyer <vishaliyer1 at gmail.com> wrote:

> The VW example leads another question- search v/s filtering. Its
> always 'faceted filtering' not
> search, isn't it?
>

----

We discuss this in the Faceted
Navigation<http://quince.infragistics.com/#/Main/ViewPattern$pattern=Faceted+Navigation>pattern
in Quince. One of the points in the rationale is this:

This is sort of the inverse of the common “Advanced Search” screen that many
have employed over the years, and it is better because it doesn’t require
the user to know or specify these details up front. It is encouraging and
usually optional because they’re not confronted with a big, complex form
they have to fill out in order to start exploring the data and, if
implemented well, it actually encourages more exploration by fast updating
based on progressive facet selection.

I think the distinction in facet kinds (i.e., those you might pre-select for
a search vs filter on with facets) should be based on an understanding of
the most crucial facets for your domain *that people are likely to know (and
want to filter on) up front*.

For instance, in flights, knowing place of departure/arrival and dates are
the most crucial and are the most known of all the facets. Maybe keyword is
the most feasible starting facet in a domain like Amazon that sells
everything, but you can also choose a department if you think you know it
(though it's not required, which is important).

Also, it seems to me that if your total set is (and will remain) small
enough not to cause anxiety and/or perf problems, you could even just start
with an all listing and filter down from there. But I guess this might be
playing with fire unless you do enough testing to validate.

-a

27 Jul 2009 - 2:20pm
Audrey Crane
2009

See also the related discussion on filters:
http://www.ixda.org/discuss.php?post=43844

In usability of the site I'm working on and others, which has mostly
newbies as users, dynamic filters have the drawback of being miss-able
("Is that navigation on the left, or...?) and scary ("Ahhh! What is
that? Too complicated!") Not universal reactions to every
implementation, but consistent across several users and several
sites.

There seem to be some new tools starting to emerge. The uBid example
listed in the filtering thread is interesting.

You could also think of Pandora as a sort of clever filter, where
there are too many variables to ever present in advanced search or a
set of filters, so you make one choice, and then refine from there by
liking or disliking proposed matches. The system manages the filters
behind the curtain, but it's happening nonetheless.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Posted from the new ixda.org
http://www.ixda.org/discuss?post=44132

27 Jul 2009 - 5:58pm
AlokJain
2006

Vishal,

To your core question, the choice of static v/s dynamic is
essentially a question of your "design language" you want to use.
In other words if you do it in one place, people expect it
everywhere. I always imagine any software as a dialogue between user
and the computer. If the computer responds to a user question in one
way then user forms a mental model of how the software operates and
expects similar behavior everywhere.

Take an example of an app that extensively uses inline editing (jut
click on something and start editing) , then for a specific area due
to whatever limitation you ask the use to click on a link to edit
something. User would make mistakes if they are used to of the other
parts of the software.

If the design language you establish throughout is very dynamic, then
the search should not be an exception. So consider this also in the
design decision.

I the search is the major function and is establishing the design
language for the dialogue between the user and the software, then it
is really a question of skills and money. The dynamic model in my
opinion bring following value:
1. Faster response
2. Maintains continuity so user doesn't have to re-orient
3. Communicates the changes (or has the potential to) much better

Cheers
Alok jain (AJ)

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Posted from the new ixda.org
http://www.ixda.org/discuss?post=44132

27 Jul 2009 - 2:38pm
poomoo
2009

Although the 'dynamic' version is better because of the instant
feedback/results, both versions need to be implemented for two
reasons:

1. Users without JS - they need to navigate through page refresh
2. SEO - Search Engines can't index dynamic pages

Ambrose, I'm very interested in seeing the discussion you linked
although, I'm very much less interested in installing Silverlight to
do so :(

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Posted from the new ixda.org
http://www.ixda.org/discuss?post=44132

28 Jul 2009 - 3:25am
Timesheet
2009

Alok

Your point is correct to some extend as Dynamic model would be more
expressive when we take a limited variation in the data set to be
presented, as in the example of flight booking - after providing the
initial details of departure and arrival there is not much variation
left to present the list of flights available,
whereas if we go for browsing something like laptop or Timesheet
software
then considering the variation availabele in the market you cant
filter them with few variables, and in this case the no. of variable
would increase drastically whcih would increase the processing time
for Dynamic model and here Static model would be the best option.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Posted from the new ixda.org
http://www.ixda.org/discuss?post=44132

Syndicate content Get the feed