Updating search results on-the-fly

16 Oct 2006 - 10:33am
8 years ago
2 replies
281 reads
jbellis
2005

In our desktop application, which is essentially a document management/pubishing system, we have a powerful, multi-purpose search function. After retrieving a list of hits, users can perform any number of actions on the result rows. For instance, they can look for unassigned items and assign them (to users).

The question is, "Should the results list automatically remove rows that are disqualified each time the user performs an action?" For instance, if you find 10 unassigned items and assign one, should the results remove that one... without the user explicitly re-executing the search?

There are reports of users being confused because the results do not currently auto-update, and in fact, I can recall experiencing this perception myself. I believe there's a split between "search" and "worklist" that's at play here.

Note a subtlety: There's no issue about automatically "re-invoking the search." We won't consider that; it would be disruptive. I only want to know, should I recommend that we automatically remove disqualified rows?

Are there models out there for searches that auto-update? Or should I just accept that if it supports "worklist" functions, it should meet the expectation of updating, like a worklist?

Thanks, Jack

Comments

16 Oct 2006 - 10:52am
Mark Canlas
2003

"Should the results list automatically remove rows that are disqualified
each time the user performs an action?"

Going purely off of anecdotal experience...

Consider the "Unread mail" search folder in Outlook 2003. It searches for
all unread items in your various mail folders. Depending on your settings,
the act of merely clicking on an item will "mark [it] as read". Unread is
bold, read is normal text. The list doesn't update unless I navigate away
from the screen (specifically, it doesn't take away items from the list, but
will add them if more unread mail comes in).

I think this is the best model, provided you give feedback to user that an
action has occurred. Sometimes, actions can be performed unintentionally,
especially those executed by clicking or manipulating row items (sometimes
you just fumble or jiggle the mouse around and stuff happens. Cat across the
keyboard?).

If you provide feedback and don't change the results, in the event of a
mistake, it allows the user to change the items back to the way they were.
Otherwise, the results would have disappeared from the list, leaving the
user to wonder, "What did I change... Oh well. It's gone now. I wonder if
that was bad."

Conversely, look at iTunes. Let's say I'm search for "Pop" songs and a rock
song comes up mislabeled as pop. The moment I change it to Rock, the song
disappears from the search list immediately. What if I made a mistake? What
if I had more info to update? Not only was it a rock song, but it was done
in the 80s by so and so artist. The file is gone from the list. Oh well.

What's a "worklist"?

-Mark

16 Oct 2006 - 11:19am
jbellis
2005

Mark,
Great point, marking/highlightinging/bolding. That answers some concerns
about losing the original list.

By worklist, I'm suggesting that there's a distinction between purely static
information retrieval vs. a list that you take action on. It's probably an
unsupportable distinction, though. Even searching the web, there's the
action of visiting the hits, and the links change color.

Thanks, Jack
----- Original Message -----
From: "Mark Canlas" <mark at htmlism.com>
>
> Consider the "Unread mail" search folder in Outlook 2003. It searches for
> all unread items in your various mail folders. Depending on your settings,
> the act of merely clicking on an item will "mark [it] as read". Unread is
> bold, read is normal text. The list doesn't update unless I navigate away
>
> I think this is the best model, provided you give feedback to user that an
> action has occurred. Sometimes, actions can be performed unintentionally,
> What's a "worklist"?
>
> -Mark
>
>

Syndicate content Get the feed