Re: WebBookmarks Queryable



On Tue, 2005-05-31 at 00:05 +0200, Martijn van Beers wrote:
> > If the actual link is an attribute it shouldn't be too hard for Hits to
> > recognize that hits from bookmarks and from web history are identical,
> > and only show one of them.
> > 
> > Maybe Hits should be a little smarter and be able to recognize related
> > and identical or related hits in things like bookmarks, web history and
> > email threads?
> >
> > Tiles could then be contained in some kind of grouping widget which can
> > be collapsed or expanded to show related and or identical hits.
> 
> I don't think you should ever present one link as multiple hits. As a
> user, I really don't care where the link was found, I just want to know
> about stuff that is relevant to my query.

I agree, but mostly when searching an unknown body of information
(external search). Very often a local search is made to *locate*
information you know or think that you have, and you may even have some
idea of what it looks like and where it came from.

> I can see how you might want to differentiate results depending on from
> which source they came (e.g. giving a link that is in bookmarks a higher
> score than a link that is just in the history), but that doesn't mean I
> want to see a double result if a link is in both, or that I even care
> where the result came from.

There is also the case where you get a hit in the web history (on the
actual page) of a bookmarked item, although not in the limited
information available in the bookmark itself. If Hits knows that the
pages are identical it can present the hit as the bookmark with the
history item as a related item with highlighting of the terms in the
relevant excerpt. 

The bookmark item may be easier recognizable to the user than the
(possibly <title>-less) history item.



-- 
Eirik Mikkelsen <eirikmik gmail com>




[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]