[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Myexperiment-discuss] Characteristics of workflows
From: |
Andrea Wiggins |
Subject: |
Re: [Myexperiment-discuss] Characteristics of workflows |
Date: |
Fri, 15 May 2009 13:34:10 -0400 |
On May 15, 2009, at 12:02 PM, address@hidden
wrote:
Marco,
faceted search & browsing with sorting on result fields of choice?
(I
am becoming addicted to the DBLP faceted browser)
That's more or less what I was thinking as well, especially if some of
the fields are limited in the range of values they present.
I find most of the items in Andrea's list not only reasonable but also
quite simple to deal with, as these are just metadata items that you
keep around, with the possible exception of "structural queries" i.e.,
on the workflow topology, which could be fun but can also be
challenging
to offer to users in some intuitive way.
Yes - I was thinking specifically of things that can be automatically
extracted or assigned. It reduces user overhead, is more reliable than
human assignment of some kinds of metadata, and structural features
are pretty objective but tiresome to handle manually. I think that
iteration strategies might be a useful thing to be able to examine/
index as well, but have had a hard time thinking up a use case for
that one.
Workflow topology could be an interesting characteristic to dig into a
little bit; network analysis measures could potentially be applied to
evaluate some of the topological characteristics in a useful way. I
was also thinking that an effort that tries to get at "complexity"
might start by having site users rate workflows (based on what they
can see online) with a simple Likert scale, maybe three or five
points, of low to high. It would be interesting just to see what the
inter-rater reliability would be, but to the degree that there is
agreement, the rated workflows could then be more closely scrutinized
for regularities in the structural characteristics that might
differentiate the groupings in a way that can be automatically
determined.
I particularly like the "quality-oriented" filters, i.e. on metadata
completeness (or other quality features)
It's also possible that making component metadata a part of the search
and discovery tools may provide some incentive to supply metadata. It
can be difficult to evaluate the potential usefulness of a workflow
without the metadata or code comments, or both.
Presence of code comments is another thing I'd like to be able to use
as a search facet, but it's also potentially trickier to handle. My
research group has done some NLP analysis on textual messages that
contain code snippets, and it's consistently a problematic factor for
processing the content.
oh btw:
I've seen this on another
social network site, labelled "include all photo-less?" which is
inelegant but perfectly descriptive.
you mean, as in the personals :-)
Actually, it's a social network for knitters and crocheters,
Ravelry.com - in context, "include all photo-less?" refers to a record
about an object made of yarn. :) The users are incredibly devoted and
grateful fans of the site, and it's some of the best SNS design I have
ever seen, bar none. It's pretty impressive but has members-only
access with a waiting list (to manage capacity), so awhile ago I
posted more about the site design and why it's so awesome at http://www.anikarenina.com/2008/06/14/raveling/
.
Cheers,
Andrea
Andrea Wiggins
PhD Student, School of Information Studies
Syracuse University
337 Hinds Hall
Syracuse, NY 13244
address@hidden
www.andreawiggins.com