SemWeb Design Guidelines, was: Re: Getting on, and a virtual Meeting [via Semantic Web Interfaces Community Group]

Hi Roberto,

On 11/26/12 4:19 PM, Roberto García González wrote:
> I'm particularly interested in the design guidelines you mention. Are
> you talking about guidelines when developing user interfaces on top of
> semantic data?
Yes. And I can see two main contexts: Common UIs and innovative UIs.

Common UIs would be interfaces like entity views, schema editors and 
browsers, or faceted navigators. Things that have been created and 
explored before, partly with elegant examples available, but that still 
tend to look awful when done by SemWebbers.

Innovative UIs would be based on particular capabilities of semantics or 
RDF. The stuff we see in Hollywood movies...

It's rather hard to create guidelines for the latter (probably by 
definition, "guidelines for innovation" is pretty close to an oxymoron), 
so I would focus on the former for now.

> I'm really interested in this topic. I've already visited the user
> experience evaluation based on heuristics
> (http://en.wikipedia.org/wiki/Heuristic_evaluation) but they are usually
> too generic and in most cases reduce the evaluation to making the user
> interface follow the conventions of Web applications (or desktop or
> mobile applications depending on the kind of heuristics).
I agree. But looking at RDF-driven apps, I often think even those 
generic principles could be worth spreading a little more ;)

> It would be really interesting to build a list of guidelines specific to
> interfaces on top of semantic data but, is there something special in
> such user interfaces? And, if it is so, is that desirable from a UX
> point of view? Shouldn't it be better that users didn't have to bother
> about the underlying technologies?
Can't agree more, how can we identify that "special" and can we reduce 
it to something common? There seem to be some characteristics associated 
with the data structures we are trying to turn into applications, and 
with the app creators in the RDF world, which repeatedly lead to poor 
user experiences.

For example, RDF is by design meant to be re-purposed by arbitrary 
consumers, but generic applications usually require a higher cognitive 
burden than single task-optimised UIs. Is there a middle ground?

Or: RDF is meant to connect federated objects, which seems to require 
context-independent identifiers (hence the use of URIs), which in turn 
require a much higher cognitive effort than centralised simple labels. 
Could we get rid of them (including QNames) in UIs, or avoid the 
cognitive friction somehow? Could a system provide them if needed, e.g. 
by Data Engineers?

Or: Semantic technologies attract very smart people, and smart people 
tend to throw smart algorithms at tasks, including UI generation. Add a 
tendency to declarative approaches and you end up with designer 
nightmares like Fresnel. What could be a better approach that would work 
well with typical interaction design processes?

So ideally, I would love semweb UIs to behave just like ordinary UIs, 
but I think it is harder to get there due to the underlying technology, 
data structures, and social context.

One of the first things a person aware of usability heuristics, gestalt 
principles etc would probably do with a Linked Data page is grouping 
predicates in a meaningful way (labels, types, geo, social, commerce, 
external links, ...). It should be possible to come up with guidelines 
to simplify things like this. How far could these tasks be pushed down 
to the client (where designers feel more comfortable) before we lose 
benefits that require server-side smarts (such as search engine-friendly 
markup)?

More questions than answers still, but maybe a starting point...

Cheers,
Benji

>
> I would really appreciate hearing the opinions about these issues from
> people from the swisig list.
>
> Best,
>
>
> Roberto García
> Associate Professor
> Universitat de Lleida
> Jaume II, 69
> 25001 Lleida, Spain
> Tel. +34 973 702 742
> http://rhizomik.net/~roberto
>
>
> On Mon, Nov 26, 2012 at 1:33 PM, Benjamin Nowack <mail@bnowack.de
> <mailto:mail@bnowack.de>> wrote:
>
>     Hi,
>
>     My name is Benji Nowack. I'm working as an independent consultant,
>     creating mainly front-ends for (often RDF-based) web applications.
>
>     Won't have much time to contribute, but I'm interested in bringing
>     more end-user awareness to the process of building Linked Data tools
>     and applications. A few design guidelines could probably go a long
>     way already.
>
>     Cheers,
>     Benji
>
>     Paola Di Maio wrote:
>
>         Hope those of you who attended ISWC2012 had a fun and productive
>         week. Here is
>         how I suggest we kick this off: 1.  If you like, introduce
>         yourself to others!
>         and say a few words about what you'd like to do here 2. Please
>         take a look at
>         our draft agenda so far, and help edit items
>         http://www.w3.org/community/__swisig/agenda/
>         <http://www.w3.org/community/swisig/agenda/> 2. Let's [...]
>
>
>
>         ----------
>
>         This post sent on Semantic Web Interfaces Community Group
>
>
>
>         'Getting on, and a virtual Meeting'
>
>         http://www.w3.org/community/__swisig/2012/11/18/getting-on-__and-a-virtual-meeting/
>         <http://www.w3.org/community/swisig/2012/11/18/getting-on-and-a-virtual-meeting/>
>
>
>
>         Learn more about the Semantic Web Interfaces Community Group:
>
>         http://www.w3.org/community/__swisig
>         <http://www.w3.org/community/swisig>
>
>
>

-- 
Benjamin Nowack

   irl: florastr. 53, 40217 düsseldorf, germany
phone: +49.177.5241935
   web: http://bnowack.de/

Received on Tuesday, 27 November 2012 10:10:38 UTC