Re: ACTION: LeeF to summarize goals of resource gathering/recommending

Lee,

I certainly see the value of what you call 'adjudicated showcase' (I
have to keep up with my English:-). The problem I have: what is the
mechanism to rank the content of the list? Is it a mechanism that will
survive beyond the existence of this IG?

The immediate answer is to look at what amazon does, or the various
tagging clouds or similar (let us call that 'collective intelligence').
We could certainly try to setup such system. This may work well with
books and tutorials. Are we sure this would work well with software tools?

I am just trying to raise issues here, obviously. 'Collective
intelligence' mechanism may work well with a whole range of 'smaller' (I
know, define smaller:-) products and open software, which are relatively
straightforward to describe. For example, Jena is obviously richer in
functionality than, say, rdflib (forget about the different languages
for the moment). The comparison of Jena and, say, redland would become
more difficult, but the collective intelligence could still work.

But if we move to larger commercial products, I am not sure that the
same mechanism would work well. Comparing efficiency of large datastores
(say, Oracle against Kowari, stuff like that) requires much more
delicate comparisons, test, quite a lot of work. The same when it comes
to evaluation of larger products (@Semantics' full information
integration system, the various metata servers from cerebra, aduna,
franz inc, etc). I am not sure that the lambda developers would do such
comparisons in a reliable fashion. Reviews of more complex software is a
business by itself after all, there are people who make a living of it.
Can we replace that with any sort of automatic procedures?

There is the /. model, the amazon & co model for books, the tag clouds
kind of approach like on flickr, some sort of linkage ranking which is,
after all, what google does, the 'star'-s on a number of on-line
software stores, ...? Which of those would work for us? Which of those
could be self-sustained on long term?

I am just asking stupid questions... but I think we should have an
agreement on this before doing anyhing specific.

Ivan

Lee Feigenbaum wrote:
> I claim victory on this action.
> 
> See: http://esw.w3.org/topic/SweoIG/TaskForces/InfoGathering#preview
> 
> Wiki edits are welcome, but for possible email discussion, here's the text 
> I added:
> 
> """
> Information Gathering Output
> 
> Lee: I think that there are two possible types of collateral that an 
> Information Gathering task force could produce. The first is set of 
> comprehensive resource lists. This would be a set of lists of all Semantic 
> Web books, articles, tools, applications, tutorials, demonstrations that 
> we know of. Parts of these lists could be created automatically (e.g. with 
> DOAP or by scraping Amazon.com), parts could be created semiautomatically 
> (e.g. by maintaining sweo tagged links in del.icio.us and pulling those 
> into the output) and parts could be curated manually. What would this 
> collateral look like visually? Wing has suggested using SIMILE's Exhibit 
> to visualize these comprehensive list in a faceted search/browsing manner. 
> Kingsley has (I think) suggested using ?OpenLink's ODS infrastructure to 
> present this data.
> 
> The second option for what collateral could be produced is what Karen 
> called an adjudicated showcase. For this option, we would settle upon some 
> manner of determining the most mature, most accessible, and, indeed, best 
> resources in each category we examine. We would feature a select number of 
> books, or tutorials, of tools, and of demonstrations.
> 
> Personally, I prefer this second option. I think that limited, 
> high-quality lists are of much greater value than comprehensive lists for 
> educating and reaching out to people who are not already Semantic Web 
> believers. A developer wishing to try out SW technologies for a new 
> project will fare better with a list of the top 10 tools to implement SW 
> solutions rather than a comprehensive list of hundreds of tools in various 
> states of maturity. A LOB-manager wishing to familiarize herself with the 
> SW world would likely prefer a select list of the top 3 SW-for-business 
> books or presentations than a comprehensive list of scores aimed at 
> various audiences and in various degrees of polish. Of course, such 
> showcases could link to comprehensive lists, so perhaps we can have our 
> cake and eat it too...
> """
> 
> 
> Lee
> 

-- 

Ivan Herman, W3C Semantic Web Activity Lead
URL: http://www.w3.org/People/Ivan/
PGP Key: http://www.cwi.nl/%7Eivan/AboutMe/pgpkey.html
FOAF: http://www.ivan-herman.net/foaf.rdf