RE: RE: [hcls] A map of the Semantic Web for life science and health care

Matthias,

I think your proposed definition of 'Map' is to general and can be misinterpretted too easily:

I see several maps from your description:
1. Basic life science concepts ( Uber Semantics? why not define with OBO?)
2. Data communities (and their record types)
3. Larger research and clincial interactions (roles and context)
4. ???

Calling it "Mapping the Semantic Web for life science and health care" is very confusing to me. Perhaps pick one of the above items, state it clearly, and focus only on that space.

Also, I think we should  get in the habit of saying "Proposed..." on some of these wiki pages ; otherwise it sounds to imposing without discussion or process...

Eric


-----Original Message-----
From: samwald@gmx.at [mailto:samwald@gmx.at]
Sent: Fri 6/22/2007 2:34 PM
To: Kashyap, Vipul; Eric Neumann
Cc: public-semweb-lifesci@w3.org
Subject: Re: RE: [hcls] A map of the Semantic Web for life science and health care
 
The wiki page http://esw.w3.org/topic/HCLS/HCLS_semantic_web_map has been updated. The wiki page now contains instruction on how you can edit the contents of the map (you need to edit the Graphviz code on the wiki page, it's simple).

After a lot of thinking I have finally decided on the style for the visualization. I chose a geographical metaphor, here is an example (with arbitrary connections):
http://neuroweb.med.yale.edu/senselab/temporary_files/map-draft-2.png


--- Short description of the visualization style ---

* Fields of information in life science and health care are represented as separated 'countries'.
* If two 'countries' represent related fields of information, they share a common boundary.
* Semantic Web resources (mostly ontologies) are represented as 'towns' in these 'countries'.
* If two ontologies are connected (by sharing URIs), this is visualized as a 'railway line'.
* 'Towns' have different 'sizes'. The size depends on three scores: Interoperability (how well does it play along with others?), coverage (how much of its field of information does it cover?) and maturity (is it still very buggy and rapidly changing?). Each score is a number between 1 and 3. The end score for each ontology is the lowest of these scores, i.e., an ontology only gets a score of 3 when all of the separate factors are scored with 3. For example, if someone creates a very large (score 3) and mature (score 3) ontology, but it does not even try to interoperate with anything else (score 1), the whole ontology will only achieve a score of 1. The higher the score, the larger the size of the 'town' on the map.
* 'Lakes' can be used to separate 'countries' that happen to be positioned near each other, but are unrelated.

--- Reasons for choosing this visualization style ---

* Easy to draw.
* Intuitive.
* Easy to extend in the future. Ideally, we could make bi-yearly snapshots of the map and observe how our 'towns' are growing, become more numerous and interconnected, and spread out into areas that were previously uninhabited.
* It expands the metaphor of bioinformatics nation(s) coined by Lincoln Stein (link to Nature article). This metaphor was also used recently by Carole Goble in her talk at Banff (pdf file).
* Relatedness between fields of information can be visualized as common boundaries. Compared to 'clouds', this approach has the advantage that 'countries' can stretch and bend around other 'countries', which gives more freedom in representing relatedness. It also saves space.
* It allows us to differentiate between ontologies that connect to other ontologies and those that do not. If there are several disconnected ontologies in one area, we have a problem that needs to be solved.
* It shows not only the resources that we have created so far, but also the goals we have to meet in the future ('uninhabited areas'). This is very importand to set things into perspective.

--- Next steps ---

The next step is the creation of the graph with the Graphviz syntax. Instructions for doing this are on the wiki page. The current graph generated by Graphviz is at
http://neuroweb.med.yale.edu/senselab/temporary_files/sw_map.gif

I will update the graph during the next few days, but help would really be appreciated. The current version only contains those things that came to my mind first, so only the neuron-related stuff is well represented at the moment. 
I'm also open to suggestions for re-organizing the 'countries' / fields of information. Of course, compromises need to be made, and it should not get too complicated.

** If you see that something important is missing from the map, please go ahead and add it **


cheers,
Matthias Samwald

----------

Yale Center for Medical Informatics, New Haven /
Section on Medical Expert and Knowledge-Based Systems, Vienna /
http://neuroscientific.net




Vipul wrote:
> Matthias

> Thanks for kick-starting this effort.
> 
> I have cleaned up the page and introduced a set of "categories". 
> 

> Have also added a series of "maps" from the WWW 2006 Tutorial on Semantics
> in
> HCLS.
> 
> These include
> 
> -          Current and Goal State Ecosystem Diagrams of all the
> stakeholders in
> the HCLS space
> 
> -          An Activity Venn Diagram across HCLS
> 
> -          Data and Knowledge Taxonomies and Maps for different parts of
> the
> HCLS space.
> 
> As usual, comments and feedback welcome.
> 
> As of now, I have uploaded the PNGs into the wiki as attachments. Should
> you
> want to store them in the Yale Server and link them
> 
> to the wiki page, I will e-mail them to you as well.... (separate e-mail
> to
> prevent cluttering of HCLS list).

> 
> Cheers,
> 

> ----Vipul

-- 
Psssst! Schon vom neuen GMX MultiMessenger gehört?
Der kanns mit allen: http://www.gmx.net/de/go/multimessenger