I invite everyone attending e-Biosphere to join in a challenge entry that will i) illustrate the human sensor-net concept, and ii) be a collective demonstration of the semantic web in action. (To be fair, we should probably exclude ourselves from consideration during voting. But if we feel like we've won, I'll buy beer for everybody.) There will be a number of ways to participate:

1. Post pictures of London wildlife during the conference, and encourage friends back home (or in other interesting parts of the world) to also post pictures. The metadata will be published in RDF, and integrated with background natural history data that is already in RDF. This will enable us to pose and answer queries like "Show observations of species that are invasive where they were observed."; "Show the Northernmost observation of the Asian Longhorned Beetle"; "Show observations of predators and prey of threatened species"; etc.

If requested, pictures will be sent to a crowdsourcing tool for species identification.

2. You can volunteer to help identify images.

3. You can specify queries that you would like to see supported. This is important for our demonstration, since part of the social/semantic web's discrete charm is its flexibility in responding to ad-hoc queries. A wiki will collect queries.

4. You can contribute RDF, and help assemble the glue (e.g., owl:sameAs statements, etc.) that will bind all the RDF together.

For reporting observations, you are welcome to use Spotter [1], but can also simply submit data via spreadsheet or email.

Details on the other modes of participation will be forthcoming.

Here's a brief description of the concept in the context of eco-blogging:
http://ebiquity.umbc.edu/blogger/2007/0 ... -released/

(For the challenge, photos need not be embedded in a blog. They simply need to exist somewhere on-line, e.g. on Flickr.)

Suggestions are most welcome.
Joel.

1. http://spire.umbc.edu/spotter/

Thanks Joel for invitation!

Your suggestion is excellent. This is a wonderful opportunity for all of us to take part in a leading edge design experiment of biodiversity informatics.

I'll present this invitation also in the European Environmental Agency, Copenhagen, where I have been invited by Professor Jacqueline McGlade Jacqueline, Executive Director of European Environment Agency (EEA) to take part in an EU senior level expert meeting. The title of the meeting is challenging: ”Global setting for European environmental monitoring and observation: measuring what we must manage”, May 13 – 15, 2009.

Mauri

_________________
Dr. Mauri Ahlberg FLS
Professor of Biology and Sustainability Education
University of Helsinki, FINLAND

Joel,

A couple of thoughts.

Firstly, blogging is so last century! Surely the platform of choice for this sort of thing would be Twitter (see discussion at http://steve-wheeler.blogspot.com/2009/ ... c-web.html ). 140 characters is perfect for triples. Have people tweet with hash tags and aggregate those.

Secondly, owl:sameAs? How many people attending will know that that means? Furthermore, my understanding is that owl:sameAs is making a very strong statement, and is best reserved for statements that two URIs refer to exactly the same digital resource, e.g.

http://www.google.com owl:sameAs http://tinyurl.com/dehdc

or, perhaps

http://www.ncbi.nlm.nih.gov/nuccore/FJ237610.1 owl:sameAs http://www.ncbi.nlm.nih.gov/nuccore/227937158

In other words, owl:sameAs asserts identity. It seems dangerous to assert owl:sameAs for actual things (see http://www.connotea.org/tag/co-reference for some papers on this).

Rod

Mauri, Thanks. The EEA folks have their hands in a lot of citizen science, and it would be great to get them involved.

Rod:
1. My idea for the challenge was that people needn't blog, but simply post pictures on-line. (Though if some luddites want to blog, that's great - we'll handle it.) I definitely agree that harvesting Twitter feeds is something we should do (both within and without the challenge).

2. This is one of the biggest cans of worms I know.
I agree that we'll usually avoid owl:sameAs, but for different reasons than the one you give.
It is true that sameAs is too strong for most of our purposes. But it can give us a lot. For example, if we want to smush together everything everyone knows about a particular taxon, we need to deal with the fact that we're unlikely to be using the same URIs for that taxon. And so we need to somehow "equate", e.g.,
http://spire.umbc.edu/ethan/aedes_aegypti
with
urn:lsid:ubio.org:namebank:2593947

That way, if I assert
ethan:Aedes_aegypti rdfs:subClassOf viralVector

and someone reports an observation of
urn:lsid:ubio.org:namebank:2593947

and someone queries for observations of viralVectors, the observation will show up.

So should we say
http://spire.umbc.edu/ethan/aedes_aegypti
owl:sameAs
urn:lsid:ubio.org:namebank:2593947
?

There are (at least) two problems with doing so. One is that most of us model species (and other taxa) as classes, and using sameAs requires treating them as individuals, i.e. using OWL Full. The other problem, as you point out, is that owl:sameAs asserts identity, which means that any property of one is a property of the other. But that includes properties like where a taxon is on the tree of life, and, for Aedes aegypti, there's disagreement about that.

The thing is that, aside from properties regarding where things are on the tree of life, we generally do want the properties of ethan:Aedes_aegpti to adhere to uBio:2593947. To give a real example: one of the main points of our semantic eco-blogging exercise was to show how easy it is to define a "class of concern", and then to find all the observations of things belonging to that class. If I'm interested in species from the Global Invasive Species Database, I can simply define the class ISSG-GISDThing (as I do here: http://spire.umbc.edu/ontologies/lists/ISSG-GISD.owl) and then make class membership assertions about all the species in the database:

http://spire.umbc.edu/ethan/Aedes_albopictus
rdfs:subClassOf ISSG-GISDThing

etc.

I can further define a class ThingsICareAbout, and then assert
ISSG-GISDThing rdfs:subClassOf ThingsICareAbout.

Now I can issue queries like "Show all observations of things that I care about." (This would be a good way to manage Twitter feeds, if the world takes your advice and starts semantic twittering.)

----

Note that owl:equivalentClass gives the same problems and benefits. (If A and B are equivalent classes, then they share all superClasses and subClasses.) Further, semantic web browsers and rdf reasoners are, typically, highly idiosyncratic with respect to the built-in semantics that they provide. I've been meaning to do a collection of experiments to determine precisely how a variety of rdf constructs behave under a variety of rdf engines, but haven't yet.

In the meantime, since no lives are at stake, for the purposes of this demonstration, I propose using sameAs (for individuals) and equivalentClass (much more often, for classes).
I concede this may be a mistake.


Joel.