[rdfweb-dev] New tool: BEAT.THE.FOAM v. 0.1

Zaczek hydrozag at poczta.onet.pl
Wed Dec 1 09:58:55 UTC 2004


>I\'m worried about the smushability of FOAM data. (My concerns could
>mostly be resolved if foam:memes was a functional property. It\'s more
>work for the smushers, but c\'est la vie.) However, you seem to be
>worried about the sets of interests and how they change. You may want to
>consider not talking about the Person at all, since you don\'t seem to be
>using the tuples that it generates.

Hi. I think you are right that the FOAF with a memetic profile is just
a starting point. A parser could analyze a bunch of such files and
generate a pure FOAM document for each meme. It could track how
many "hops" there are between memes: 0 if they are in the same memetic profile,
1 if they are one <foaf:knows> away. There could be a syntax for describing that.
There's still a lot of work to be done here obviously.

This analysis can give us a map of linguistic closeness of memes. 

I have thought that there would be a benefit to stick with the FOAF syntax 
somehow. I was thinking of using some existing tools for that, such as FOAFNaut.
If we substitute a Meme to a Person, then we can use existing FOAF-aware software for
looking at memes. I mean disguising a Meme with properties that make it
look like a Person to a FOAF parser. Is this legitimate? 


>After reading the "Why you would need it" and "Algorithm" sections of
>[1] and comparing it to what you\'ve written above, I fear I may not
>understand your goals. How do you find out how a meme got to a certain
>node? The RDF you\'ve written doesn\'t seem to be able to answer that
>question. I like your idea of studying meme propagation, but I\'m worried
>that you haven\'t the right classes and properties to do that yet.

When you raised this question, the Dublin Core came to my mind. It has
a "Creator" and a "Date" which could be attached to each meme, but it
doesn't have a versioning vocabulary. So PRISM with its <prism:isVersionOf>
might be necessary. The "creator" is the first person using a meme, those
who replicate it might refer to the original meme with isVersionOf.


The example application I made - BEAT.THE.FOAM has to deal with the
fact that DEL.ICIO.US as a datasource is not of course FOAM-aware.
It doesn't make apparent how a user started using a tag and where it
was gotten from. We can only assume that if the user is subscribed
to somebody's feed and a tag appeared there at a certain time, and
later the user used the same tag, we are faced with meme propagation.
It's possible to extract information about tags and the date of their use from
the delicious RSS feed (each item has a date and the tags used, so we can 
associate the date with a tag, but the same tags are used in different items so
it's a bit tricky. I would use the earliest occurrence, but then the RSS as such
does away with earlier items and you never see them again.)

The ideal tool for gathering data about meme propagation is the Google Desktop
Search. It knows everything you read and write and therefore could hypothetically 
produce a fairly good assessment of the memes that a user replicated using a PC.
But Google doesn't produce RDF and we don't want to disclose everything
we read and write in the form of FOAM feeds :)

Sorry if I wander off purely RDF topics here. Anyway, in order to be really able
to track the spread of memes we not only need an extension of the FOAM vocabulary
with something like PRISM, but also FOAM-aware software that the users could actually
use to replicate memes. Blogging software is ideally suited to be extended in such a way.

Greets,
Zaczek





More information about the foaf-dev mailing list