Evan Martin ([info]evan) wrote in [info]lj_research,
@ 2004-02-24 01:31:00
Previous Entry  Add to memories!  Tell a Friend!  Next Entry
foaf
LiveJournal now provides FOAF data for all users, via www.livejournal.com/users/username/data/foaf.

It includes interests data and friends lists. No word yet on how cache-friendly (read: scrape-able) this is, but it ought to be nicer than at least the userinfo pages.

(Post a new comment)


[info]ciphergoth
2004-02-24 04:49 am UTC (link)
I don't think it counts as "scraping" when it's meant to be machine readable...

I'll ask what usage they mind and what they don't in [info]lj_dev.

(Reply to this) (Thread)

[info]jamesd
2004-02-28 09:58 am UTC (link)
"I'm going to enforcing a new policy on bots that scrape the site (userinfo/FOAF/fdata/etc)". Since Brad inludes FOAF in scraping, it's included.

(Reply to this) (Parent)(Thread)

[info]ciphergoth
2004-02-28 10:45 am UTC (link)
Yes, it's clear from Brad's post in [info]lj_dev that it's OK for robots to make bulk requests for FOAF data, so long as they follow his User-Agent policy. I think this is the best way to collect friends list information and I plan to update TrustFlow to use it.

Incidentally, Brad (and [info]evan) is slightly mis-using the word "scraping" here - strictly speaking, it only counts as "scraping" if you're reading by script what was only ever meant for human consumption, but FOAF data is meant to be machine readable.. However, if you read "spidering" where "scraping" is written then the intent is entirely clear.

(Reply to this) (Parent)

[info]theorb
2004-02-24 05:31 am UTC (link)
I should probably dig in the spec for this, but I can't find the spec...

What is the mbox_sha1 field?

(Reply to this) (Thread)

[info]crschmidt
2004-02-24 05:36 am UTC (link)
SHA1 Hashed version of 'mailto:youremail@address.com'.

A unique identifier to you, so that FOAF data about you can always map who the "person" you are is, becuase no one else should have your email address, so no one else is likely to have your SHA1sum as a unique identifier.

(Reply to this) (Parent)(Thread)

[info]theorb
2004-02-24 05:48 am UTC (link)
Ahh! That makes a decent bit of sense -- in fact, it was my guess, after penetrating a bit of heavy standardseese. On the one hand, it isn't that useful -- I have quite a number of email addresses that I use regularly, and I doubt they sha1 the same way. On the other hand, I don't see a better WTDI that doesn't have severe privacy restrictions, or at least the appearance of them. (I can see a US-SSN:878-486-4879 pseudourischeme being handy here, though the danger exists of somebody computing SHA1s for every possible SSN. Note also that the first three digits can be guessed at by year and state of birth.)

(Reply to this) (Parent)(Thread)

[info]omnifarious
2004-02-25 04:39 pm UTC (link)

Actually, it should be a hash of your public key. :-) But, that will have to wait awhile until addressing things directly to public keys has wide protocol support.

(Reply to this) (Parent)

[info]omnifarious
2004-02-25 04:41 pm UTC (link)

Of course, I'm biased because I'm working on a protocol that addresses things to public keys. :-)

(Reply to this) (Parent)

[info]crschmidt
2004-02-24 05:38 am UTC (link)
FOAF data is entirely memcached, so the hit on the data is supposedly very lightweight, although I obviously can't comment from the SysAdmin standpoint.

I've already made a community for FOAF discussion including information on how to use it on LiveJournal: [info]ljfoaf. I highly recommend looking at it for anyone who might have inclination to do anything with FOAF data.

(Reply to this)


[info]ex_snej373
2004-02-24 10:24 am UTC (link)
Very cool!

It would be nice if the FOAF data were enhanced to include a link to your default userpic. It's always nice for a GUI client of this data to have a graphical representation.

(Reply to this) (Thread)

[info]crschmidt
2004-02-24 03:23 pm UTC (link)
See discussion at http://www.livejournal.com/community/ljfoaf/1779.html regarding that.

The basic reason why it's not included is that FOAF doesn't have a good depiction for this form of data.

(Reply to this) (Parent)

[info]marnanel
2004-02-24 08:23 pm UTC (link)
Excellent!

If friend-of data was in FOAF, this would mean programs like Joule could ditch the API they use. It would also mean that it'd work for communities and so on. Any chance of that, or at least of a special version for that? (The FOAF standard doesn't seem to have any concept which even vaguely matches up to friend-of, but having a special URI for the version with friends-of would solve that.)

Is there any chance that could happen?

(Reply to this) (Thread)

[info]evan
2004-02-25 08:50 am UTC (link)
It may also be possible to add extra namespaces to the FOAF output. I know that's what we did for /stats/latest-rss.bml for supporting music and mood.

(Reply to this) (Parent)

[info]mcfnord
2004-03-05 04:52 pm UTC (link)
http://www.livejournal.com/misc/fdata.bml?user=mcfnord

Does this have everything Joule needs?
I'm goin' nuts about this interface.
How long has it been here?!
I love it!

(Reply to this) (Parent)(Thread)

[info]marnanel
2004-03-05 05:02 pm UTC (link)
Well, that's the interface that Joule uses, but I thought the URL wasn't supposed to be generally public. Did that change?

(Reply to this) (Parent)(Thread)

[info]mcfnord
2004-03-05 05:12 pm UTC (link)

a post made today in the memedev community.

sorry i didn't know.

this is THE INTERFACE i've walked the earth to lay eyes upon... and now i learn it's a private party!

pain!

(Reply to this) (Parent)(Thread)

[info]marnanel
2004-03-05 05:16 pm UTC (link)
Ah, no, I see it's just become public information:

http://www.livejournal.com/bots/

(Reply to this) (Parent)(Thread)

[info]mcfnord
2004-03-05 05:18 pm UTC (link)
that's great cuz i like the completeness and simplicity.

All I need is Location data and I'd swear off infopage scraping for life.

(Reply to this) (Parent)

Create an Account
Forgot your login or password?
Login w/ OpenID
English • Español • Deutsch • Русский…