On Feb 24, 2006, at 16:04, Michael Stillwell wrote:
On 24 Feb 2006, at 16:45, Joshua Schachter wrote:
would it be okay to write a robot that crawled a site and pumped
every URL it found into delicious, using tags collected from <meta
name="keyword">? (I thought this would be considered spam.) Has
someone already written something like this?
The tags and descriptions people use to bookmark a site are
collectively a summary of the message conveyed to them. They're the
consumer's point of view. This feedback is probably more valuable
than a site's ability to jockey for position by pushing their own
agenda into the database.
Using <meta> tags as a data source sounds good, but it fails in
practice. I've used an experimental bot to collect nearly 20,000
links from IRC channels and post them with <meta> keywords and
descriptions culled from the pages themselves, and the results are
less than spectacular.
Anyway, publishers tend to choose bad keywords. Many sites try to
position themselves in search engines by stuffing every conceivable
keyword is the <meta name="keywords"> tag. A real-world example:
god atheism skeptic skepticism agnostic agnosticism agnostics atheist
theist theism jesus savior saviour deus dieu dieux jehovah christ
scientology scientologist buddha pope life death reincarnation belief
believe believer naive christian christianity islam islamic bishop
reason genius intelligence religion religious buddhist holy goddesse
deesse divinity divinite divinities crucifixion rebirth faq magic
gods skeptics skeptical darwin evolution sacred sacrifice worship
dianetics rituals spirit trinity hindu hinduism brahma shiva krsna
krishna vishnu afterlife loss lord seigneur tantra tantric love
omnipotent omnibenevolent omniscient veda vedic trancendental
trancendence meditation rama scriptures incarnation salvation
siddharta righteous righteousness rite rites ritual priest catholic
catholics catholicism catholique eglise church quebec québec devotion
monk creation creationist creationism tao taoism qi shinto temple
rosicrucian cross renunciation virgin virginity sikh creator
thanksgiving jewish jew jews hebrew torah bible biblic gospel
hanukkah moses abraham noah adam resurrection lies hell heaven
christians crucifix baptism sacrament unconditional muhammad koran
mecca prophet allah muslim faith preach preacher preaching prayer
apostle apostles testament birth miracle circumcision communion
gospels polytheism monotheism pantheism spiritual immortality science
watchtower yahweh elohim rael gott theos yhvh yahveh dios godt ape
apes monkey monkeys messiah dio duvel devil satan demon angel angels
gud bog boje eloi mosque jainko diu dogma
Or a site will use the same boilerplate keywords and description for
every page despite the fact that the site is almost entirely user-
submitted, tagged, and described content:
Keywords: photography, digital photography, cameraphones, camera,
hobby photography, photo, photos, digital camera, compactflash,
smartmedia, cameras, canon, nikon, olympus, fujifilm
Description: $SITE_NAME is almost certainly the best online $WEB_2_0
application in the world. Show off your favorite $CONTENT to the
world, securely and privately show $CONTENT to your friends and
family, or blog the $CONTENT you take with a $DEVICE.
(The variables are my lame attempt at anonymizing the site.)
None of the above describes why you probably thought the pages were
interesting enough to bookmark. Automating publication of this sort
of data would eventually reduce del.icio.us to yet another search
engine. I think that would suck.
--
Rocco Caputo - [EMAIL PROTECTED]
_______________________________________________
discuss mailing list
[email protected]
http://lists.del.icio.us/cgi-bin/mailman/listinfo/discuss
