Dica4Tags

TableOfContents

= Why? =

after moelskiing evaluation, we focus on blog scenario

* Distributed DiCA

* centralized DiCa

= Use case =

show me all the users who use concepts that map to paolo's "social_software" concept.

show me the last X posts by users who use concepts that map to paolo's "social_software" concept.

show me all the users who use concepts that map to paolo's concepts.

show me the last X posts by users who use concepts that map to paolo's concepts.

= community =

what is the community used for?
to restrict the adv game, otherwise having a game with 6.000.000.000 peers does not scale!

who compose the community of user a?
1. list of explicitly added users (friends)

1. list of preinserted users

1. list of all the users

1. list of friends and friends of friends

1. list of suggested users (based on sintactic overlapping probably)

how to bootstrap the community?
[complete]

how to show the community?
= rendering pages =

for anonymous user
[complete]

for just logged-in user A
we still don't have meaningful information for her, since the adv game takes some time to produce results but we want to return smething anyway. probaly based on syntactic info.

we fetch user A's posts, extract some keywords, and we find users who use similar keywords. we retunr a list of last posts created by those simialr users.

[complete]

for logged user
[complete]

= rendering a random category for anonymous users =

in order to get an interesting one, use http://developers.technorati.com/wiki/TopTags

The toptags query allows you to get a list most popular post tags tracked by Technorati. On the Technorati site, the tag index page shows a heatmap of the most popular tags. The API version allows you to specify the lowest tag rank you would like to view.

maybe keep the list and take the first not yet shown in the last 24 hours or so, or the most emerging (from zero to 10 is better than from 1000 to 1200!)

= how do you participate in the game? =

* creating a categorized blog post (in del.icio.us, you participate "bookmarking an URL") * claiming a blog on thesite

= show anyway some posts from technorati about the chosen globaltab the user is browsing on =

using http://developers.technorati.com/wiki/TagQuery

= bookmarklets =

[complete]

= possible problems / critical points =

Dica4TagsPossibleProblems

= blog we will include =

Dica4Tags/BlogsToBeAggregated

= paolo wanted features: =

1. subscribe to a "semantic feed" (that is whatever is categorized by [every] user under what phauly calls "social software"). I THINK THIS IS SOMETHING IS TOTALLY MOISSING FOR NOW IN THE BLOGHOSHPERE AND EVERYONE WOULD LOVE IT!!! WE WILL GET SLASHDOTTED!!! --> put the service on a very powerful machine!!! very powerful!!!

at feedster at the moment you can subscribe to a certain texual search. for example i subscribed to the search for "trust" http://feedster.com/search.php?hl=en&ie=UTF-8&q=trust&sort=date [be sure to check this page, it is somewwy similar of what DiCa provides, they provide as well the rss feed for a query!! http://feedster.com/search.php?q=trust&sort=&ie=UTF-8&hl=&content=full&type=rss&limit=15 ]. yahoo and google will go as well into this direction [providing rss feeds of searches] (i think). but as you c an imagine i also must subscribe to the search for "reputation" and for "trust-aware" and "repuitation systems" .... I would love to subscribe to just "what i call trust and reputation"!! and millions of blogger will as well!!

1. add some discovery feature? popular users that used the term "social software" as phauly used it ... make sense?

1. show a pairwise comparison page where you can see how categories of 2 users are matched. this is really useful especially if i am one of the two.

-

how to populate the dataset?

-

claim your blog. how?

-

search blog / search user

some seraching funciontality should be provided. imagine that i log in and i want to find iuf my friend "phauly" with the blog "paoloblog.org" il already in. how can i do it?

= everything else =

convey meanings via icons? colors?

--

create a user --> a new peer is added into the language game

-- feedback? for evaluation and for

-- date?

-- provide the RSS for the "semantic" tag "social software"

-- use blogstreet for inferring blogrolls

-- click on bookmarklet and no chosen tag --> try to autodetect the tags

-- if you click on category of a blog that is not in the db --> "you have found a new player in the game, come back in 1 day!" or give back an approximate results in order not to annoy the browser.

-- firefox plugin that transforms the page adding (find semantically equivalent tags)

-- add to my sources

-- popular something? just show some popular

--- extract posts tagged under "social_software" and see all the terms that are most used ... in order to provide a sort of summary of "social_software"?

-- show only results from my community or results from everyone

--- blog registered --- blogs added by someone

--- blogs of your community

-- play only with your community --> means n games for n users

-- how to boostrap the community?

want social? "flexi.com"

want technical? "slahsdot"

want cutlure3? boing boig"

wnat news? "reuters, cnn, bcc, ..."

suggest blogs that have similar keywords to the ones you use in your blog?

-- what if my community is empty?

-- rss feeds can be: registered, added by someone else.

-- tag(changecolor)norati

-- anyone can claim to own "slashdot" --> many people can claim the same

-- i say i won slashdot and i say the network of slashdot is sex.com, porn.com

-- claim blogs in the same way as technorati.

-- if i claim i have sex.com and i claim you are in my community, do you get results from sex.com

-- my community is blogroll

-- in future: subscribe to phauly:social_software

-- other relationships for tags: more general than, less general than

-- feedback: i read a paper and say "this post is (NOT) related to my category"

-- there is room for people that do collaborative maintainance [ "Tags gardener (TM)" ]

-- do you plan to show all the users that match my concept? maybe 100000 for folksonomies!!!

- semantic rss feeds

- marco's posts for networks (the real ones!)

- global tag rss feed

clicking on right usernames is not too meaningful

match : match against all the community

click on "my label" leads to "matches table" (with a list of posts under this tag)

free text searches? for tags? if you don't have a blog (or a certain category) and want to see the mapping of "sex" into conor's categories. how can i do it?

blog posts have often more than 1 tag!!!

--

search in all community (a button that add all the users)

search for all the tags (a button that add all the tags)

users are dictionaries! ... (conor as german)

-

list of what are the pages (and what they do provide)

run tests with many users (many tags, many user in my community)

what user page?

what homepage?

what when logged?

what when not logged?

avoid + in tags!!!

username are only small letter and numbers and no strange chars!!!

xfn? foaf?

--

i didn't get the function proposed by conor at the end (italy -- italia -- belpaese)

-

-

-

remove some players from the game.

the one not very actoive?

--

game?

"blind date with the tag" --> try to guess the tag we have predicted but we are not showing. bleah, lame!

remix words --> do something creative with those words? invent a new tag that gets high adoption? show most used tags

similar ro the "word burst" page at http://www.daypop.com/burst/ that shows the "Word Bursts are heightened usage of certain words in weblogs within the last couple days. "

score points for every feedback?

"let's tag it" --> madly tag everything! quickly!

---

if 10.000 users map my concept "social_software" then the service is not useful ((too much info!))

restrict to my (small) community. how to set the limit? manually? empoirically?

-

"subscrive to equivalent" function near every local tag.

-

pplugin for wordpress

-

importing subscription oif someone on technorati should be easy and since we ask "your del.icio.us username" that is feasible.

--

add a "getfirefox" button somewhere. register on http://spreadfirefox.com/ and then use a identification link such that if many people click on it you get listed on http://spreadfirefox.com/ as a good spreader (this is additional advertisements!)