- posted from ...
- TechmemeFH (10559)
- malburns (3148)
- arstechnica (2996)
- nytimes (2695)
- bloghud (2453)
- mashable (2261)
- twittes (2235)
- gigaom (1740)
- venturebeat (1594)
- fluther (1483)
- weltonline_top (1336)
- diggdiggerton (1331)
- davewiner (1292)
- golem_de (1244)
- popurls (1190)
- newsticker (1189)
- guardiantech (1179)
- bild_sport (1173)
- kleverson (1164)
- bbcnews (1071)
- Pistachio (1045)
- tweetmeme (1040)
- exteenrecent (1035)
- Yedda (1010)
- kristopher (999)
- ObamaNews (975)
- timesonline (945)
- waynesutton (925)
- jowyang (913)
- StanChu (888)
- TechCrunch (883)
- erku (865)
- guykawasaki (798)
- headlinenews (753)
- pkedrosky (727)
- andybeal (724)
- dsonline (709)
- JasonCalacanis (688)
- bbcworld (674)
- oregonian (673)
- desinformado (672)
- laughingsquid (667)
- bbcengland (661)
- digg (644)
- loiclemeur (642)
- WeSmirch (626)
- acarvin (616)
- hoosgot (609)
- zetrax (604)
- chrispirillo (600)
How does it work?
First, the Linkr is collecting "friend timelines" of some english speaking twitter users. Then we parse all the tweets, looking for links in it. If there is a link, we collect the tweet and number it. For tweets with tinyurls we also reverse the URL so that you can see in your browser where the tweet realy is linking.
Most popular
This feature we realy love. TwitterLinkr is looking for all words in the collected tweets. If the word is appearing in 8 or more tweets within the last 24 hours, it's displayed on the most popular page. If it's appearing in 10 or more tweets, the word is displayed in the navigation bar of the page. There's only a little filter for stop words we don't like to display, because "this", "that" and some more are realy don't worth being displayed.
Most active users
There is also a page for the 1000 most active users. That's fun, nothing more. Btu I like to look aroung here. Funny names.
New: Recommendation Links
As you can see, some of the newer links are displayed in bold format. That's the little recommendation tool we realy love. We count all the clickthroughts, saved links and links which are showing up in more than one tweet. And voila, they are recommended. By you for all.
RSS
Yes, there's RSS. For the whole site, for every user and for every search term.
Everything: http://www.twitterlinkr.com/rss.xml
Some user: http://www.twitterlinkr.com/rss.xml?nutzer=netzpolitik
Some search term: http://www.twitterlinkr.com/rss.xml?suche=launch
Transperency
We realy like a high degree of transperency. So we don't store a lot of data, only tha data we can collect from Twitter and the user clickthroughs we need for recommendation. And yes, we like to see some statistics, so we use Google Analytics. That's it. We don't store IPs or anything.
Some geek things
Most of the page is built with PHP, and some Ajax-Calls for the recommendation tool (it's all done with prototype and scriptacolous). The data is stored in a MySQL-database. Some caching is done with plain text files. And that's it.
for all the law enforcement guys
In Germany (where the site was built) you are forced to have a thing called "Impressum". Here it is.
Verantwortlich im Sinne des Telemediengesetzes...
Peter Schink
Schwedter Str. 51
10435 Berlin
