@gklyne @wsREST Maybe you get inspired once you see our CfP :-) We can’t do much about the travel budget, besides maybe VC you in…
@IgorBrigadir Guess why our paper is called “MJ no more…”arxiv.org/abs/1303.4702aD). Full text search for “shoulders” :-@milesosbornene
RT @mafintosh: My slides for my talk are available here, https://t.co/f2c9og967l - the torrent-docker stuff starts here, https://t.co/rzEYf…
@edsu Right. For the time being, one can <del>at best</del><ins>simply</ins> roll ones own instance of Wikipedia Live Monitor ;-)
@IgorBrigadir @edsu Fair enough, but no history (no std. deviation) for “just now” created breaking news articles… But works for many cases.
@IgorBrigadir @karpathy @edsu LOL, mostly this. Initially required >1 languages, but now ≥1. Also had ≥2 concurrent editors, now ≥3.
RT @IgorBrigadir: RT @karpathy “We set α = 0.1 because that’s the only number we had time to try and it seemed to work okay #honestpapers” …
@edsu For text-only: https://t.co/FiewfKYFTc. For media galleries: https://t.co/RssL1LfBmj. Hi callback hell ;-)
@edsu Fully agree. Human in the loop makes total sense, also common “stop hashtags”, like stop words now. Internationally for all languages.
@edsu (In no way saying current settings’re perfect.) Long-term, article analysis needed & better understanding of article creation process.
@edsu Getting these thresholds “right” is incredibly hard. Also very much a matter of taste and how “breaking” you require the news to be.
@edsu Yes. Also surprised, but breaking news conditions (see paper or settings in app) weren’t fulfilled. Mostly ≥3 concurrent editors one.
RT @apassant: Sharing @YouTube #music on @Twitter: Analytics using @fbase + #BigQuery
http://t.co/B9V3Ela8Zh /cc @felipehoffa @akowal @wjar…
@netlabsorg @RubenVerborgh Ironing out some issues later today w/ Ruben, but watch this space for updates ;-)
@edsu Probably a lot of noise, think of hashtags like #RIP. Worth playing around w/, though, interesting idea! Looking forward to code :-)
@remagio I can offer http://t.co/u19tbu96L3 and http://t.co/H0md0UvUaD. Somewhat related http://t.co/Ko1acRvJQc. Cheers.
@edsu Possible: let @WikiLiveMon detect breaking news, extract hashtags from rel. microposts (based on article titles), follow $TWTR stream.
@lbjay @edsu People who like @WikiLiveMon also tend to like @mediagalleries ;-) First described here arxiv.org/abs/1403.4289.
@remagio @edsu Sure, you can simply adapt https://t.co/nhj2oivo11 to make @WikiLiveMon cover only part of the Wikiverse.
@edsu @lbjay You could adapt the “email” function (link last tweet) & extract hashtags from found microposts; then archive to DB (or email).
@lbjay @edsu Not sure I fully follow, but @WikiLiveMon already lets you archive posts related to detected events: https://t.co/VWsixnH6dD
RT @IAugenstein: New journal paper by @LeonDerczynski et al. “Analysis of named entity recognition and linking for tweets” http://t.co/W1FC…
Spent the night working on a <polymer-ldf-client> element—Linked Data Fragments client as a Web Component. (CC@RubenVerborghgh)