2. Google News
After all the negative buzz on Google Buzz, Google had made some changes to policy. Read more here . Another interesting post was from Amit Singhal titled This Week In Search . The most interesting news in that post was that , all your public buzzes are being included in real time search. Not unexpected but still important to know.
This was a demo in TED 2010 conference by Blaise Aguera y Arcas who also previously developed PhotoSynth . In this talk he talks about some of augmented reality features being introduced into Bing Maps. Frankly, the demo could have been made more awesome but it still blows you away with its potential and its background technology. Awesome job guys !
Tom Toles is one of my favorite cartoonist . I did hesitate to add this cartoon to the post. But finally, I added it as this cartoon might be one of his best so far this year.
5. Will My Paper Be Accepted at FOCS?
Another nice post from Dick Lipton. One of the questions he raised was can we write a machine learning program which can give the probability that a particular paper will be included in FOCS. (FOCS is one of the premier conferences in TCS). I am sure he was not very serious but still I think this could make a very fun project. Since I am interested in ML, may be I should try it out in spring break. (May be I can try applying it to SIGMOD/VLDB articles where I have better understanding).
I spent some time thinking about the features needed. It needs to handle temporal changes in valuation which usually can be done by some decay factors. Probably for best results, we need to split the input papers into different classes (Complexity, AGT etc) which can be done using a Naive Bayes classifier. I think for best results, we need to compare papers with the old papers in same field. So we might need multiple classifiers, one for each major field.
Authors, Coauthors, citation counts of cited papers will be important non content oriented features. If some human can rate the importance of the open problem the paper solves we can add it to the score too. Understanding the significance of a problem is not very easy to do by a machine. After this , things become murky. Verifying author’s claims or proofs is all but impossible using a program. I am not sure what all features makes sense. If you find some useful things do chime in, or you can add it to Prof.Lipton’s blog comments.
Have a nice week !