How does Google Scholar work? Is there some algorithm guessing whether some site contains scientific papers or is there some manually kept list of sources? Or some other way?
(By the way, if your employer makes a fuss because of search results for a innocent search term your employer is the problem, not missing NSFW warnings.)
As of right now, that's still the top hit if you search for "iphone gender". The best part is that not only does it somehow think it's a scientific paper, it thinks that the author is "F. iPhone Porn" and the journal is "Lea".
I was just wondering about how Google Scholar works the other day. It does seem like they can't just have a manual list of sources, because in many cases, it picks up papers from the websites of their authors, not just from large sources like the ACM or the NIPS proceedings or whatever. But even if you are able to pick out research papers when you crawl the whole web, and get author/title/citation information out of them, how do you rank them? In particular, they have citation links between papers -- but presumably any paper can only cite to older papers, so papers form a DAG, where it seems like PageRank shouldn't really work (because the most 'authoritative' papers would end up being old ones).
PageRank already has some regularization (essentially epsilon of outgoing links to everything). Also, papers don't form a DAG because people frequently forward-cite or cross-cite (because authors and their friends often have relevant work in various states of preparation/review).
I'd say it's lacking in originality and doesn't contribute anything particularly new to the field. I doubt it will attract many citations. No wonder they couldn't get it published in a decent journal.
11 comments
[ 2.4 ms ] story [ 19.6 ms ] thread(By the way, if your employer makes a fuss because of search results for a innocent search term your employer is the problem, not missing NSFW warnings.)
Not very well, apparently!
As of right now, that's still the top hit if you search for "iphone gender". The best part is that not only does it somehow think it's a scientific paper, it thinks that the author is "F. iPhone Porn" and the journal is "Lea".
(Narrowly avoided: a "seminal contribution" pun)