Google Blogoscoped

Forum

What We Learn From Google Trends...  (View post)

Andrew Hitchcock [PersonRank 10]

Thursday, May 11, 2006
18 years ago5,097 views

Damn, I tried opening all of Steve Rubel's links in new tabs, and now I'm getting forbidden messages because they think I am spamming them. Grr.

Wouter Schut [PersonRank 10]

18 years ago #

me too, very frustrating. They do not even let me prove that i am human (with a captcha thingie).

/pd [PersonRank 10]

18 years ago #

I think thats trends is a very powerful tool ..!!

Philipp Lenssen [PersonRank 10]

18 years ago #

Hmm, so I guess we shouldn't even attempt screenscraping data from Trends... something like a dictionary search...

/pd [PersonRank 10]

18 years ago #

Why screen scrap......just right mouse click and you should be able to save trendgraph as .png format

elyk [PersonRank 6]

18 years ago #

I wonder how long until they have a page showing how often people are searching for trends for certaibn terms. And then they could have a trends page for that feature. And so on, repeating infinitely

Jay Amin [PersonRank 10]

18 years ago #

Lol, elyk, that is similar to when once I was pondering about how a rate increases by a rate that incrases at a rate that increases........ :)

Caleb E [PersonRank 10]

18 years ago #

/pd, philipp means a screen *scrape* (not screenshot)
http://en.wikipedia.org/wiki/Screen_scraping

Philipp Lenssen [PersonRank 10]

18 years ago #

Yes, for example in an English dictionary there can be 27,000 words... and that's even a lot for automated queries.

Tony Ruscoe [PersonRank 10]

18 years ago #

And since there are no numbers being shown anywhere, it wouldn't be much use screenscraping the data... would it?

Philipp Lenssen [PersonRank 10]

18 years ago #

Hmm, you could use something like PHP Image to grab pixel values... you know, perform a search for
the, abba
the, apostle
the, abramah
the, abulafia
...

and then calculate the distance between "the" and the bottom line, which should get smaller the more people search for "abba" etc. OK, I guess it wouldn't work because there's only so many pixels in a PNG...

Philipp Lenssen [PersonRank 10]

18 years ago #

Correction: "the" would pretty much always be higher, not lower then the graph for the other words. So you'd have to measure the distance between "the" and the other word, I guess...

Philipp Lenssen [PersonRank 10]

18 years ago #

But it doesn't sound very feasible! I guess Centuryshare is a better algorithm for that, and it also goes back to any year in history...
http://blogoscoped.com/archive/2003_06_11_index.html

Forum home

Advertisement

 
Blog  |  Forum     more >> Archive | Feed | Google's blogs | About
Advertisement

 

This site unofficially covers Google™ and more with some rights reserved. Join our forum!