Microsoft's Answer To Google's PageRank Algorithm: Less Privacy?

Google
أرسل من قبل BlackSigma في السبت, 2008/07/26 - 1:38pm.

BrowseRank, Microsoft claims, can deliver better search results by measuring user behavior, including where users go and the amount of time they remain at those pages.
Microsoft researchers have published a paper describing a new method for determining which Web pages are the most relevant for a given keyword search query.

Google relies in part on its PageRank algorithm to determine what's relevant. Microsoft's answer to PageRank is BrowseRank. 

The academic paper, "BrowseRank: Letting Web Users Vote for Page Importance," was co-authored by Microsoft researchers Bin Gao, Tie-Yan Liu, and Hang Li; Zhiming Ma of the Chinese Academy of Sciences; Yuting Liu from Beijing Jiaotong University; Shuyuan He from Peking University; and Nankai University's Ying Zhang.

PageRank treats Web links as votes for relevance. The paper's authors say PageRank can generate inaccurate results "because links can be easily added and deleted by Web content creators."

BrowseRank, Microsoft claims, can deliver better search results by measuring user behavior: the Web pages Internet users visit and the amount of time they remain at those pages.

"The more visits of the page made by the users and the longer time periods spent by the users on the page, the more likely the page is important," the paper states. "With this graph, we can leverage hundreds of millions of users' implicit voting on page importance."

The paper mentions that the data set used to test BrowseRank was cleansed to protect user privacy. "All possible privacy information was rigorously filtered out and the data was sampled and cleaned to remove bias as much as possible," it says.

And that's the only mention of privacy in the document.

But privacy almost certainly would be an issue for Microsoft, if it tried to implement this technique without some means of anonymization. Keeping records of every Web site that every Internet user visits, as a way to determine relevant search results, would have huge privacy implications

the source .....


( )

خيارات عرض التعليقات

اختر طريقتك المفضلة لعرض التعليقات و اضغط "حفظ الإعدادات" لتفعيل تغييراتك.
صورة foaad
أرسل من قبل foaad في الأحد, 2008/07/27 - 3:55am.

Great algorithm, one problem though...
No one has this data, and no one will!!


صورة BlackSigma
أرسل من قبل BlackSigma في الإثنين, 2008/07/28 - 1:47am.

كتب foaad:
No one has this data, and no one will!!
off course no one already has this data it will be built step by step
but why NO ONE WILL ?????


صورة foaad
أرسل من قبل foaad في الإثنين, 2008/07/28 - 4:48am.

كتب BlackSigma:
but why NO ONE WILL ?????
How could you collect this data?
You need to install a monitoring software
either in all Internet website or in all Internet users machines (the browser itself could do this job).
Which one of the above two options you think is possible?!
I for once would not use a browser that sends away information about what sites I visit and how much time I spend in each of them, if not for privacy reasons for bandwidth reasons.


صورة BlackSigma
أرسل من قبل BlackSigma في الإثنين, 2008/07/28 - 8:05pm.

كتب foaad:
Which one of the above two options you think is possible?!
actually the second one Smile
as long as collecting such information can be done through cookies it wont be that difficult
toolbar + cookies and that is or just a small plug ins
but in term of privacy , here the problems begin