How to prevent Google Adsense to cheat clicked analysis

Filed under: AdSense — Wrote by Lees on Friday, September 28th, 2007 @ 6:17 am

As online advertising popularity, pay By Per Click (is clicked every time pay) mode is accepted by everybody gradually. But subsequently and the precautionary approach that the problem that come is Fraud Clicking is in eyebrow nimble, because this will matter to mode of this kind of advertisement directly whether live for a long time and whether the income source that becomes a kind of true website owner.

Google Adsense system introduces how to set out from systematic angle below prevent to click deceit, the hope prevents to click phonily to the online advertisement system of other can have very good direct effect:

1] hit = clicks a time / the frequency that browses in all.

The hit is the method that whether a judgement has the key that does not have Fraud Clicks, the advertising hit that can think a website to go up exceeds 10% imply this what.

#of Click/# Of Viewed

2] is clicked enclothe rate / independent IP, this distributings in if have; of individual IP (click / browse) = is clicked enclothe rate exceeded the systematic error of 3 times there will be cogged suspicion inside limits.

Example, exemple Buddha browsed 16 webpages from the user of 129.119.200.1, clicked 4 advertisement, and whole advertising hit ” from [the calculative in 1] arrives ” be 5% , so computation gets:

%5 X 16 =~ 1, variance is Sqrt(1) = 1, click enclothe rate =4/1 = 4, distributing according to the gauss on maths, this kind of probability is less than extremely one of.

Ratio VS IP Distribution

3] hit ” click enclothe rate ” / IP/ time

Undertake an analysis to the hit according to time alignment, if be in certain time paragraph on have overt peak value, so the likelihood that this will think to potential deceit is clicked.

Ratio VS Time

The time of 4] webpage Load and advertisement click the analysis with wrong time, and the analysis that twice every time needs alignment between Click

[The time of webpage Load and advertisement click time to differ] should be distributinging Possion of a Poisson

Distribution, and every two the time difference between Click also should be a Possion

Distribution, if this time is written down with the second, the word that is more than 25 seconds basically presents the figure that gauss distributings.

[Time Of Loading - Time Of Click] Distribution VS Possion

[Time Difference Of Two Clicks] Distribution VS Possion/Gaussion

5] is aimed at the analysis that Proxy clicks

Changing IP to undertake clicking can saying to be in is before most solve hard most discover cogged way hard, proxy was used to have the method that clicks phonily with respect to most when compatriots undertakes the Boost of Alexa probably, whether be but here should carry the source of retrorse inspect IP only,to contained the server of Proxy function to be able to know.

Reverse Proxy Check

6] is aimed at the analysis of Http_agent

The analysis of the time alignment of Http_agent/ time, peak value exceeds 3 variance to need to examine

7] is aimed at the analysis of Http_referral

The analysis of the time alignment of Referral/ time, peak value exceeds 3 variance to need to examine

There still is a very useful amount on 8] whole effect:

Reveal cost 1000 times all be worth / independent IP

This aux will be able to is more direct the moving computer that discovers Spam Clicking and give ban.

Overall Ratio VS IP

Even if I gave out here of above avoid cogged measure, but did not forget:

What wrong person comes than just person forever is much, those who come is violent.

  -

No comments yet. Be the first to comment this post.

Leave your comment

Copyright © 2007 Google Adsense College.
Powered by GoogleSchool. All Rights Reserved.