As reported by the media, AOL reSearch was kind enough to publish some datasets from their internal AOL-searchengine. Seeing how the terms of use only allow the usage for scientific research, naturally we have not used them but just performed general evaluations.

By now, AOL has publicly apologized for the publishing of the searchqueries. While AOL is still running internal investigations, the New York Times is reporting that the first AOL-user has been identified through her searchqueries. AOL reSearch is not offering a download of the data through their wiki anymore but mirrors are still offering them.

All in all, they protocoled 36.389.577 searchqueries of us-American origin in the time from march 1st to 31st. For this they recorded anonymized user-ID, the query, time of the search, the selected URL as well as the position of the selected URL. The 657,426 unique users put in 21.011.340 unique searchqueries and clicked on 19.343.540 URLs. The files have an unpacked size of 2175,76 megabytes.

Clickrate by Rankingposition



It is interesting so see that the clickrate goes up a little for the last spot on the first page (#10) and then, on the second page, declines sharply. The large differences between position 1 and 3 are also remarkable.

Most clicked on URLs



By request here the evaluation of word-count per searchquery. It seems that the trend for the 2.- or 3.-word has not reached AOL-users yet.

Google has a large lead over MySpace. Only then do we get Yahoo, Wikipedia and Amazon. Having Bank of American on position 10 can be explained by the fact that AOL-users are often searching for URLs and will then click on the first result. The label “Army of Lamer” still seems to hold ...

Time of Search



Increasing from noon to 9 pm and decreasing again during the night. This should be familiar to most from the access-statistics.

Most-searched-for Keywords



The big players (Google, Yahoo,MySpace, Mapquest) are represented in the top15. It is notable how often domain names are searched for – people are even searching more often for MySpace.com than for MySpace.

Words per Search



By request the evaluation of words per searchquery. The trend that was reported a few weeks ago about the 2.- or 3.-word search has not reached the common AOL-user.