Sunday, January 29, 2012

Automatically Finding Danish Typosqutters

Typosquatters are people that buy domains with a name close to an existing domain. Usually the domain they squat is popular and therefore has a high risk of people typing the wrong URL in their browser. Example: is a typosquat domain of

The problem with typosquatters is that they are hurting the domain they squat, by hijacking users. Many typosquat domains also contain malware, advertisements and are used to do black hat SEO and otherwise hurt the user experience on the Internet. In this blog post I'm going to create an automated procedure of reliably finding typosquatters of .dk domains, and at the same time present the data found in the process of creating this procedure.

The Process
1. Finding Popular Domains
To begin with, I created a list of popular domain names ending in .dk. I used Alexa to get the list of domains as they have filtering by country.

2. Filtering
Alexa provided not only .dk domains, but also .com domains. I removed all the .com domains, even though many of them have .dk counterparts. I also removed all domains that had a length of 4 or less as they would produce a lot of false-positives in the next step.

3. Finding Typosquat Domains
I used the Damerau-Levenshtein distance algorithm to find the number of edits required between two domains. Having small domain names gave me a problem of too many false-positives. would have a distance of 1 to, and is not a typosquat domain.

A lot of the domains timed out or 404
Next I made a list of 862.000 active danish domains by data-mining 4 different data-sources. The reason I had to data-mine instead of getting a complete list from the dk TLD, is that dk-hostmaster does no longer provide the public with a full list of dk domains. According to their statistics page, they have 1.156.476 domains, so the list I created is by no means complete.

4. Marking The Positives
Manually checking the results reviled that a lot of typosquat domains looked the same and had the same layout. I saved the sites in a collection of their own, and if the distance to those sites was below 20, I marked the site as a typosquat domain.

5. Marking The Negatives
I downloaded the HTML for all 100 popular sites and compared the results of all potential typosquat domains to that HTML. If the distance was 50 or below, I marked the page as an alias (The owner of the popular site typosquat his own domain to prevent others from doing it).

The Results
A total of 746 potential typosquat domains were found by comparing the 100 popular sites to the 860.000 danish domains. Out of those, 138 confirmed typosquat domains were found and 287 aliases were found.

 The following domain usernames were found i the process:

DL4695-DK JT6619-DK HHL113-DK WA3375-DK
DB7060-DK IM2838-DK KB11475-DK UIG3-DK
DA13215-DK DB7120-DK DS11319-DK YX7-DK
EOA89-DK MM18535-DK AAI56-DK HX23-DK
LX17-DK FMI19-DK MK20537-DK NP1743-DK
DA12540-DK JL10879-DK DL4843-DK IDF3-DK
BBL46-DK IAL33-DK PT3833-DK T13205-DK
VD1208-DK MM19124-DK JM11900-DK EL2826-DK
S12017-DK NL1786-DK ELJM1-DK MM19927-DK
AS20785-DK XQ1-DK MK17141-DK EH4651-DK
XA441-DK XA440-DK EN1576-DK

Each username can be looked up here. Doing so will result in 3531 domains with a high concentration of typosquat domains.

The Case Of Sedo
About 90% of the domains are owned by Sedo, a domain hosting company that advertises with the ability of giving your site more traffic. They do so by linking the typosquat domains to your site.
Not only does Sedo use typosquat domains, they also buy popular domains that have been shut down or have expired.

Their tactics are relentless, they even use designs of popular sites like Facebook as a way of tricking users. They also use different designs with content related text on their typosquat domains to lure people into clicking on their links.

Typosquat site owned by Sedo
A lot of the sites that are on the typosquat list created by dk-hostmaster are the very same domains I found in my analysis. Even tho they have closed a lot of sites belonging to the same username (DS11319-DK), they do not close the account or otherwise punish the the typosquatters.

No comments:

Post a Comment