idn.bz

My name is Jose Augusto and I have been doing IDNs since Christmas 2005. Most IDNers know me from IDNForums or dnlocal, also as Lidador. I have more than one thousand IDN names and keeping up with what each one means and is worth (it changes over time) is a very time consuming and tiring job.

With my main renew season arriving and being a script dude, I thought it was the right time to bring to light this idea I had all along of putting together a script that would allow me to just look at a score and decide whether or not to renew an IDN domain. I also had the domain idn.bz doing nothing so I choose to spend 3 days working on a script instead of 3 days looking into a long list of which domains to renew.

The result is idn.bz: a complex algorithm to automatically evaluate and rate a term used on an IDN domain!

How to use it is really simple: just enter the term on the box, choose the language and hit the button! You can enter full domains like "crédito.com" or a multiple keyword term, like "energias renováveis" (be sure to enter multiple words separated with spaces - imagine what it would take for the script to decide where a word ends and another begins...). You can also enter keywords in punnycode format.

The algorithm takes into account more than 10 factors and correlates them together.

• - The language
• - The number of search results
• - The Google Trends
• - The availability of the term on the main domain extensions
• - The translation to English
• - The reverse translation back to the original language (to find premium terms)
• - The size of the domain
• - The Adwords ads
• - Some other minor factors
• - The interdependencies between all the above
• - And, finally, the secret source, which heights more than 25% it's where the real "subjective" rating of a term is made

And the script does much more like, for instance, detecting the Arabic "the" or considering the Spanish ñ!

Two are. The minimum # search results a premium term should have (in order to be considered premium - this varies depending on the language) and the language factor.

Current values are, for the :

• PT: 4.000.000
• ES: 6.000.000
• FR: 5.000.000
• DE: 5.000.000
• RU: 6.000.000
• JP: 8.000.000
• CN: 5.000.000
• AR: 2.000.000
• KO: 5.000.000
• HE: 1.000.000

• PT: 6
• ES: 8
• FR: 7
• DE: 10
• RU: 7
• JP: 10
• CN: 9
• AR: 8
• KO: 9
• HE: 5

It a subjective figure that takes into account two things: number of people speaking that language x IDN sales trends. For instance, there's no doubt Portuguese is a much more spoken language, but PT domains sale very badly.

After many months of fine tuning the rating algorithm it was time to move to the next logical step: calculating the estimated value of a domain.

As most of the veteran idners now agree the rating algorithm produces reasonably trustful rates, all that was needed was to put together the idn.bz rates with the sales value of all the idn domains we could find. On doing so, we were able to build a prediction matrix that would output the value of an idn domain given its rate.

The equation that estimates the value is a+bx, where:

and:

and where x and y are the sample means AVERAGE(known_x's) and AVERAGE(known y's). The x are our rates and the y are the sales figures.

As you can easily notice, it's a simple linear regression equation.

Of course, the more sales figures the better, so if have been involved or know of some idn sales, please get in touch with us and send in those numbers. It would make the prediction model much better.

The ideal format for sending the data is: pynnucode,lang,value,year, on a .txt file, one each line. For instance: xn--u9j040kliq,ja,25,2008

We use only data from sales on the following extensions:

• Portuguese, French, Russian, Greek, Arabic, Chinese Traditional and Korean: .com only.
• Spanish: .com and .es
• German: .com and .de
• Japanese: .com and .jp
• Chinese Simplified: .com and .cn

Yes, here , the official forum thread for this tool / Yes, please do!

Or else, here at IDNForums.com

Yes, algorithm current issues are:

• - Not accurate for Geo domains: will never work... anyone has a list of all world cities and countries, including population?!
• - Not accurate for Latin plurals: working on it...
• - Not accurate for Russian/Greek capital letters: SOLVED !
• - Not working on Firefox: SOLVED !

Also, keep in mind most of the times the wrong figures are due to Google inaccurately translations and stuff, and thus, results will get better in time even thought the algorithm stays the same, as Google tools are improving each day.

The top 10 listings are dynamic, as they are assembled in real time from the top 10 best rates stored on the database. These lists change everytime someone searches for a term that tops the ones already on the list. There is one Top Ten list for each language.

As this is a dynamic tool, some terms might be trendy on certain seasons, for instance winter sports domains might not get ads during summer and thus be valuated less than during winter. But, hey, it makes sense right?! Another example are terms who got more popular, languages that grown it online presence, and more sudden jumps might also noticed when a term gets a more accurate translation on Google translator. BTW, if you think your term is wrongly translated and you know better than Google, you should use the link on the Google translate page to suggest a translation. You would be doing everyone a favor. :)

Hey, in case you haven't noticed yet, we measure almost everything online by Google standards! The algorithm depends 57% (exactly) on Google and, as said, it will get better as Google translation, trends and search results improve.

• お金 (Money - Japanese) Score: 99.76 (highest score I could gett so far...)
• سفر (travel - Arabic) Score: 97.82

Remember those ascii look-alike "musìc"?! lol

Version 1.0 (Early 2008)

• Alpha one :)

• - Three new languages added: Chinese Traditional, Greek and Hebrew
• - idnclub forum added to forum buzz frame
• - Chinese IDNer bug fixed
• - Language factors adjusted according to user inputs

• - Fully compatible with all browsers (I got rid of Ajax)
• - Input box is bigger
• - Input box on all pages
• - Input box is cleaner
• - It retains the used language selected
• - You can use a direct link on your blog post / forum sale thread.
• - Some minor bugs fixed

• -Fixed: Google translate results are back
• -Fixed: Wikipedia removed from chinese terms due to jumping out of the iframe. Link remains.
• -Fixed: Argorithm changed due to the way google translates Arabic. Google returns "term" for "the term" on Arabic.
• -Added: A link to arabic results without the "the".
• -Added: The detection 'い' '・' 'の' '的' on Asian languages. These can ruin a domain value.
• -Added: IDNF auctions listing.

• -Fixed: Google trends fetching moved to another server due to Google overload ban.
• -Fixed: Algorithm tweaked due to Google geo showing ads only to targeted countries. And thus not showing them to our servers (based on USA)
• -Added: The estimated value of a term [dot] com. Read complete info above on this FAQ.
• -Added: A database. Now all queries and resultant values are stored on a database for future data mining. The database makes possible the new dynamic top listings (more below) and the calculus of the estimated value of a domain name, as this is based on previous sales.
• -Added: Cache. Now all queries get cached for a certain amount of days. For now it’s 10 days. The effect is blazing fast loading speed of popular queries. (And giving Google a rest)
• -Added: Dynamic Top Ten Lists. These listings are now assembled in real time from the top 10 results stored on the database. There is now one Top Ten list for each language.
• -Added: Naver and Baidu links to Korean and Chinese results; Yandex and Rambler to Russian; and keywordadvisetoolplus to Japanese domains queries.
• -Added: A Paypal donate button: $5. ;)

• -Fixed: Naver, Baidu, Yandex and Rambler link utf encoding. (Thanks Bramiozo)
• -Fixed: Downtime: moving away from Dreamhost!
• -Added: 4 new languages; Hindi, Thai, Turkish and Ukrainian.
• -Added: direct link to register available idns at once.
• -Added: clickbank products pub

• -Added: Russian transliterations
• -Added: 1 new languages: Vietnamese
• -Added: Introduced Google Local variable on Google queries (&gl=)
• -Added: Algorithm now counts ads from Google's local pages as shown to local users
• -Added: Result page now shows Google page frame exactly like the country local user would see it

• -Fixed: Dramatic speed improvement (on first time term checks)
• -Fixed: Naver links fixed (again). They are back to UTF-8 from EUC-KR
• -Added: Days to expire on non available domains (will show in red if domain has expired!)
• -Added: Now shows other extensions availability: .org .info .ws .tv and .cc (this uses Ajax and does not slow down the page load, as it will only show up when finished)
• -Added: idn.cctld checks! Available only for Portugal, Germany, Japan, Korea, Turkey and Twain (Chinese Traditional) cctld domains.

• -Added: Two new languages: Danish(Dansk) and Swedish(Svenska).

• -Fixed: .ws availability corrected. website.ws bans for maximum allowable queries per day. Muted when results are not available.
• -Added: Babelfish Translations. Available for chinese,chinese-traditional,french, german,greek,italian,japanese,korean,portuguese,russian and spanish.
• -Added: Variant double checking. In case of too good to be true domain availabily results, the script does a variant checking (slower) on the background to ensure the domain is really free. Next time you refresh, result will be accurately updated.
• -Added: Now shows since when the domains have been registered. (the year they were first registered)
• -Added: .biz extension checking.
• -Added: One new languages: Italian(Italiano).

• -Added: Real time expected traffic and CPC values from 102 countries! (Country only shows when there are results available)
• -Added: Yahoo.com search results.
• -Added: Bing translator. Available for arabic,chinese,chinese-traditional,french,german,greek,italian,japanese,korean,portuguese,russian and spanish.
• -Added: Three new languages: Farsi, Hungarian and Czech.
• -Fixed: Menu changed to English and sorted from A to Z.

• -Added: Virtual keyboard: available in Russian, Hindi, Thai, Arabic, Persian, Greek and Hebrew.
• -Added: Auto detecting of language: when manually selected OR if you leave the url lang= blank.
• -Added: Twitter search results: on Twitter you don't conclude a term is popular by the number of search results, but instead by how long ago that term was last used. If you see "about minutes ago", you can be sure it’s popular. Also don't forget to click the "Translate to English" button inside the Twitter frame, for better scrutiny on a term.
• -Added: Four new languages: Norwegian, Dutch, Polish, Bulgarian. Norwegian, Polish and Bulgarian with IDN.ccTLD check!
• -Added: Detailed translation link, as Google translation now has dictionary like results for some translations.
• -Added: Name.com IDN suggestions link.
• -Fixed: Greek results. It was using the wrong language tag.
• -Fixed: Russian results. It was failing to convert some results to lowercase and thus the comparison would fail. I had to code a replacement function to PHP’s 5 default strtolower, which fails on utf-8.

Just do a search and copy the url on the address bar or use the following format: http://idn.bz/rate.php?keyword=KEYWORD&language=LANGUAGE

That's the idea, or else what good would ratings be if no one could check them out? There is, however, one thing that is I wanted to factor in Adwords figures using Google’s API calls. However, each Adwords API call is paid! I am open to suggestions on how to support those payments...