Multi-lingual Keyword Stemming and Link Relevance

[pathikbd]

Keyword stemming and link relevance has made life easy on the search engine side.

However, keyword stemming technique is being predominated by English. But, what about other languages? What about introducing international keyword stemming that takes all the languages in concern?

If a site has replicated webpages in other languages (like Arabic, Hebrew, Chinese, Japanes, Korean, and all other languages that doesn't use roman/latin alphabet) will it be punished for not following the link relevance technique?

People would want to get traffic from non-English speakers also. What could be done or is it already been done?

Asif Anwar: eMarketing Consultant in Bangladesh

[brian.mark]

According to Mikkel at the last SES, most of the internationalized versions of search engines have technology that is far behind. The engineers in Mountain View and San Jose as well as Redmond haven't seemed to understand languages other than English real well, so many spam techniques don't get penalized in the international versions and advanced technologies like stemming haven't made their way over yet.

If you can get listed in those regional engines, the game is a little easier - just do everything you've ever been told not to do and you'll rank #1. (Just kidding).

Seriously, since search engines are run by American companies, the other language versions of those engines are far behind in technology. There is just a language barrier right now, but they're starting to hire the right people to address some of that.

Brian.

[pathikbd]

Thank you Brian for the update.

The problem is we people in Bangladesh think that we are always staying behind. But, lets pray that, we would have that technolgy working for future.

Actualy, keeping that in mind, I am making my site future proof. Do you or anyone think its worth it?

[brian.mark]

Of course that's worth it. If you don't work on what's coming up (you can see that easier than those of us in the US since the US version is ahead of your localized version), then you'll be fully prepared when those changes do occur. When that happens, those that weren't prepared will be playing catch up and you'll look like a genius.

On that note, if you know of any online retailers that aren't utizing froogle to the full extent, then they too will be playing catch up at some point in the future according to Danny Sullivan. Personally, I hope he's right because our sites would be doing just dandy.

Brian.

[pathikbd]

Dear Brain

Starting from Alta-vista to Google, almost all have their translation plugins. Why not put that into their engines to recalculate the algorithm?

However, since my pages are single and in diffrent languages, like: French, German, Italian, Spanish, Chinese (Simplified), Korean, Japanese (Russian, Portugese, Turkish, Arabic, Pesian, Bahasa - Planning to add); will it influence my link relevance or outbound linking strategy?

If it does, then I may have to take different inbound linking strategy for different languages.

[brian.mark]

Originally Posted by pathikbd

Starting from Alta-vista to Google, almost all have their translation plugins. Why not put that into their engines to recalculate the algorithm?

The real problem is that yes, they can translate the page, but that doesn't mean it's done well. See http://www.webproworld.com/viewtopic.php?t=50499 for what I mean. The example I gave there is as follows:

"I'm translating this text from English to Spanish and back again."
=> "Estoy traduciendo este texto de inglés al español y a la parte posteriora otra vez."
=> "I am translating this text of English to the Spanish and the posteriora part again."

That's about as easy of a sentence as I could think of to try that with. If you saw that second sentence, would you understand what it meant? Probably not, and because they can't properly handle straight translation they also can't handle stemming in other languages very well. It'd be best if you use all variations that you want to rank for on your pages and in your backlinks.

Even in English (their primary focus, right?) the search for singular and plural returns different results in most cases. That's the simplest form of stemming and it produces different results, so I don't expect more complex variations to be handled very well any time soon.

Brian.