Take the 2-minute tour ×
German Language Stack Exchange is a question and answer site for speakers of German wanting to discuss the finer points of the language and translation. It's 100% free, no registration required.

Question: Does anyone have any advice, or know of websites or other resources that could inform me, on how Germans do web searches in their language?

Background: We have a search engine for large corporations that helps their employees buy items with the best discounts, and tweaking our search engine for English has resulted in complaints from German users (who where never really happy with it anyhow). We know what language a user speaks, so we can optimize searches differently based on the language used (and can furthermore sometimes detect the language typed in by looking in various dictionaries), but first we need to understand the German language, how it is put together, and what the users will expect us to do when they run searches.

Specifically, one of the big problems we think we are having is with compound words. When we started out, in most languages at least, we would match any word containing the word the user asked us to search in. This became a problem when carpet showed up in searches for car or pet, so we refined our search so it would just show plurals and stuff. That works in English because we have fewer compound words. It doesn't seem to work in German as well. We also don't know what else we are missing.

About me: I took three years of German in high school (20+ years ago), so I understand the basic grammar of the language and how it changes endings and forms compound words, but I can't really read it anymore (I adopted a child from China, so I've been learning Chinese instead), and now know just enough German to know that I don't know anything about German.

Danke.

P.S. I'm not sure what tags this belongs in, so please fix if incorrect.

share|improve this question
    
Using Google’s autocomplete with random German words comes to mind (see also XKCD). Apart from that, I can only guess that word composition and special characters may be a problem. –  Wrzlprmft Jun 5 at 23:21
1  
Why don't you just listen to your users? They should know best what they want and why. Contact a few, find out what they were trying to do, how they tried to do it (look at your logs, too), and why it didn't work with your search engine optimization. –  dirkt Jun 6 at 5:26
    
FWIW, I think that's a great question: I never thought about mechanics of a language influencing search algorithms. I may be biased by my computer science background, though. :> –  Raphael Jun 6 at 6:54

2 Answers 2

up vote 3 down vote accepted

I'll try list some concerns that may separate searches in German from searches in English.

  • Compounds

    This is on your list already. You definitely want to match "Hut" with "Stoffhut" und "Anglerhut". It might make sense to rank results higher if they match multiple search terms in a non-compound way, i.e. "stoff hut" results in "Stoff für Hüte" first and "Stoffhut" then (if I wanted the latter, I'd have searched for "Stoffhut", arguably).

  • Plurals

    Plurals (and other forms) are less regular in German. If I search for "Hut" I want to find items with "Hüte" in their titel, and vice versa.

  • Umlauts

    Browsers, keyboards and product titles may deal with umlauts in different ways. To be safe, replace "ü" with "u" and "ue" for additional hits, and "ue" with "ü" in the other direction. (This may even be supported by your DBMS directly; buzzword phonetic search.)

General concerns include

  • synonyms ("Mütze" vs "Hut" vs "Kappe" ...) for more matching hits and
  • "Boolean search" for refining searches, e.g. "car pet -carpet" solves the compound problem.
share|improve this answer
    
Thanks for the answer. That is what I was hoping for. Does German use plural forms or otherwise modify the "base" form when making compounds? I get that someone may search for AnglerHüte, but is there a compound word that has Hüte in the beginning or middle (or some other pluralized or otherwise modified word in the beginning or middle)? Or is it always the last part of the compound word that pluralizes? –  Guy Schalnat Jun 6 at 16:35
    
@GuySchalnat: Modified nouns can certainly occur in the beginning or middle; consider "Geigenbauer" (a person who builds violins, plural) or "Schaffenskraft" (die "Kraft des Schaffens", Genitiv). –  Raphael Jun 6 at 17:39

Your question is too generic to give a specific answer. English and German are sister languages and any search phrase that an English or American person might enter, might as well be entered by a German (translated of course). I don't think that there are significant cultural differences in the way how the searches are formed. What Germans (ok, sample size is 1) might actually wish for when trying to look for shopping items is to restrict the results to businesses that:

  • Ship to Germany
  • Have a German translation of their site
  • Allow payment using methods that are popular in Germany (Überweisung)

Can you be more specific about the complaints that Germans typically have when using your engine?

share|improve this answer
    
One of the big problems we think we are having is with compound words. When we started out, in most languages at least, we would match any word containing the word the user asked us to search in. This became a problem when carpet showed up in searches for car or pet, so we refined our search so it would just show plurals and stuff. That works in English because we have fewer compound words. It doesn't seem to work in German as well. We also don't know what else we are missing. –  Guy Schalnat Jun 6 at 1:43
    
German is famous for the ability to stack words together to form compunds. Without the "world knowledge" that a carpet has nothing to do with cars or pets you are pretty much lost. in other words, you need more context... –  Stefan Schroeder Jun 6 at 18:18

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.