Seite 1 von 1

Browser preferred language in Yacy search

BeitragVerfasst: So Mai 11, 2014 5:08 pm
von sto
Hello all forum users,
I am quite new in the field of Yacy but I am definitely supporting it. I have been running a senior node 24/7 for a few weeks now (dome_cirrus) and I could crawl a few million pages of French + English wikipedia and various newspapers.

The most frustrating thing for me is the language management in the search. When I look for, let's say, Obama, I will have the wikipedia results in many languages, which is absurd. I want the results in the languages I can read (namely French, English and a bit of German).
The solution would be to add /language/XX at the end of the search request. But I don't want to do it. It is not user friendly at all.

Easy solution suggested to developers: get the browser preferred language, as DuckDuckGo does, to present the results in that language (user's language, let's say German), then in English (universal language), then in French/Russian/Spanish/... (other languages than browser's preferred that the general user will generally not understand)

Is it feasible?

This idea leads to a concept of language sub-networks. When someone makes a search in French, it would be good to ask French peers first if they have results for the query. It may not apply with English, of course, as everybody indexes some English pages.
I have no idea if such a tuning of the DHT system is feasible.

Thank you for reading. Please reply if you have better solutions. I would be glad to have some feedback on these points.

PS: I copy-pasted my original message from http://www.yacy-forum.org/ as there does not seem to be much activity there.

Re: Browser preferred language in Yacy search

BeitragVerfasst: Sa Mai 31, 2014 8:21 am
von biolizard89
sto hat geschrieben:Hello all forum users,
I am quite new in the field of Yacy but I am definitely supporting it. I have been running a senior node 24/7 for a few weeks now (dome_cirrus) and I could crawl a few million pages of French + English wikipedia and various newspapers.

The most frustrating thing for me is the language management in the search. When I look for, let's say, Obama, I will have the wikipedia results in many languages, which is absurd. I want the results in the languages I can read (namely French, English and a bit of German).
The solution would be to add /language/XX at the end of the search request. But I don't want to do it. It is not user friendly at all.

Easy solution suggested to developers: get the browser preferred language, as DuckDuckGo does, to present the results in that language (user's language, let's say German), then in English (universal language), then in French/Russian/Spanish/... (other languages than browser's preferred that the general user will generally not understand)

Is it feasible?

This idea leads to a concept of language sub-networks. When someone makes a search in French, it would be good to ask French peers first if they have results for the query. It may not apply with English, of course, as everybody indexes some English pages.
I have no idea if such a tuning of the DHT system is feasible.

Thank you for reading. Please reply if you have better solutions. I would be glad to have some feedback on these points.

PS: I copy-pasted my original message from http://www.yacy-forum.org/ as there does not seem to be much activity there.


Personally I wouldn't want the HTTP header to determine what results I get. I think it would be reasonable for YaCy to support a set of configurable "default languages" so that if I choose French and English as my default languages, YaCy should automatically include those constraints on searches unless I choose otherwise for that specific search. It would also be fine with me if YaCy detected my HTTP header languages the first time I used it, and asked me if I wanted to use those as the default languages for future searches.

Re: Browser preferred language in Yacy search

BeitragVerfasst: Mi Jun 04, 2014 9:23 pm
von sto
Thanks for your feedback, this solution looks 100% OK to me.