Last modified: 2013-12-13 20:38:52 UTC
Searching for expressions that have been set between guillemets (« and », "french quotes") yield no results. Compare this http://tls.theaterwissenschaft.ch/w/index.php?search=Top+Dogs&go=Seite&title=Spezial%3ASuche to http://tls.theaterwissenschaft.ch/w/index.php?title=Spezial%3ASuche&profile=default&search=«Top+Dogs»&fulltext=Search and this: http://www.google.ch/webhp?tab=ww#hl=de&q=top+dogs+site:tls.theaterwissenschaft.ch&oq=top+dogs+site:tls.theaterwissenschaft.ch
Hello, Thank you for your bug report. Google seems to ignore French quotes, as demonstrated in the following URL: https://www.google.com/search?btnG=1&pws=0&q=%22%C2%AB+Top+Dogs+%C2%BB%22
Yes – I guess I should clarify: Phrases on wiki pages that are set between guillemets can't be found by the internal search, probably because the search index saves those phrases as "«Top" and "Dogs»" instead of "Top" and "Dogs". Example: The page http://tls.theaterwissenschaft.ch/wiki/Volker_Hesse, contains the sentence "H. inszenierte unter anderem 1993 die deutschsprachige Erstaufführung von Tony Kushners «Angels in America», die Dürrenmatt-Collage «Fritz», 1995 Coline Serreaus «Weissalles und Dickedumm», 1996 die Uraufführung von Hürlimanns «Carleton», 1997 Schnitzers «Liebelei» und wurde mit den Ensembleprojekten «In Sekten» (1994) und «Top Dogs» (1996, Textgrundlage: →Urs Widmer) an das Berliner Theatertreffen eingeladen." Searching for "Ensembleprojekten", "Textgrundlage", "Berliner Theatertreffen" will find this page. Searching for "Fritz", "Weissalles", "Carleton" and "Liebelei" will yield no results, as those words are all set in-between guillemets in the aforementioned sentence. Searching for those last phrases with guillemets in place (as "«Fritz»", "«Carleton»" or "«Liebelei»") will produce results.
(In reply to comment #2) > Phrases on wiki pages that are set between guillemets can't be found by the > internal search, probably because the search index saves those phrases as > "«Top" and "Dogs»" instead of "Top" and "Dogs". Ok. This depends on the tokenization system being used: it's probably not an issue with Lucene or Cirrus/ElasticSearch, what search is that wiki using? How much control do we have on the tokenization for the standard MediaWiki search which IIRC may use MySQL directly or something like that?