Shalom, isn't the StandardAnalyzer supposed to take care of "forking" in case of ideograms? I.e. use a Japanese-friendly analyzer for japanese characters and an English-friendly analyzer otherwise.
As Jack pointed out, edismax is nifty to expand a query on multiple fields. If you need to do more than a "linear expansion", you'd need to write your query component, which works quite well to inject particular business logic (e.g. prefer documents with a title that matches fully the query, ....). Paul On 7 juil. 2013, at 18:29, Shalom Ben-Zvi Kazaz wrote: > Hi, > We have a customer that needs support for both english and japanese, a > document can be any of the two and we have no indication about the > language for a document. ,so I know I can construct a schema with both > english and japanese fields and index them with copy field. I also know > I can detect the language and index only the relevant fields but I want > to support mixed language documents so I think I need to index to both > english and japanese fields. we are using the standard request handler > no dismax and we want to keep using it as our queries should be on > certain fields with no errors. > queries are user entered and can be any valid query like q=lexmark or > q=docname:lexmark AND content:printer , now what I think I want is to > add the japanese fields to this query and end up with "q=docname:lexmark > OR docname_ja:lexmark" or "q=(docname:lexmark AND content:printer) OR > (docname_ja:lexmark AND content_ja:printer) " . of course I can not ask > the use to do that. and also we have only one default field and it must > be japanese or english but not both. I think the default field can be > solved by using dixmax and specify multi default fields with qt, but we > don't use dismax. > we use solrj as our client and It would be better if I could do > something in the client side and not in solr side. > > any help/idea is appreciated. ?
