The Language Converter can't support search results in Chinese wikis. It shows results only in zh-hans when typing zh-hans, and only zh-hant when typing zh-hant.
For example:
https://zh.wikipedia.org/w/index.php?title=Special%3A搜索&profile=default&search=2015年希腊纾困公投&fulltext=Search - no result
https://zh.wikipedia.org/w/index.php?title=Special%3A搜索&profile=default&search=2015年希臘紓困公投&fulltext=Search - 3 results
Description
Description
Related Objects
Related Objects
- Mentioned In
- T158203: Test and analyze new Chinese language analyzers
T138857: Serbian language search differentiates between Cyrillic and Latin alphabets - Mentioned Here
- T158203: Test and analyze new Chinese language analyzers
T138857: Serbian language search differentiates between Cyrillic and Latin alphabets
Event Timeline
Comment Actions
I guess you're asking: The LanguageConverter can't support search results in Chinese wikis. It shows results only in hans when typing hans, and only hant when typing hant.
Comment Actions
Sorry but it is still unclear here what are steps and settings to reproduce the problem, what is expected, and what actually happens. Feel free to edit the initial task description and title of this ticket.
Setting to "stalled"for the time being.
Comment Actions
will need to investigate the lang convertor...this might take some time. moving to the later bucket for now.
Comment Actions
But kindly Serbian also uses language converter, maybe also happened on Kazakh, Kurdish, Tajik, Uzbek?
Comment Actions
This should be helped by the new language analyzer being tested in T158203.
Both 2015年希腊纾困公投 and 2015年希臘紓困公投 get 14 results, and 2015年希臘紓困公投 is the top result in both cases.
Please feel free to try out the new language analysis and compare it to the current Chinese Wikipedia. (The content is week-old copy of Chinese Wikipedia and it has the search index and snippets, but not the articles.)
Comment Actions
I believe this is fixed. The examples given now return the same number of results (15), though in a slightly different order.
Chinese language wikis are now converting Traditional characters to Simplified characters for indexing and searching. It isn't perfect, but it generally works well. The slight difference in order in the results comes from slight changes in the scoring for exact matches between query and article. So, for example, a query with Traditional characters matches an article with Traditional characters a bit better, and the same for Simplified characters in query and article. It's usually does not make a huge difference, but it is definitely enough to reorder results that already have similar scores.
I suggest closing this ticket.
As for Serbian and others that use the Language Converter for changing the writing system used to display articles—unfortunately this requires a different kind of software, specifically an Elasticsearch plugin, that can do more or less the same job but do it inside the search engine. For Chinese, I found an external plugin that did the job. There may be something available for Serbian (see my soon-to-come comments in T138857).