Page MenuHomePhabricator

Bosnian Wikipedia Search Engine Visibility and Page Views
Open, Needs TriagePublic

Assigned To
None
Authored By
Srdjan
Feb 19 2020, 5:12 PM
Referenced Files
F31622895: docker.png
Feb 19 2020, 5:12 PM
F31622891: USB-C.png
Feb 19 2020, 5:12 PM
F31622893: raspberry.png
Feb 19 2020, 5:12 PM

Description

It has recently come to our attention (the Bosnian Wikipedia community) that there's a staggering discrepancy in page views to the Bosnian Wikipedia (bs.wikipedia.org), and lack of search engine visibility.

Upon investigation, we have concluded that the cause is a decent chunk of views not coming from Google's Knowledge Graph, as is the case for other languages.
The following examples are provided from an incognito browsing session, with the language set to Bosnian, and the region to Bosnia and Herzegovina (&hl=bs, &gl=ba).

It looks like there's some sort of fallback set, but it does not include the Bosnian Wikipedia or the Bosnian language isn't explicitly defined.
This cuts the amount of traffic to the Bosnian Wikipedia considerably since most people use Google to look things up, and devalues the hard work our community members put into making new articles as very few people will actually get to see them.

If this is an issue on MediaWiki's end or with some local configuration, let's work out a solution. If it is not, since we are unable to contact Google directly, we were hoping someone here would be able to do so on our behalf.

I've already contacted the Wikimedia Foundation, and a representative suggested I file this ticket, assign it to the Language Team, and tag it with Discovery Search.

Event Timeline

This seems a valid task, but I'm not sure what search platform can do to help. We are primarily familiar with the internal implementation details of various search/discovery based services run inside the WMF. I'm not sure who to pass this on to that might know something about google and how they index/rank our sites.

It seems Google really ignores Wikipedia in Bosnian language. Latest example is COVID-19 article that was present on Bosnian wikipedia long ago, but Google will not generate the box on the right. Wikipedia in Croatian has added COVID-19 few days ago, a much shorter article then on Bosnian Wiki, and visitor count has skyrocketed over at Croatian Wiki, while plummeting on Bosnian Wiki. I am not sure where the issue is, but hopefully we can find a solution in the long term.

Removing Language-Team and Discovery-Search as this is out of scope for these teams.
Reminds me of T236241. Maybe Product-Analytics could help a bit, but not sure (see that task).

Note that Cantonese Wikipedia also experience similar phenomenon vs Chinese Wikipedia, according to on-site discussion record.