Talk:Wikimedia monthly activities meetings/Quarterly reviews/Discovery, October 2015

From Meta, a Wikimedia project coordination wiki

Comment on the 0 results search[edit]

Hi folks, as an information professional, I search for more obscure things in the article space than the average user. I really don't mind getting a "0 results" message, and in fact, find this quite helpful at times. If it's not there, it's not there. This can be exactly what I was attempting to find out. "We don't have it" is straightforward and honest, and should not be regarded as a problem. It saves a lot of time for people like me to get a "0 results" message, and it helps us figure out when we need to look in other places. You can waste a lot of time as an information professional when databases refuse to display "0 results," and instead keep presenting things that are only vaguely related to your search term.

A "we don't have that exact spelling, may we show you a related spelling" message can be helpful, but it certainly isn't required in all circumstances, and can actually be rather annoying at times.

The searchable corpus here is just Wikipedia, not the entire Web, so we can reasonably expect "0 results" to remain a constant factor in searches here.

People may actually be very happy to see "0 results" in Wikipedia if they are searching for non-public, proprietary, or personal information. "0 results" is a useful tool for seeing how search is working, but should not be assumed to be synonymous with a dissatisfied user.

A related metric mentioned in the course of your discussion that would be of interest to those of us working on more obscure historical topics and locations is whether our articles are ever read by human eyes, or only by bots and other automata ... :) --Djembayz (talk) 16:25, 17 October 2015 (UTC)[reply]

Alternate approach to "0 results"[edit]

Are there any indications that can be drawn from the analytics as to what kinds of "0 results" searches are actually searches for encyclopedic content? Looking at this could be a big waste of time, or it could show patterns of what's missing in Wikipedia. If it shows that a sizeable number of people are really more interested in "how-to," "business directory", "biographical directory", or "travel directory" information than encyclopedic content, you could view it as an opportunity to provide a general referral to a different type of website ... --Djembayz (talk) 16:45, 17 October 2015 (UTC)[reply]

You should probably read mw:User:TJones_(WMF)/Notes/Survey_of_Zero-Results_Queries. :) Nemo 19:41, 17 October 2015 (UTC)[reply]
Thanks, @Nemo bis:, that's informative. Looks like TJones is on the right track already. It does in fact appear that searches for non-encyclopedic content are taking place. There's really no telling what user intent is simply by looking at search terms, but it does seem possible that a fair number of those searches could be seeking to confirm whether the term is in the encyclopedia or not. In that case, adding a message telling the user that it's an encyclopedia, not a directory of business, biography, travel or how-to information won't be helpful. The user would still want to confirm whether their term appeared.
It appears from the report that although manually working through a semi-random sample every so often would be useful, and has the potential to reveal some unexpected things, it won't necessarily turn up anything that makes a huge dent in the 0% results.
As an outsider looking this over, it appears that there is an ongoing need to check that third party apps and systems are playing nicely with Wikipedia search, and that the more automated sorts of queries are working. Truncated search terms that seek to narrow the type of data retrieved (eg., "paint") and abbreviations look like some of the more low hanging fruit.
Conclusion here remains that since the Wikipedia database attempts to represent the sum of all encyclopedic information, rather than the sum of all information, getting 0% results will remain one way that users determine the overall scope of the database for themselves. --Djembayz (talk) 13:48, 20 October 2015 (UTC)[reply]