Wiki Search Sucks

DMS General
#1

So…
We all like to complain about the wiki search. And I tottaly get it. But in looking to make it better, I promised (somewhere on here) to take note the next time I found a thing that didn’t work worth a plug nickle.
So, here it is.


Try any way I can think of, short of using Google, I can NOT find that file.
it’s not in the file list.
Nor the “special pages” as suggested here by Ken

Even by name, the file list/media search will NOT turn it up.
So what’s up with that?
I can see it was linked to the “Financial Committee” wiki page by @Lampy but why is the metadata non-existent. The file appears to me to be absolutely unfindable using the wiki search, and I’d like to know why, and if we can do anything to fix that (other than tell folks to use Google).
Or was it just my turn to come up completely flat on the wiki-search-fu?

Thank you for reading.

2 Likes
Wiki Improvements and Search
#2

What sucks even more is that I get zero results with the site:wiki.dallasmakerspace.org tag on Google searches.

2 Likes
#3

Well. To be honest, I NEVER use that, because I find it just as easy to put the /wiki on the end as to put it at the beginning. But, I hear you.
There’s no content on that site. it’s just a redirect to the actual site. So I can at least explain why I can’t find it that way. I’m not sure what would need to happen to fix that, but it’s outside my available scope.

1 Like
#4

One problem is the lack of keywords. For example “deduction” does not appear in the document / form or on the wiki page.

If the wiki supports it a few of these may help…

Should the wiki search use Google Search instead of the built-in search? (I believe I have seen other wikis that do that.)

1 Like
#5

The lack of keywords is, i think, caused by the fact that this file is stored in the wiki datastore, but is not part of the wiki. I think that’s WHY it has no metadata tied to it. And why we can’t change that metadata.
But I could be wrong.

Sounds like a question for the wiki herder. I think that’s @AndrewLeCody but might be someone else(s?) at this point, probably on the @Team_Infrastructure .

Here’s a 8 year old write up that makes it sound easy peasy: https://meta.wikimedia.org/wiki/How_to_add_Google_search_to_your_MediaWiki_Search_Results_Page

1 Like
#6

For wiki searches to work on attached files, the easiest thing is to reference the file in the actual wiki page and describe the contents using key words the user might need when searching. it is far from fool proof but generally effective.

1 Like
#7

We’re also hella behind on MediaWiki. So far back that I don’t care to do the upgrade because I legitimately feel it may result negatively.

1 Like
#8

@bscharff, back when you were running Infra, didn’t you chat with @AndrewLeCody about upgrading to the WYSIWYG version of the software? What was the outcome of that?

1 Like
#9

I didn’t talk with Andrew about it, it was with Alex. I started to make a clone of the environment (the wiki/ directory and database) but as I read the upgrading guide and then looked into the extensions we used, I realized it was going to be a big ordeal and it would take a TON of time to upgrade and then slowly figure out all the resulting errors.

It’s a worthy effort, but I’m sad to say I don’t have the time to commit to it now. Looking back, it probably should’ve been higher on the priority list.

2 Likes
#10

media wiki has support for elastic search but in general the built in search is way better than what google crawls since it is directly searching the sql backend while google is just pulling from html and has thier machine learning to rank keyword weights for blogs instead of wikis.

#11

Agreed, easier to just use a github wiki or github page even.

#12

But, for example, this particular instance, there appears to be nothing IN the SQL backend for the wiki search to index. Or so is the appearance to me. Supporting the notion that google does a better job than the internal search, is that things get found with a google search of the site, whereas not with the internal.

moving the (pertinent) existing data to a completely new wiki/format sounds like EVEN MORE work. With zero first-hand knowledge, though, what do I know?

#13

moving the (pertinent) existing data to a completely new wiki/format sounds like EVEN MORE work. With zero first-hand knowledge, though, what do I know?

Github wiki supports mediawiki format so its just dumping both the files and articles via a sqldump and then iteratting that with python and urllib2 to post it into github.

1 Like
#14

Found a new example.
This time, partially user…error? Not really “error” so much as “misapplication” of the search function; what follows we all probably already know, but sometimes need reminded, and sometimes is more crucial to success than others. Googler & its ilk has made us lazy…


You’d think, once you know the BOD meeting happened in October 2017, you could just search for 201710 in the wiki, and it would turn up any files/pages with that in the title, certainly, or on the body, right?
Wrong. You’ll get nothing out of that search.
BUUUUUT
Add the wildcard (201710*) and wham-oh. Results of all several pages with 201710something. Including the BOD meeting.

By the by, “201710 site:dallasmakerspace.org/wiki” turns up those same results without the asterisk. See what I mean about Googler making us lazy? :slight_smile:

PS: if you search for the EXACT name/date, it will find it, too: 20171020 works fine on the wiki search.