Support » Requests and Feedback » Plugins search results is never relevant :(

  • Hi,

    Both on WP plugins page as well as on WP back end plugins search, the results are rarely relevant & accurate.
    Some plugins appear on any keyword search results, while relevant ones are not displayed on the 1st page.
    For example… try to look for Admin CSS plugin by Scott Reilly.
    Use keyword “Add Admin CSS” & you’ll find it on 3rd page if you have the patience to keep on looking, after 1st page includes irrelevant results, such as WP Fastest Cache, Metaslider, Hummingbird & more 🙁
    The same issue exists even if you use the author’s name in the search.

    I cannot understand how WordPress affords such a search engine which is totally the opposite of being user-friendly.

    Made my point.
    Thanks,
    Ahrale

    The page I need help with: [log in to see the link]

Viewing 7 replies - 1 through 7 (of 7 total)
  • WordPress uses text-based search for plugins.
    Plugins with a large number of active installs, who only mention a particular keyword in their description, are thus more likely to be displayed on the first page or two because of this.
    In order to get more relevant search results, it would be necessary to give a higher weighting to the title and tags (since there are a limit to the number of tags that can be displayed).
    Have you considered opening a Trac ticket?
    Please note that actionable tickets are significantly more likely to receive traction. This means that your recommendations for improvement would need to be very specific.

    Thanks, @carike,

    It’s obvious it uses basic search with no relevance.
    This is what I wonder about… How many years it will keep on being this way, without anyone thinking it is about time to fix this issue.

    Also, many times appear plugins that are no longer supported or updated.

    Unfortunately, I don’t know what’s Trac ticket is & couldn’t find it here on the forums.

    I just hope that I’m not the only one who understands there is no logic in the current WP plugins search.

    Thanks again

    Here is a previous meta ticket related to the relevance of plugin search results:
    https://meta.trac.wordpress.org/ticket/4930
    The issue here is often that it is plugin authors themselves that are quibbling over their particular search ranking.
    As you may be able to imagine, it is not possible for all plugin authors to be happy about their ranking, which may account for fewer volunteers wanting to take on a project related to search.

    I (and others) are of the opinion that the search places too high an emphasis on the number of active installs.
    Personally, I support a filter-based system, which puts the results back under the control of the user, rather than a purely algorithmic approach.
    That way a user could exclude a plugin that has not been updated in the past 6 months, search could be expanded to include results that include a particular keyword anywhere (like the description) (my preferred default being only searching title, slug, tags for the keyword), etc.

    A new meta ticket can be created using the following link:
    https://meta.trac.wordpress.org/newticket

    Moderator Samuel Wood (Otto)

    (@otto42)

    WordPress.org Admin

    It’s obvious it uses basic search with no relevance.

    No, actually, the plugins directory uses an elasticsearch engine with very high amounts of relevance.

    You can find more about this here, and elsewhere on the data.blog site.

    https://data.blog/2017/03/15/improving-relevance-and-elasticsearch-query-patterns/

    Edit: BTW, this is all open source. You can see the search parameters in the meta trac code browser.

    Some people don’t like the way the search works. That’s fine. That is inevitable. Those are not the people we’re building it for.

    I’m sorry Samuel…

    When I use a very specific keyword & get results of totally irrelevant plugins that are not related at all to the keyword subject, You cannot claim this elasticsearch engine gives high amounts of relevance.

    This is completely not true!

    I’m not here to argue. I only wanted to point out something that can be highly improved, but if you deny the truth I rather quit & stop this discussion.

    Thanks for the reply.
    & thanks again @crackie for the link.

    • This reply was modified 4 months, 1 week ago by Ahrale.

    > https://data.blog/2017/03/15/improving-relevance-and-elasticsearch-query-patterns/

    That is a fascinating article on many levels 🙂

    > Some people don’t like the way the search works. That’s fine. That is inevitable. Those are not the people we’re building it for.

    I have to respectfully disagree with that.
    I genuinely understand not wanting to build the internal plugin search for developers, but focussing on end-users instead.
    The thing is that, increasingly, users without any technical background are the ones to indicate that they can’t find what they are looking for in the plugin search.
    I’d add that, not having any plugins in the repository, I don’t have a “stake” in any specific search ranking myself.

    The issue seems to be that highly-focussed, single-purpose / set-and-forget plugins performed better in the previous search.
    This is understandable, since the number of visible tags are limited – and a multi-purpose plugin may thus not be as likely to show up for a “secondary” or “tertiary” / auxiliary function / purpose.
    There was also an assumption that smaller plugins could not handle exponential growth.
    A side-effect of boosting a high number of active installs in particular, is that it stifles competition, with higher barriers to entry for new plugins / new authors. I don’t believe that that is a good thing for the ecosystem (and the end user in particular).
    Multi-purpose plugins are a LOT more likely to have a high number of active installs.
    If a user is looking for one type of function, they may now have a lot of code on their system that they will never use / need. While a multi-purpose plugin may contain a particular function, it may not necessarily do it well / optimally / as well as a different plugin.

    By all means, using these boosting signals for tabs like “Popular” or “Recommended” is perfectly acceptable.

    When you search a specific plugin name and it does not show up on the first page results though, that seems to me like seasoned developers telling end users what they should want, instead of asking users what they want.
    Although I do not believe that this is out of malice, it is still not empowering users, which, at the end of the day, is what Democratizing Publishing is all about.

    Thanks, @carike,

    for your attitude & reply. It is appreciated.
    Yes, I believe the current way, totally ignores the real need of the end-users and seems even not to care about the end-users.

    Plenty of explanations can be given, but in the bottom line, the current system is surely not friendly for WP users.

    I can only say, I hope it will be changed & some logic enters this part of WP.

    All the best to all & thanks again for listening & replying.

Viewing 7 replies - 1 through 7 (of 7 total)
  • You must be logged in to reply to this topic.