Massive Matches Query?

  • Agrajag27

    (@agrajag27)


    9 years, 1 month ago

    Our site has been happily using this plugin for some time to drive our complex relations for quite some time (Click here to see an example).

    We run a movie review site and our biggest challenge is relating films. Genre is generic and is a nice final touch, but too broad. Using Keywords has gotten us into a bit of trouble as now we have 2,000 keywords. Many are used once or are simply synonyms of another word (and thus not relating). In literature there’s a big question about the true number of plot elements that really exist. Vonnegut argued that it was six. There’s a good argument for 20.

    My question is, say we shift over to just support say 30 keywords for relation (plus the dozen or so genres (Action, Adventure, etc.) You’ll have literally hundreds of films matching on “Quest, Revenge, Rivalry, Sci-Fi, Thriller”.

    What happens then? If 100 posts relate and I just want to show say, six, how is that resolved? Will all 100 show the same six all the time?

    Is this a bad idea from a performance standpoint? I would think less terms means easier indexing.

    Lastly, in case it’s not covered above, can the 100 be randomized so that if it matches 100 then just X are shown from that set?

    Thanks again. Hope all is well on your side.

Viewing 2 replies - 1 through 2 (of 2 total)
  • Plugin Author keesiemeijer

    (@keesiemeijer)

    9 years, 1 month ago

    It’s difficult to say. If you assign many terms to all posts, let’s say 50, is a bigger performance hit than the total amount of terms that exist (lets say 5000).

    How many terms do you assign on average?
    How many posts are there in total?

    I would definitely decide on a set of terms that have meaning and is used across all posts. This will help you index your posts better and help with performance.

    What happens then? If 100 posts relate and I just want to show say, six, how is that resolved? Will all 100 show the same six all the time?

    Could be, if they are closely related (have common terms).

    Is this a bad idea from a performance standpoint? I would think less terms means easier indexing.

    Yes, the less terms per post the better the performance.

    Lastly, in case it’s not covered above, can the 100 be randomized so that if it matches 100 then just X are shown from that set?

    Yes, set the number of related posts to 100 and use this in your theme’s functions.php file.
    https://keesiemeijer.wordpress.com/related-posts-by-taxonomy/recipes/#better_random_posts

    Thread Starter Agrajag27

    (@agrajag27)

    9 years, 1 month ago

    Thanks for the response:

    1. Each post gets Genre tags that come into play. Those number from 1 to perhaps a max of 5. Average would be 3. Then it also gets Plot Keywords added. In the past these ranged from 3-8 from a total of over 2,000 (thus our desire to standardize) with maybe 5 being average. So, each post would have an average of 8 total terms.

    2. There are literally thousands of posts with hundreds added each year to the total.

    3. I’m not sure we’re on the same page with my last request on randomizing. My number of “100” was just an example, not any sort of hard limit. I’m simply giving the example of a case where a given post happens to match 100 (could be 200, 128, 12, 422, etc.) Obviously we only display a few matches (generally 6).

    IF Post 12345’s terms find 130 matches, I THINK it makes the best sense for users to not just always see the same 6 each time, but instead allow a means for them to have the potential to see all 130, just 6 at a time. Thus, randomizing the output seems the best approach to my mind. Make sense, or is there a better way to go?

    Thanks again.

Viewing 2 replies - 1 through 2 (of 2 total)

The topic ‘Massive Matches Query?’ is closed to new replies.