This topic contains 13 replies, has 2 voices, and was last updated by 
- Posts
Hello. I am using ASP and it’s all working great.
I am running a Multisite network of two websites, main one in Hebrew and the second is in English. I’ve created (duplicated) a new search instance from my main one and recreated the Index Table with both sites selected (previously only the main website was indexed). UI and results show just fine, but Autocomplete and Keyword Suggestions do not work. I’ve tried changing a few settings in the Index Table/Search sections like using Google Keywords for suggestions and only Google keywords showed up.
Funny thing was, before I indexed the other website as well as the main one, Autocomplete and Suggestions worked, but the results themselves didn’t show.Any help on this please?
Thanks in advance.-
This topic was modified 1 year, 11 months ago by
profilesoft70.
Hi,
Sure, can you please add temporary FTP and back-end access? I would like to test the issue, as well as check the configuration and do debugging if neccessary.
Best,
Ernest Marcinko
If you like my products, don't forget to rate them on codecanyon :)
You cannot access this content.Thank you!
I have tried on both installations, but it seems to work all right on my end.
Best,
Hebrew autocomplete: https://i.imgur.com/oOo573R.png
Hebrew keyword suggestions: https://i.imgur.com/D7U0Khi.png
English autocomplete: https://i.imgur.com/Mk6UTHE.png
English suggestions: https://i.imgur.com/clD6gon.png
Ernest Marcinko
If you like my products, don't forget to rate them on codecanyon :)
At the last image it doesn’t suggest anything altho your keyword “tst” is one character away from the item with the word “Test” in it. So suggestions still don’t work in English…
That is because a “smarter” similar keywords algorithm is applied to the taxonomy term sources, and not the post title sources. That is because of performance reasons. While the keyword “tst” is only one character away from “test” it is extremely difficult to gather and compute the suggestions without making a huge impact on the server perfromance.
Best,
Please mind, the plugin tries as best as possible to match suggestions, but of course the google keyword suggestions source is far superior – at it comes directly from google, and it does not have to consider performance related issues.
Ernest Marcinko
If you like my products, don't forget to rate them on codecanyon :)
Makes sense, but it’s still weird because in the Hebrew site you have items that contain the word “טסט” and when you type in “טסטט” or even “טסטטט” it shows all the suggestions, but on the English site when you have an item with the word “digital” – you type in “digitall” and it doesn’t return anything. They are both configured the same way but it seems that the word matching mechanism works a lot better in Hebrew.
Is it possibe that those suggestions in hebrew are maybe from post tags or other selected taxonomies? That would explain it pefrectly.
Best,
Ernest Marcinko
If you like my products, don't forget to rate them on codecanyon :)
No, it takes the suggestions from the titles. Would it help to include the title text in the post tags? Why is that less expensive to load?
Interesting, probably the SQL Soundex algorithm works better in hebrew for checking similarly sounding words. Moving to tags could actually help, it should not be too expensive either. I would probably recommend creating a new custom taxonomy for that if you can, in case you want to delete them later, it makes it simpler.
Best,
Ernest Marcinko
If you like my products, don't forget to rate them on codecanyon :)
Why would searching in post terms/tags be less expensive than in titles?
Anyways, we can’t use this at all, even at the Hebrew site, because we need all the websites to have the same logic.
It would be great if you found a solution for this, we’re really into it.
Thanks in advance.In short: It is faster to simply select taxonomy terms and process them within PHP to find possible similar keyword matches – because they are more likely to be very short words. It works nicely even for cases with over 50 000 terms. In case of titles the table is much bigger, and most titles are multiple words long, which exponentially increases the comparision process. Not to mention the possible filter criteria, exceptions during the initial select query etc.. It would not be an issue whatsoever if the processing power and memory on regular hosting servers was much greater, but we can not risk that unfortunately. We try to limit the plugin to run on the customers local server instead of charging monthly fees for external processing services.
Best,
This feature will be improved for sure in the future as we are adding improvements and new features. Do not worry, the solution for the terms was added not too long ago, and it seem to work really great in comparision how it was before. I’m sure we can find a way sooner or later to somehow use that on titles without the performance issues.
Ernest Marcinko
If you like my products, don't forget to rate them on codecanyon :)
Thanks for the detailed reply. We’ll be looking out for these improvements, but for now, do you think that copying the titles into the post tags would make it work? That case we can just create a hook that does that on post save.
By the way, one of our admins accidentally deleted the account we’ve set up for you, if it logged you out then you can log back in as I’ve recreated your account with the same credentials.Thanks again.
You are welcome!
That should probably work in my opinion. The terms table has much better indices, so there should be no performance issues either.
Best,
Ernest Marcinko
If you like my products, don't forget to rate them on codecanyon :)
-
This topic was modified 1 year, 11 months ago by
You must be logged in to reply to this topic.