I am very impressed with hard work put together by teams in translating, collecting, recording, line-by-line, maintaining the suttas here on the sutta central, thanks for all the hard work done by entire team. I often read suttas, as well use the audio function to listen to them
Having said that, the search function on the portal has some improvement areas, most often it doesn’t produce what I am looking for.
E.g. when searched for mahanidana sutta SuttaCentral produces 312 results none contains mahanidana sutta, when same done in google gives first link as [
Search should allow inline search like from:person and/or title:sn.46.51 which handles both simple and advanced searches
Objective should be to provide results in 1-3 at most pages 25 / 50 items, any more are not useful
Search should type ahead to complete the words from indexes
Search should remove dictionary words, punctuation marks, stop words etc as well to be more precise
Search should also include translating pali words to english and vice versa using different dictionaries like Abhijñā vs Abhijna both should provide similar results as we don’t expect users to type those specific characters
you can use apache solr / elastic search etc. which are industry leaders and open source to provide similar features and indexing, more like this etc.
or use some google integration which can be useful
We’ve had many discussions about the Sutta Central search function here and here and here and here … a final resolution is still in the works!
I created a custom Google search for SuttaCentral some time ago… you can access it at bit.ly/SuttaCentral - Using this for your search term of “mahanidana sutta” will get you the following result…
We certainly do need to improve the native search function on SuttaCentral though (currently we are using Elastic search)… let me assure you that the team, led by Ven @sujato is on it!!
Personally the only feature I would love is to checkmark/filter which nikaya/collection is visible. So if I search “phala” I can make it only show me results from AN,DN, MN, and SN. That would be amazing.
Thanks for the feedback, Faujidoc has already responded well.
But just some tips. One of the things our current search does it it tends to work by adding words, so “mahanidana sutta” is guaranteed to return lots of false positives.
“mahanidana” gives two results, one of which is to the correct sutta. “mahanidanasutta” also gives two results, one of which is correct. So that’s not too bad.
You’ll get better results if you just enter the sutta id “dn15”.
The goal of search should be to return relevant results no matter what the exact form of the input. But that is a hard problem. We get so used to searching on Google that we forget, it was by solving this (actually a vast number of very hard problems) that they became one of the most powerful companies in history.
So the way I see it, if Google sometimes does a better job than we do, that is to be expected.
I think the result is because it’s in the body of the text (see below) - searching other suttas using the same method does not return a result (e.g. “samkhitta sutta”) . Trying to understand elasticsearch Bhante Sujato’s tips helped a lot