Searching with a wildcard character (part 2 :-)

If I'm performing an inline search for a word using a wildcard character at the end of a word (like του* for example), the results take a second or two (I'm using SBLGNT btw). But if I search using a wildcard character at the beginning of a word (like *υσα for example), it can take anywhere from 6-12 seconds on my M2 MacBook Air.

The search is slower if there are more hits, but it seems to me that the search is exponentially slower if the asterisk precedes the rest of the word.

Is this normal behavior? And if it is, can it be fixed? In Accordance, both these searches are instantaneous, and as we have more users switching to Logos, they'll be uncomfortable with this performance.

Find more posts tagged with

Desktop App

Comments

Justin Gatlin

I get the same results that you do. The same thing happens searching English Bibles (for example *ing versus pro*). This is expected behavior for most search engines, since they generally read from left to right and so wildcards at the end require far more steps. Accordance must be smart enough to do searches from right to left when the wildcard is at the end of a word, which is cool.

Bradley Grainger (Logos)

The search is slower if there are more hits, but it seems to me that the search is exponentially slower if the asterisk precedes the rest of the word.

Is this normal behavior?

Yes, this is "normal" for Logos. Wildcard searches are slower (and prefix wildcard searches are much, much slower).

And if it is, can it be fixed? In Accordance, both these searches are instantaneous, and as we have more users switching to Logos, they'll be uncomfortable with this performance.

Technically, yes, it can be fixed. (There are various tricks you can use with specialised indexes, such as "rotating" a single wildcard to the end of the word, so that wil*ard becomes ardwil* which can be implemented as an efficient prefix search. I would guess that Accordance does something clever like this. Or maybe the list of terms in a single book is just so small Accordance can brute-force check them all quickly.)

However, this has been a "known issue" in Logos searching since Logos 4 came out and it's not high on the priority list to fix any time soon.

Mark Allison

However, this has been a "known issue" in Logos searching since Logos 4 came out and it's not high on the priority list to fix any time soon.

Thanks for the detailed response Bradley. It helps to know that it's not just me or my computer. Being a new user can be pretty frustrating that way.

Right now Accordance is a dumpster fire. I know many Accordance users who are abandoning that sinking ship (to use another metaphor) and looking at Logos as an option. There are some pretty radical differences between the two apps, mostly related to Logos' historical commitment to the pastoral market, and Accordance's to the academic market. However, Logos and Accordance have both worked hard to meet in the middle, and I think Logos is much better at original language research than most Accordance users give it credit for.

A lot of BibleWorks users moved to Accordance when it shut its doors. If Accordance continues on the same downwards trajectory, there will be two groups of users who work with original language research all day, are used to lightning-fast searching in both Greek and Hebrew, and are looking for other alternatives.

All that to say that Logos might want to take a look at adjusting some of their priorities to meet the needs of this group of users who are VERY committed to Bible software and aren't afraid to spend money on the tools they use every day.

MJ. Smith

Being a new user can be pretty frustrating that way.

A second approach is to ask why you use wildcards? It may be that Logos solves the problem in a different way. What are you looking for that requires the use of wildcards?

Mark Allison

What are you looking for that requires the use of wildcards?

A user on the Accordance forums was looking for every inflected form of a Greek in the NA28 that ended with υει, and needed to also see the lexical form of each word.