When I scan this page, it does not interpret “l” correctly, as Roman numeral 50. See logs and images below.
5367.Logs.zip
I don't know that Roman numerals are supported, but in any case, I would doubt this bug would get fixed. The most common reading of the reference would be Psalm 1:13. It would be difficult to "fix" this bug without creating a new one.
If you didn't tell me that was supposed to be an L, I would have assumed it is the number 1 with a period after it and I'm not a computer but a reasonably intelligent native English speaker. I don't think it is reasonable for OCR to detect that.
(cap L) LLLL (lc L) l llll (one) 1 1111 Times New Roman on a pc all look the same [here the different font shows l & 1 differ]
(lc L.) l.l.l.l.l.l.l. [on my CPU close together] (Times New Roman on a pc)
(one .) 1.1.1.1.1. [on my CPU more distance apart then lc L] (Times New Roman on a pc)
Only difference between (one) and (lc L) is spacing
Good luck finding an OCR program that can tell the difference [Depending on font]
[My OCR saw the first two lower case L . as one. and the rest as lower case L. (l.l.l. as 1. 1. l. l. )
But it saw the (Lc L) lllll as all ones (Times New Roman on a pc)
The font on the forum (as seen by me) has just a vertical line for the lc L and the one looks like a flag on a pole.
You guys had some good points, but I still think something is wrong. Let's take things in sequence:
As you can see from my example below, it does work sometimes. So the problem is probably not with the OCR (although that is a good point), but with how Logos App handles the output of the OCR. What I think it should be doing is one of the following:
Just my 2 cents, coming from the perspective of a programmer. I admit I don't know how OCR works, but if it is anything like some other stuff I have dealt with, I imagine the Logos App gets a bunch of texts from the OCR library. It then interprets the text, looking for verses, as best it can. I just think the logic could be tweaked to improve this, is all. Yes, I freely acknowledge this is a weird case, just saying there might be some easy logic tweaks to improve things a little bit, and I bet the logic changes would be pretty quick and easy and cheap. But maybe not.
Now, on to the example that worked. In the below case, Ps lxxxvi . 11 gets converted to Ps 86:11, like it should have been. This proves that the Logos App sometimes handles the roman numeral "l" correctly. Note that in this case, the roman numeral "l" looks almost exactly like the numeric "1" in the verse, but it still handles it correctly.
Obviously, Roman numerals works because of the Amos passage
It isn't obvious at all based upon this particular example. The only "verse 21" in Amos happens to be 5:21. However, your Psalm 86 example is evidence enough for me.
I admit I don't know how OCR works, but if it is anything like some other stuff I have dealt with, I imagine the Logos App gets a bunch of texts from the OCR library.
They do and the mistakes that come must be manually fixed. OCR has improved, but it is still not perfect. I can tell you that none of the photos you have uploaded were of "good" quality, which diminishes the ability of OCR to function properly.
In addition, many FL resources require a multi-language OCR because of Greek and Hebrew words inserted. Single language OCR is much simpler.
