is there an effort in progress to fix the parallel resources which aren't actually parallel?

Page 1 of 1 (16 items)
This post has 15 Replies | 0 Followers

Posts 144
Josiah | Forum Activity | Posted: Sun, Apr 28 2019 9:20 AM

If I'm reading a Tertulian defense and want to check another translation, I don't want to be taking to the Table of Contents of a work that doesn't even include the particular defense I'm reading...  It's bad enough the way all the DSS are indexed in parallel resources, but it shouldn't be hard to fix the Church Fathers references?  Right?

Posts 24472
Forum MVP
Dave Hooton | Forum Activity | Replied: Sun, Apr 28 2019 5:45 PM

Please provide examples of what should be parallel.

Dave
===

Windows & Android

Posts 2948
SineNomine | Forum Activity | Replied: Sun, Apr 28 2019 6:17 PM

Josiah:
If I'm reading a Tertulian defense and want to check another translation, I don't want to be taking to the Table of Contents of a work that doesn't even include the particular defense I'm reading...  It's bad enough the way all the DSS are indexed in parallel resources, but it shouldn't be hard to fix the Church Fathers references?  Right?

Don't get me started on tagging of St. Augustine...

Posts 144
Josiah | Forum Activity | Replied: Sun, Apr 28 2019 6:27 PM

SineNomine:

Josiah:
If I'm reading a Tertulian defense and want to check another translation, I don't want to be taking to the Table of Contents of a work that doesn't even include the particular defense I'm reading...  It's bad enough the way all the DSS are indexed in parallel resources, but it shouldn't be hard to fix the Church Fathers references?  Right?

Don't get me started on tagging of St. Augustine...

I'm relieved you at least understand what I'm talking about.  From the above, it seems like Dave doesn't understand the issue.  I'll try to take some screenshots.

Posts 144
Josiah | Forum Activity | Replied: Sun, Apr 28 2019 6:46 PM

Dave Hooton:

Please provide examples of what should be parallel.

Let's look at some Ancient Literature...

...hmm, I always like Tertulian...

...ah, hah, witty and sardonic, I love it, let's check another translation...

...Logos found two other Parallel resources...

...let's check the Catholic University of America Press translation...

HUH?!

...oh, I see.  "De exh. cast.  6" isn't actually in this book.  The whole book is just haphazardly tagged as being Parallel to "Tertulian" and this is the last place I left off reading (if I had never read it before, it would open to the title page).  hmm....that's not helpful...  It'd probably even be better 'untagged' than tagged so haphazardly...

--

Another classic example is open up a DSS.  Take a look at the Parallel Resources.

Literally ALL 200+ scrolls are considered parallel to each other, when in actuality for any given verse only a few ever overlap.

Posts 24472
Forum MVP
Dave Hooton | Forum Activity | Replied: Sun, Apr 28 2019 6:51 PM

Josiah:
it seems like Dave doesn't understand the issue.

I tried a few Tertullian datatypes in ANF3/4, and I don't have any parallel resources i.e. other resources with that milestone. But I have several resources that reference Tertullian e.g. Dictionary of the Apostolic Church, History of the Christian Church, A Dictionary of the Bible (Hastings), TDNT.

Dave
===

Windows & Android

Posts 144
Josiah | Forum Activity | Replied: Sun, Apr 28 2019 7:06 PM

Dave Hooton:

Josiah:
it seems like Dave doesn't understand the issue.

I tried a few Tertullian datatypes in ANF3/4, and I don't have any parallel resources i.e. other resources with that milestone. But I have several resources that reference Tertullian e.g. Dictionary of the Apostolic Church, History of the Christian Church, A Dictionary of the Bible (Hastings), TDNT.

Yeah, I posted screenshots above, it should make sense.  I have pretty much every package except Reformed and SDA plus a lot of individual purchases, so it probably requires a certain collection of resources. 

The "Cited By" tool is great for finding resources that reference something.  There are also cases there where someone thinks it's fun to reference COS 1 or something (THE ENTIRE VOLUME) so it comes up as false positives on nearly every search, but that is not Faithlife's fault, that's people who didn't cite their sources well.  Faithlife just tags them as they are written.

Posts 144
Josiah | Forum Activity | Replied: Sun, Apr 28 2019 7:11 PM

Dave Hooton:

Josiah:
it seems like Dave doesn't understand the issue.

I tried a few Tertullian datatypes in ANF3/4, and I don't have any parallel resources i.e. other resources with that milestone. But I have several resources that reference Tertullian e.g. Dictionary of the Apostolic Church, History of the Christian Church, A Dictionary of the Bible (Hastings), TDNT.

Interesting you mention Milestone.  If I open "Tert., De exh. cast. 6" in ANF4 I ALSO have no Parallel Resources.

If I type {Milestone <Tertullian Tertullian, On Exhortation to Chastity 6>} into a search, I get both ANF4 AND the Paulist Press translation from my Screenshot above.  AND even more interestingly, neither of the two false positives that were showing up on the Parallel Resources tab.

So, it seems like if they could get the Parallel Resources tab to populate with a Milestone search (instead of doing whatever crazy logic it is now), it might actually work the way we want.

Posts 13141
Forum MVP
Mark Barnes | Forum Activity | Replied: Mon, Apr 29 2019 1:29 AM

Josiah:
If I'm reading a Tertulian defense and want to check another translation, I don't want to be taking to the Table of Contents of a work that doesn't even include the particular defense I'm reading...  It's bad enough the way all the DSS are indexed in parallel resources, but it shouldn't be hard to fix the Church Fathers references?  Right?

Josiah, we've been here before ;-).

The issue you raise is important to me, but it's not a problem with the references. Instead it's a performance issue.

When you click the parallel resources menu, Logos prepares a list of all resources that share that datatype (i.e. have the Tertullian milestones). It can do that fairly quickly. At that point, it could then check each of those resources to see whether they have that exact reference. That would be much slower, especially for Bible references when you might have hundreds of resources to check.

So instead of that, Logos add metadata to each resource that gives the approximate range of references that are included. These ranges can be checked very quickly and resources where the current reference isn't within their ranges are rejected.

This works very well for most resources. For example, a commentary on a single Bible book can be quickly eliminated from parallel resource lookups for other Bible books. But it doesn't work well for complex datatypes (like some Fathers), or resources that have lots of gaps in their milestones (like study Bibles).

There is a way of getting an accurate parallel resource list, and that's by clicking the plus button next on the tab bar. That will generate a new pane, with two columns. The large left-hand column will list recent resources, but the small right-hand column will list parallel resources using the slower, accurate method.

My own belief is that most computers are now quick enough to use the slower method without pain, and periodically plead with them to do so. But Faithlife either don't share that belief, or don't want to allocate resources to fixing the problem.

Posts 144
Josiah | Forum Activity | Replied: Mon, Apr 29 2019 7:51 AM

Mark Barnes:

Josiah, we've been here before ;-).

Haha, yes.

Mark Barnes:

The issue you raise is important to me, but it's not a problem with the references. Instead it's a performance issue.

When you click the parallel resources menu, Logos prepares a list of all resources that share that datatype (i.e. have the Tertullian milestones). It can do that fairly quickly. At that point, it could then check each of those resources to see whether they have that exact reference. That would be much slower, especially for Bible references when you might have hundreds of resources to check.

So instead of that, Logos add metadata to each resource that gives the approximate range of references that are included. These ranges can be checked very quickly and resources where the current reference isn't within their ranges are rejected.

I am not unsympathetic to this explanation, but it is not a hardware limitation.  Hardware is only taxed because of outdated design and algorithms.  It's a case of whether they want to overhaul the code or not.  This is, of course, non-trivial and there is always an issue to introduce new bugs, they must balance the time-cost investment vs. other features.  Generating new sales vs. keeping existing customers.  As I mentioned before it isn't monetized, it doesn't obviously sell new resources (though people might buy more translations if Parallel Resources worked better, it's a non-obvious connection).

Anyway, the Milestone search executes in milliseconds.  Heck, if they're worried about performance, make it a Cloud-only feature and crunch it server side.  But client-side shouldn't be an issue with proper indexing and algorithms.

Mark Barnes:

There is a way of getting an accurate parallel resource list, and that's by clicking the plus button next on the tab bar. That will generate a new pane, with two columns. The large left-hand column will list recent resources, but the small right-hand column will list parallel resources using the slower, accurate method.

I appreciate this, and do not want to appear uncooperative, but just in case other people are reading the thread and think this proposal is an acceptable solution, I am obligated to point out that it absolutely does not work correctly.

Here's the resource tab for the same Tertulian section as my example above:

As you can see, it is even worse!  And this is with probably 70% of my library hidden (I don't like clutter).  I can only imagine what it would look like for people who do not hide resources.

Posts 13141
Forum MVP
Mark Barnes | Forum Activity | Replied: Tue, Apr 30 2019 8:32 AM

Josiah:

I am not unsympathetic to this explanation, but it is not a hardware limitation.  Hardware is only taxed because of outdated design and algorithms.  It's a case of whether they want to overhaul the code or not.  This is, of course, non-trivial and there is always an issue to introduce new bugs, they must balance the time-cost investment vs. other features.  Generating new sales vs. keeping existing customers.  As I mentioned before it isn't monetized, it doesn't obviously sell new resources (though people might buy more translations if Parallel Resources worked better, it's a non-obvious connection).

Anyway, the Milestone search executes in milliseconds.  Heck, if they're worried about performance, make it a Cloud-only feature and crunch it server side.  But client-side shouldn't be an issue with proper indexing and algorithms.

I agree completely. It was (IMO) a poor original decision choice that made lookups expensive (originally there was no master lookup index for milestones). But whatever the original choices, it ought to be have been optimised by now.

Josiah:
I appreciate this, and do not want to appear uncooperative, but just in case other people are reading the thread and think this proposal is an acceptable solution, I am obligated to point out that it absolutely does not work correctly.

That's very strange. I've never seen that behaviour before. 

I tried it for the reference Tert., De res. 2, and it showed just two resources, as I'd expect. But then I tried it for Tert., De mon. 1 and it showed dozens of unrelated resources.

That's either a bug, or perhaps if there are no parallels it just shows random junk?

Posts 144
Josiah | Forum Activity | Replied: Tue, Apr 30 2019 9:08 AM

Thanks for your reply.  Has to be a bug, because I do have two legitimate translations of the letter.  So not a case of no parallels.

Posts 13141
Forum MVP
Mark Barnes | Forum Activity | Replied: Tue, Apr 30 2019 1:27 PM

Josiah:

Thanks for your reply.  Has to be a bug, because I do have two legitimate translations of the letter.  So not a case of no parallels.

What reference were you using?

Posts 144
Josiah | Forum Activity | Replied: Tue, Apr 30 2019 1:51 PM

Tert., De exch. cast. 6

Posts 13141
Forum MVP
Posts 144
Josiah | Forum Activity | Replied: Sat, May 4 2019 6:39 AM

Thank you!

Page 1 of 1 (16 items) | RSS