I purchased L4 Gold this week, (hello overtime) which means I've got some new toys to play with.
I do not understand what the "Cluster Graph" is for.
Would someone please kindly explain it to me? If/Why you use it?
Much appreciated!
It compares how close each version is to each other for that verse.
Personally I don't find it all that helpful. It's primary usefulness is to see if there is a cluster where translations are close to each other in the way they translate that verse.
EDIT: So basically the graph you posted basically tells you that those versions use different words and are not similar.
I'm glad you asked. I don't understand, either, since I don't know what the X and Y lines are measuring. I'd love to know more about a lot of the tools. I'm hoping more docs will come out as the push to get fixes out the door dies down! I do that all day so I know what it's like. [:S]
So essentially it's telling you that in the verses you display that the CEV is closer in wording to the NLT than it is to the ESV or KJV. It means that the KJV represents some usage that is apparently shared by the KJV and CEV. Looking at how the NLT seems to be just a bit further away from all the others, than the others appear to be from each other that the NLT is "significantly different" in word choice. Of course we would expect the KJV to be significantly different from a modern version and hence you'll note that it also visually appears to be distinct from the others.
[Y] Yall sure are a helpful bunch! TY!
I don't know what the X and Y lines are measuring
Bingo. Kaye hit the nail on the head. The scales of each Cluster Graph vary and so the graphic is totally relative and of minimal use. For example, look at what happens when I add Lexham to the example above ...
What previously appeared to be a rather loose cluster (KJV, ESV, CEV, NLT), is now much tighter compared to Lexham. Again, a relativistic, graphical representation. No absolute scales. No idea as to what the X, Y, or Z axis represent (if anything).
Also, I cannot figure out the significance of the dynamics of the graph as the dots move around the 3D screen. Presumably, we are looking at the cluster from different vantage points (left click on the graph to start/stop the motion). You do not seem to be able to control the vantage point, however. And, again, without knowing what the X, Y, or Z axis represent or their scale, it becomes a totally relative representation. Perhaps the dynamics are included just so you can get a well laid out PowerPoint slide?
More significant than the graphics, however, is the fade-in, fade-out word count comparisons. Again, I do not see that the User has any control over which comparison is viewed except to click through until he/she sees the one that interests. I would think that a non-fading, separate table showing the complete, word count comparison matrix might be more useful.
My guess is that this feature is still a work in progress and that we will see this become far more useful/informative as the L4 team fleshes it out. For now, it is rather limited and awkward. Patience ... the L4 developers are pretty clever and have shown themselves to be several steps ahead of us on many occassions!
No absolute scales. No idea as to what the X, Y, or Z axis represent (if anything).
The only represent a relative spacial distance in word choices in the selected pericope among the represented texts.
The XYZ axis are only important in that they show a greater distance in difference between translation 1, 2 and 3, etc.
It's not the best card in L4's hand ... I think the programmers were trying out some ideas (which is fine) and it got left in, but it doesn't add much.
If you check out IBM's ManyEyes, you'll see many examples of visualization that are "just 'cause I can" ... however, over time one learns to "read" these visualizations just as you learned to "read" traffic signs and pie charts. It's like avant garde music, you've got to learn to hear again.
I do not understand what the "Cluster Graph" is for. Would someone please kindly explain it to me? If/Why you use it?
I would add to this question the question, 'what do direction, distance, and sequence signify?' In other words, I get the basic idea of the chart, that different versions are closer in translation if the dots are closer together; but I don't understand if the different distances and directions mean anything.
???
I would add to this question the question, 'what do direction, distance, and sequence signify? ... but I don't understand if the different distances and directions mean anything.'
...
but I don't understand if the different distances and directions mean anything.'
The whole Idea of the cluster graph is to gain a very fast, very visual conception of how closely translations are worded. Not it is not likely to be the most used graph you have access to but it does have it's merits. If you take five or so modern translations, say: NASB, ESV, RSV, NIV, TNIV and HCSB on a passage and you notice that the NASB and ESV are farther apart than you'd expect but that the NASB is clustering closer to the NIV/TNIV you might consider how the translation philosophies have created such a move. It may even help when certain words can be translated quite differently depending on a host of factors. These graphs in some of those more thorny passages might point you toward which translations to provide a little extra attention to. And there lies the strength and purpose of the graph so I'll say it again.
By immediately providing a visual concept of the difference in translations you are pointed toward which translation(s) should recieve a little extra attention in your studies.
"just 'cause I can"
Haha - love it! MJ may have just stumbled onto the true answer for the existence of this graph!
One other theory might be that the Cluster Graph has been slipped into the program in order to sort out the left-side-of-the-brain types from the right-side-of-the-brain types. Those who need an absolute scale to derive meaning = left side. And those who "feel" that it is meaningful just as it is (relatively speaking, that is) = right side.
I am not exactly sure why "they" would do this, but I am sure that it is all part of the conspiracy ...
I went to a Logos demo once at an Iron Sharpens Iron conference, and if I remember correctly, they explained what the X and Y axes stood for. One of them (don't remember which) measures how closely the meaning of each word in the passage matches the meaning in the original language, and the other shows whether the entire passage was translated litterally or more as a paraphrasing. I don't know how much my explanation helps, because (1) this kind of measuring seems rather subjective and (2) I can't remember which axis is which nor which direction is which.
Maybe these words from the ol' help file will help,
Choose the 2D or 3D display. • With 2D, the chart shows only how close or far apart the selected Bible versions are to each other based on word usage. Bibles that use similar wording will be clustered closely together; Bibles that use different wording will be spaced farther apart. • With 3D, the results are animated, randomly displaying a line between two of the resources, with the number of words in common and the percentage difference visible in the lower-left corner. Click in the display to start the animation. It will change every few seconds to show the difference between another two Bibles, until you click in the graph to stop it. Click again to restart the animation. 5. Clicking on a Bible label or dot in the graph will open Text Comparison with that Bible listed first.
From how I read this help file, the first Bible listed becomes the base or standard for comparison, with the rest of the Bible versions listed. So the cluster graph is a visual comparison for the text comparison feature.
Hope this helps.
I think that I am going to resurrect this thread.
Mainly because I find that most of the more learned answers are as vague as the application of its usage. As to the X, Y, and Z axis, well if you can grasp a third dimensional drawing then you got it, if not then it’s still: nothing. Because you never see the actual movement of the "Z".
So is the KJV closer to the original Language than the NET, but then the LEX has a better grasp on translation than the CEV which beats out the TNIV which had a close running next to the RSV. But then I clicked the NASB and the focal point of the X, Y, Z rotated to display the ESV was more politically correct than the NIV.---HUNH?
So, perhaps we can get some better responses as to how this application, that we have had for a spell can be better applied. I got the add-in moons ago for other purposes but I got to admit, I am drawn to this just to watch the rotational shift…
Thanks.
[hopefully, before the MS.Manners comes streaming through the door; I am not attempting to be rude or disrespectful, but sometimes ignorance can be bliss.]
before the MS.Manners comes streaming through the door;
This time you don't need Ms. Manners' suggestions for which I thank you.[;)]
One of the purposes of a cluster graph when applied to natural language is to force the viewer to not read too much into it. Consider the factors that go into determining the similarity/difference between two texts:
Note, I still haven't looked at phrases, clauses, sentences etc. which have even greater complexity as one tries to determine if two sentences share a deep structure or are fundamentally different in their understanding of the sentence. Now think of the heuristics that assign values and weights to each of the elements - there is no exact measurement, only a value that represents the intuition of some computational linguist that generates values that "feel right" or come close to the values assigned by hand coders for translations in a test/training set. Jurafsky & Martin's Speech and Language Processing is a common textbook that will give you typical heuristics, algorithms, examples ... The rationale for providing the results visually on undefined axes is specifically to avoid having the user read exactness into a value that is anything but.
It sounds to me as if you have a pretty good grasp of how to read the cluster graph. And I suspect you have the good sense to have no inclination to want to create one.[:D]
Unfortunately, this is completely incorrect. (I'll follow up to make sure that Logos presenters aren't disseminating this type of information.)
I think that I am going to resurrect this thread. Mainly because I find that most of the more learned answers are as vague as the application of its usage. As to the X, Y, and Z axis, well if you can grasp a third dimensional drawing then you got it, if not then it’s still: nothing. Because you never see the actual movement of the "Z". So is the KJV closer to the original Language than the NET, but then the LEX has a better grasp on translation than the CEV which beats out the TNIV which had a close running next to the RSV. But then I clicked the NASB and the focal point of the X, Y, Z rotated to display the ESV was more politically correct than the NIV.---HUNH?
The X and Y (and Z) axes have no intrinsic "meaning" (such as literalness or translation philosophy, or anything else). (As fun as it would be to have X mean readability and Y mean heresy, the cluster graph doesn't work that way.)
Bibles that are close together in the graph use similar words. Bibles that are far apart in the graph use dissimilar words. "Words" are what you would normally consider words in English, but capitalization is ignored, and stemming is used to group words with the same root (e.g., "believes" and "believing") together.
This is much clearer in Libronix3 (my underlining):
DescriptionThe report displays a graph that shows the relative similarity of multiple Bible versions. The similarity is computed based on differences in word usage between the versions. The distance of a point along the horizontal or vertical axis has no significance; the only important measurement is the distance between two points on the graph, which is proportional to the difference between those two versions. Bibles that use similar wording will be grouped closely together; Bibles that have different wording will be spaced far apart.
I also like the L3 version since you can easily move between languages, if the passage supports more than 1 language.
Plus you can adjust the stemming depending on how tight you want the comparison.
Plus the screen includes instructions, a nice luxury for those who didn't program the screen.
My apology's Bradley, I was truely being facetious. It was mentioned and I ran with it, again My Apology.
This I grasp immensely, it goes without saying, although I think some read into it more than what is really there. I tried using just the greek text and the results were quite different that what I would have imagine, especially with certain passages[I got a pic, I will load it shortly]. The part that be-dazzled me the most was in the knowing that certain scholars were used on many the same translation projects, yet the results were different for those translations….it would make one think that they would almost 'stack' themsleves....
Dmb, yes I do like the L3 version better, maybe it could revert back.....
MJ- oohh the greatest of ease of how we use and take for granted nano-technology....
sorry for the second post. but notice the locations of the same manuscripts with the same text but one is 2D and the other 3D....[disregard the LXX, I was attempting to throw it off]
for a really kewl picture show type in the text and run thru the division of Jn 3
