Suggestion: Please Fix Topics and Interesting Words

Stein Dahl

Topics and Interesting Words are nifty looking - but I can't see all the words.

All of the words are on the screen but some of them are so small that I'm not able to see them.

I'd like to suggest adding an option to be able to view these 2 features - Topics and Interesting Words - differently. For instance, maybe the words could appear in a list format (which would be much more readable) - with the most important words toward the top of the list and the less important ones toward the bottom.

That would be extremely helpful to me - anyway.

Find more posts tagged with

Comments

tom

[Y]

Sean Boisen

Stein, Tom:

In fact we're working on alternative display for Topics that should be more readable. I'm not sure about Interesting Words, though i'd personally favor an alternative there as well.

Damian McGrath

https://community.logos.com/discussion/comment/97835#Comment_97835

In fact we're working on alternative display for Topics that should be more readable. I'm not sure about Interesting Words, though i'd personally favor an alternative there as well.

Please.

tom

https://community.logos.com/discussion/comment/97835#Comment_97835

Stein, Tom:

In fact we're working on alternative display for Topics that should be more readable. I'm not sure about Interesting Words, though i'd personally favor an alternative there as well.

Thanks Sean,

I believe the way Logos has decided to display topics and interesting words is a clear example of people saying that L4 is not user friendly - it is design for those computer programmer type of people.

The display looks cool, but I cannot make head or tales of what it is telling me. Do the different colors mean anything? Words are different sizes, what does that mean? Does the position of the words mean anything? Some words are printed out horizontally and others are printed out vertically, what does this mean? There are too many questions that the user must answer before he/she can use it.

Now if the user knows all of the answers to the above questions, is this word a different color than that word, or is it an optical illusion because of the colors of the words next to them? The same issue comes with the size of the font and the location of the words.

If a user has to go and find a key to be able to decipher the report, than the report has a bad design.

Ward Walker

https://community.logos.com/discussion/comment/97906#Comment_97906

I don't mind the present format, but agree with tom's line of thought--I'd assumed that the larger size text was triggered by some liguistic heuristic to draw the user to the more signficant word, and perhaps colors were linked to part of speech/TVM. Maybe if a mouse-over led to an explanatory pop-up?

Not being Greek/Hebrew/Aramaic savvy, I hadn't gone into any further depth of thought on that yet.

In one of my Layouts, I noticed that the font size for those reports got crazy small for me...they seem to hard-enforce an inch right/left margin, which becomes a limiting factor when restricting the PG to a modestly narrow (40%) column on my portrait-oriented 1200x1600 monitor. If I float the PG, the text becomes readable again...wish it would stop enforcing such a huge margin though; nothing else does that.

Rosie Perera

https://community.logos.com/discussion/comment/97934#Comment_97934

I don't mind the present format, but agree with tom's line of thought--I'd assumed that the larger size text was triggered by some liguistic heuristic to draw the user to the more signficant word, and perhaps colors were linked to part of speech/TVM.

I have no idea what the colors are for, but the relative sizes of the words have to do with frequency of occurrence in the passage.

Kevin A. Purcell

https://community.logos.com/discussion/comment/97952#Comment_97952

I have no idea what the colors are for, but the relative sizes of the words have to do with frequency of occurrence in the passage.

According to Morris Proctor at the Henderson NC Camp Logos the color and size are have to do with one thing. It makes it look neat and NOTHING else.

Mark Watson

https://community.logos.com/discussion/comment/98007#Comment_98007

I have no idea what the colors are for, but the relative sizes of the words have to do with frequency of occurrence in the passage.

According to Morris Proctor at the Henderson NC Camp Logos the color and size are have to do with one thing. It makes it look neat and NOTHING else.

But they do make an interesting bulletin or cover sheet for a group of handouts!

MJ. Smith

https://community.logos.com/discussion/comment/98020#Comment_98020

According to Morris Proctor at the Henderson NC Camp Logos the color and size are have to do with one thing. It makes it look neat and NOTHING else.

This surprises me. In Wordles, size indicates frequency. Color and orientation mean nothing. This would mean that Logos "plagarized" Jonathan Feinberg's presentation without retaining his meaning. That would be as silly as creating a pie chart in which the size of the slice meant nothing.[:O]

Mark Barnes

https://community.logos.com/discussion/comment/98007#Comment_98007

According to Morris Proctor at the Henderson NC Camp Logos the color and size are have to do with one thing. It makes it look neat and NOTHING else.

If he said that, Morris is wrong. The size has to do with relative importance (i.e. roughly speaking it's the ratio of the frequency of the word in this passage to the frequency in the whole Bible).

tom

https://community.logos.com/discussion/comment/98134#Comment_98134

According to Morris Proctor at the Henderson NC Camp Logos the color and size are have to do with one thing. It makes it look neat and NOTHING else.

If he said that, Morris is wrong. The size has to do with relative importance (i.e. roughly speaking it's the ratio of the frequency of the word in this passage to the frequency in the whole Bible).

Thanks Mark for this info. Now I know why this report made no sense to me (L3 or L4), and it comes from the fact on what I think is important. For me, those pesty little words like 'not' are very important.

Sean Boisen

https://community.logos.com/discussion/comment/98156#Comment_98156

Size definitely represents relative importance of the individual topics or words to this passage. Color, however, is essentially random art, as is orientation.

(For words, it's essentially frequency as previously mentioned. The computation and statistics are rather more complex for topics/concepts, because a dictionary article on a topic might include lots of references.)

tom

https://community.logos.com/discussion/comment/98273#Comment_98273

Size definitely represents relative importance of the individual topics or words to this passage. Color, however, is essentially random art, as is orientation.

(For words, it's essentially frequency as previously mentioned. The computation and statistics are rather more complex for topics/concepts, because a dictionary article on a topic might include lots of references.)

Thanks again Sean for clearing this up for us.

Can you please bring this question back to your design team, "If the users of L4 who know what they are doing (which apparently includes Morris Proctor) are having a hard time deciphering this display, how useful is this display? This question needs to be answered truthfully because every new user to L4 is going to go through this same process of trying to figure out what the report is trying to tell him/her.

MJ. Smith

https://community.logos.com/discussion/comment/98291#Comment_98291

This question needs to be answered truthfully because every new user to L4 is going to go through this same process of trying to figure out what the report is trying to tell him/her.

Er .. ah.. I guess I'm not an L4 user because I didn't go through the process, nor did my son (when I showed him a mal-formed one). Wordles are merely a variation on the tag clouds many of us are used to seeing. I agree that they are not entirely intuitive. The typical human mind looks for patterns even when there is none - so trying to add meaning to the colors makes perfect psychological sense.

Scott S

https://community.logos.com/discussion/comment/98273#Comment_98273

Color, however, is essentially random art, as is orientation.

I'm glad color is not used to convey information. Red-green color blind folks like myself, might have difficulty in some circumstances. We are 7-10% of the population.

Bob Pritchett

https://community.logos.com/discussion/comment/98291#Comment_98291

Can you please bring this question back to your design team, "If the users of L4 who know what they are doing (which apparently includes Morris Proctor) are having a hard time deciphering this display, how useful is this display? This question needs to be answered truthfully because every new user to L4 is going to go through this same process of trying to figure out what the report is trying to tell him/her.

The colors simply help distinguish the words. This is like a map of the US in which each state is colored differently to emphasize the boundaries. Purple states aren't different from the green states, unless of course there's a legend indicating that certain colors map to certain data points.

We're looking at adding a toggle between the "cloud" and a "list". One of the reasons I prefer the cloud, though, is that these aren't precise lists -- and even frequency stats aren't precise indicators of rank or importance. When we used a list in the past we'd get people saying "Why is this word on the list?" or "How could you think that word x was more important than word y in this passage?" The cloud intentionally (and gently, I think) obfuscates these things, and does what a statistically generated report should do: offers a number of interesting possibilities to explore, not a perfect ranked list of stats to imbue with authority.

tom

https://community.logos.com/discussion/comment/98424#Comment_98424

Bob Pritchett">

Can you please bring
this question back to your design team, "If the users of L4 who know what
they are doing (which apparently includes Morris Proctor)
are having a hard time deciphering this display, how useful is this display? This question needs to be answered truthfully
because every new user to L4 is going to go through this same process of trying
to figure out what the report is trying to tell him/her.

The colors simply help distinguish the words. This is like a
map of the US
in which each state is colored differently to emphasize the boundaries. Purple
states aren't different from the green states, unless of course there's a
legend indicating that certain colors map to certain data points.

We're looking at adding a toggle between the
"cloud" and a "list". One of the reasons I prefer the
cloud, though, is that these aren't precise lists -- and even frequency stats
aren't precise indicators of rank or importance. When we used a list in the past
we'd get people saying "Why is this word on the list?" or "How
could you think that word x was more important than word y in this
passage?" The cloud intentionally (and gently, I think) obfuscates these
things, and does what a statistically generated report should do: offers a
number of interesting possibilities to explore, not a perfect ranked list of
stats to imbue with authority.

Thanks Bob for responding to my post.

For the most part, I would say that I agree with you. Because stating where we agree would create a
very boring post, I am going to state where we disagree and why I disagree with
you.

You related the "cloud" to a map of the United States. I personally, as a user, see this cloud more
like a graph than a map. This being
said, I am going to go with your map image first. As you stated, colors are used in a map of
the United States
to point out the boundaries between states.
If a map has the same color for all of the states, then data (the state
boundaries) has been lost. Therefore, color
is not used on a map to simply to make it look pretty. Color is used, even on a simple map of the United States,
to express information.

We are so trained that when we are looking at something that
is expressing information (map, graph, etc. . .), we interpret colors as represent
data. Sometimes this data representation
is direct (green = growth), and sometimes this data representation is used in
an indirect fashion (as in your map of the United States example).

Like I said above, I view this “cloud” more as a graph than
a map. Maps are used to represent geographic
data (state boundaries, where I-5 starts and stops, etc. . .).
Graphs are used to represent statistical information, and this ‘cloud’
is attempting to represent statistical information.

As you stated above, “A statistically generated report
should offer a number of interesting possibilities to explore.” A couple of questions that a statistically
generated report can produce are, “Why were profits so low for the month of January? Why has the average temperate for the month
of February increased over the past ten years?”
A statistically generated report will offer a number of interesting
possibilities to explore only if the data it is representing is done in a clear
and precise manner. This is why your statement,
“The cloud intentionally (and gently, I think) obfuscates these things” struck
me as being odd. While I think “obfuscates”
is a very cool word that should be used more often, I do not think obfuscating
information is helpful. By definition, “to
obfuscate” something is to make something obscure or unclear, especially by
making it unnecessarily complicated (Encarta Dictionary). When things are obscure, unclear, or
unnecessarily complicated, people will not spend much time investigating what
the chart is trying to express because it takes too much work.

The difficulty as it relates to “interesting words” is how
does someone classify a word as being “interesting?” People are going to use different criteria to
classify a word (or phrase) as being “interesting.” For me personally, the word’s definition and
how does the author use the word can makes a word “interesting” much more than
how many times the word is used. To use
a couple of examples from this past Sunday’s RCL Gospel reading (Luke
13:31-35).

Here is the text for Luke 13:35 ἰδοὺ ἀφίεται ὑμῖν ὁ οἶκος ὑμῶν.
λέγω [δὲ] ὑμῖν, οὐ μὴ ἴδητέ με ἕως [ἥξει ὅτε] εἴπητε, Εὐλογημένος ὁ ἐρχόμενος ἐν
ὀνόματι κυρίου. Οὐ (not) is not very interesting when it is
by itself. It occurs 3757 times in the
LXX/NT. Μὴ (not) Is also not
very interesting when it is by itself.
This word occurs 4217 times in the LXX/NT. Now when these two words (οὐ μὴ ~ not not) occur
together, they become very interesting.

Backing up one verse, here is Luke 13:34 Ἰερουσαλὴμ
Ἰερουσαλήμ, ἡ ἀποκτείνουσα τοὺς προφήτας καὶ λιθοβολοῦσα τοὺς ἀπεσταλμένους πρὸς
αὐτήν, ποσάκις ἠθέλησα ἐπισυνάξαι τὰ τέκνα σου ὃν τρόπον ὄρνις τὴν ἑαυτῆς νοσσιὰν
ὑπὸ τὰς πτέρυγας, καὶ οὐκ ἠθελήσατε.
Here, Jesus is referring himself
as a ὄρνις (hen – a protective mothering figure). This word only occurs four times in the
LXX/NT. For me, this is very
interesting. As a matter of fact, this
is what I preached on yesterday.

Reading between your lines, I believe that you want L4 to aid
your customers in examining scripture. You
do not want L4 to create lists that are imbuing with authority simply because
L4 created the list. Anytime L4 creates
a list, a cloud, a thing-a-ma-jig, it is going to have authority because L4
created it. The best way to get around
the “imbuing with authority” issue is to allow the user to create an “algorithm”
on how he/she deems what words are interesting.
Because I think this is almost impossible to do, I believe the second
best way to get around the “imbuing with authority” issue is by giving us the
raw data on how often the word appears (in is form and lemma) in the LXX/NT, how
often the author uses the word in his book(s), how often the author uses the
word in the text that is being examined.
I believe this is the only way you will be able to avoid questions like,
“Why is this word on the list? How could
you think that word x was more important than word y in this passage? Why is the font for this word is larger than that
word?"

MJ. Smith

https://community.logos.com/discussion/comment/98424#Comment_98424

The cloud intentionally (and gently, I think) obfuscates these things, and does what a statistically generated report should do: offers a number of interesting possibilities to explore, not a perfect ranked list of stats to imbue with authority.

What I read between the lines of the various complaints about wordles is a desire for some of the text analytics functions. I'll admit to having read, with a wee bit of envy, the "infer" search for intertextual studies in a white paper (implemented in one of your competitors). The complaints also seem to circle around a misunderstanding of what is being conveyed. Better documentation on the feature might also quiet some complaints.

Bob Pritchett

https://community.logos.com/discussion/comment/98895#Comment_98895

Reading between your lines, I believe that you want L4 to aid
your customers in examining scripture. You
do not want L4 to create lists that are imbuing with authority simply because
L4 created the list. Anytime L4 creates
a list, a cloud, a thing-a-ma-jig, it is going to have authority because L4
created it. The best way to get around
the “imbuing with authority” issue is to allow the user to create an “algorithm”
on how he/she deems what words are interesting.
Because I think this is almost impossible to do, I believe the second
best way to get around the “imbuing with authority” issue is by giving us the
raw data on how often the word appears (in is form and lemma) in the LXX/NT, how
often the author uses the word in his book(s), how often the author uses the
word in the text that is being examined.

You are correct. Giving stats as part of "Important Words" -- in my opinion -- over emphasizes stats, when what makes a word important is more subtle, and there may not be any perfect algorithm for determining it. Again, why we use the Wordle -- it's an exploratory tool, not a statistics report. (The words are sized with stats, and pretty clever ones, but they still aren't perfect. We emphasize words whose frequency in this passage isn't necessarily high, but whose frequency in this passage is different than their normal frequency in other nearby passages.)

If you want stats on a word, you can run a Bible Word Study and get stats in the Bible, a graph of relative density in Bible books, occurrences in other corpuses, etc.

Do you really want that kind of data exposed here in Passage Guide? (Which we intend to be higher level than the drill-down in the Bible Word Study report.)

Bob Pritchett

https://community.logos.com/discussion/comment/99092#Comment_99092

I'll admit to having read, with a wee bit of envy, the "infer" search for intertextual studies in a white paper (implemented in one of your competitors).

The reason we don't have an INFER command at this point is because it seems like the kind of thing better "pre-run" against a massive corpus than run by each user on their own machine. And we have in fact done this; internally we've got scripts that run every NT verse against the whole OT, looking for similar matches. However, for the Bible we don't seem to get much interesting new stuff -- men and women steeped in the text have identified and cataloged many more (and with more discernment) cross references than a "5 of 6" word matching algorithm ever will. I'm not sure how INFER is superior to having the Treasury of Scripture Knowledge.

Now that type of cataloging hasn't been done between every text in the system; that's why we did the "fuzzy" search feature in Logos 3. (It's actually cooler than INFER, in my opinion, because it uses a two pass system that finds more potential matches at the word level before filtering for matches at the character level. So it finds things even with spelling and conjugation differences.)

There's a very special "super fuzzy" search embedded in http://bible.logos.com. If you enter a multi-word phrase there, you should see pretty cool fuzzy hits.

In the Logos 4 generation we're interested in doing even smarter identification of "interesting" related phrases, allusions, etc. across different texts. But finding word matches usually isn't enough, especially when using translated texts. So we're digging deep in the academic research in this area, and exploring ideas like "semantic distance," among others.

These queries will be very time and memory intensive, so it's more likely we'll pre-run them and serve up the best results through some Tool UI, or else offer them via a server callback. But the goal is to identify both tight similarities -- the same phrase, even if differently translated, or a Church father's loose quote of the text, or even a modern commentator's quote of a poem or hymn lyric -- as well as loose "thematic" similarities, like two paragraphs in two different commentaries that are making the same point about a passage.

MJ. Smith

https://community.logos.com/discussion/comment/101969#Comment_101969

Fantastic! My own curiosity is about similarities in early translations - LXX, Vulgate, Peshitta, Old Gothic etc. that are not in the original Greek or Hebrew. Sounds like we'll get there in my lifetime.[8-|]

Kevin A. Purcell

https://community.logos.com/discussion/comment/98273#Comment_98273

Size definitely represents relative importance of the individual topics or words to this passage. Color, however, is essentially random art, as is orientation.

I may have misunderstood what he was saying at the training. He may have meant just the colors when he said it did not have any significance. But I know that more than one person there thought he meant the color and size so you guys at Logos may want to make this clear and make sure he makes it clear too. He does a great job otherwise. And I am using Logos more and more. Can't wait to get my favorite tools in there.

tom

https://community.logos.com/discussion/comment/101962#Comment_101962

Do you really want that kind of data exposed here in Passage Guide? (Which we intend to be higher level than the drill-down in the Bible Word Study report.)

Hi Bob, and I am sorry for not responding to your post sooner. I have been extremely busy because of Lent.

Again, anything Logos produces is going to carry weight, and it will lead a user in one direction or another. There is no way around it. When the information is not clear, people ask question, and this is why you received the questions like,

"Why is this word on the list?" or "How could you think that word x was more important than word y in this passage?"

I do believe MJ. is correct when she said,

"The complaints also seem to circle around a misunderstanding of what is being conveyed. Better documentation on the feature might also quiet some complaints."

I think a link under the Wordle that will open up another window that explains what the user is seeing would be very helpful.

To answer your question, yes I would. This is because I enjoy doing word studies.

MJ. Smith

https://community.logos.com/discussion/comment/101969#Comment_101969

However, for the Bible we don't seem to get much interesting new stuff -- men and women steeped in the text have identified and cataloged many more (and with more discernment) cross references than a "5 of 6" word matching algorithm ever will.

The place that I would actual want to use the infer-style/fuzzy-style search is not Bible against Bible but Prayer Book against Bible.For example, of the 4 Jewish siddur that I have, only one makes any effort to identify short quotations or near-quotations of Biblical text.

Michael F Storz

https://community.logos.com/discussion/comment/97835#Comment_97835

I would appreciate that too!

MJ. Smith

https://community.logos.com/discussion/comment/127790#Comment_127790

Since this topic has been resurrected, I'd like to state that my thoughts on this topic are changing. We are getting more questions on the forums about text analytics. The real question then becomes - do the users want Logos to do the analytics and present the data to the user or does the user wish to do their one analytics. I would suggest that Logos resources are sufficiently biased to the "North American Protestant canon" and that early translations are so sparsely represented, that end user analytics are necessary for a significant group of users. However, it is also true that the largest base of users haven't the background or interest to do the work themselves and Logos would fill a need by providing some "predigested" materials.