Moon Zoo Forum

Resources => Moon Zoo Team Area => : jules May 04, 2010, 09:31:20 PM

: jules May 04, 2010, 09:31:20 PM
Moderators to list a summary of comments about the Moon Zoo user interface here.
: Problem with favourites map
: Geoff May 10, 2010, 09:19:10 PM
I've saved some favourites from the south pole area and they are off the map - see screenshot:

: Arfon May 10, 2010, 09:59:08 PM
Hi Geoff, yes that's a problem with Google maps unfortunately.  They don't have complete coverage of the lunar surface and so sometimes you get a pin in no-man's land!  There's an update coming for Google Maps (Moon) edition soon I think which should fix this.

: Thomas J May 11, 2010, 08:00:26 AM
I'm just bumping in to get this thread in my replies.  :)
: jules May 16, 2010, 11:07:11 AM
It might be nice to have a field for comments and observations on EACH picture we examine.  That way we could quickly be able to reference unusual features not listed in the controls, or else further elaborate on the buttons we have used.
: jules May 27, 2010, 05:40:09 PM
These are my first impressions/comments on the Moon Zoo software. I realize you probably are already aware of many of these issues, but it's easiest to list all of them together.

Boulder Wars
There are lots of issues with Boulder Wars:
  • What if both pictures have nearly equal numbers of boulders?
  • What if a picture just has a few tiny boulders? Should you still select it for "Most boulders"?
  • If you make a mistake there's no back button
  • There's no favorite button
  • An illumination guide might be helpful

The data from Boulder Wars doesn't seem very valuable to me. I can see the appeal of having a binary decision that is less disputable but you're losing so much data by simplifying so much.

How big are the boulders? Are there patches or are they all over?

In my opinion it would be better to just have a single Survey tool with boulder capability.

For extremely large boulders you could have a single boulder tool just like the crater tool.

Boulder Region Input Tool
1. Select the boulder region by drawing around it or maybe by painting the region
2. After selecting the region, the user is prompted to draw a circle around the smallest boulder in the field and then the largest.

This approach isn't too difficult for the user and also gives an idea of the size range of boulders in the field. Of course, representing complex regions in your database will be difficult. You might just go with an ellipse tool, square tool, or polygon tool.

Better Crater Workflow
It seems like it would be useful to have additional crater types:old/soft, lumpy,new,etc. I think a better workflow would be to select crater types after you select all your craters. I would remove craters from the "Interesting Feature" tool and put those options all together.

1. User selects all crater sizes (the same way it's currently done)
2. User clicks "Done selecting Craters"
3. View changes to a list of craters with a cropped view of each crater (this way the orange circle isn't overlapping your crater)
4. For each crater, user selects among
  • Clear Edges (New)
  • Smooth (Old)
  • Lumpy Edges
  • Bench/Mound/Flat
  • Fresh White
  • Elongated Pit
Within this list the user can also select "Bouldery" options

You might also put crater age on a scale, so you could mark really faded (old) craters, somewhat faded, newish, sharp edges (very new).

When dragging across this scale example pictures would be displayed to the user.

An option to set all craters to a certain type might be nice as well to save time.

(Hopefully this description made sense, if not I can try to give you a more visual mockup)

Policy Confusions
  • How big should a fresh white crater be for the user to mark it?
  • Examples of Crater Chains don't make it clear how to identify them
  • What exactly is interesting? You're going to get way too many false-positives for interesting features

Other Issues
Some images are static or have bad exposure. There's no way to mark them. I saw the forum for users to copy images to, but most users aren't going to dig through the forums, find that thread, and then actually track all their static images. You should really just have a "Bad Image" button that lets you select "Static" or "Bad Exposure"

The Application is tiny. I'm assuming the images are scaled down; it would be much better to be able to resize or full-screen the application to take advantage of your full monitor size.

Hopefully these comments are useful. Let me know if you want clarification or suggestions on a particular aspect of the software.

: jules May 27, 2010, 05:42:15 PM
I'd say that the Boulder Wars employs a rather beautiful algorithm to weed out nice images from bad images.
You are assuming eac image is displayed once, which is not the case. I am thinking that on a database level, each image has a table where its "boulderiness" is stored related to other images it might have paired against in Boulder Wars. By doing a multidimensional sort, you can easily sort the images based on user input.
Let's assume you have 10 images; each of them is compared with the other 9, and since each image has its own table, you will have a total of 81 comparisons. But that's for a single user going through all pairs. Now let's assume you have 1000 users doing that; all of a sudden you have 9000 comparisons for 1 image against the other 9.
By doing a statistical analysis on users' inputs, you will be able to use a Gaussian Curve algorithm and weed out extremes, leaving the most probable values and using them as such (since there are only 3 possible options, I think it's not even necessary to apply such an algorithm). This will give you a "boulderness" factor for each image and then you will be able to precisely sort them.
Of course, in reality the algorithm would be much more complex, since one image might be part of another, and researchers will need to take this into consideration; also, the number of images is enormous, so I'd expect some heavy computing power standing behind this whole thing. But in the end, statistical analysis would give you very accurate results.
The greater the number of inputs, the greater the accuracy.

I agree with the "Bad image" feedback button.

I don't think the images are scaled down, but rather using three levels of zoom and being cropped to fit that resolution.
Generally speaking, most of what's happening here will feed a statistical analysis algorithm, so don't worry if you miss something, in the end it's the average that matters.
: jules June 01, 2010, 09:08:49 PM

This isn't a crater question, specifically, but... while doing Crater Survey, and after 20 or so images had been submitted, it went to bring up the next image and crashed.  It says "Loading Error 500: UrlNotFound".  I tried clicking on Crater Survey to reboot the process, but same thing.  I've noticed, also, that all my Moon Zoo tabs (I usually have three going at a time) tend to freeze several times a day, even though none of my other tabs do, and I have to shut them down and open new ones to start over.  Are these problems being experienced by anyone else here, or is it yet another painful display of my computer's overwhelming lameness? 

I swear, when I can finally afford a decent computer I'm going to enjoy the heck out of taking this outside and putting five or six bullets in it.
I've noticed that Moon Zoo, especially My Moon Zoo, freezes if left open for a while too.
I'll copy this over to the Team Area.
: jules June 05, 2010, 10:04:21 PM
I want an arrow, so I can point to a crater.
There are getting to be a lot of
"It is 2" left of centre, going up to the top corner. near the degraded crater"
"I can't see it"

eg. Galaxy Zoo has the purple triangle for marking objects.
: jules September 12, 2010, 04:30:18 PM
Some recent comments here:
and here: