Clickworkers results

Results

Quality analysis

This has been primarily a proof-of-concept, and some analysis has been done to see whether the clickworkers technique produces useful data.

Are the collective inputs of clickworkers comparable to the traditional method of having an expert do all the work? For the crater marking task, we now have a detailed comparison of the results obtained by combining the work of several clickworkers to what an expert would have input. [600k PDF-format document]
You can look at an example of how the individual clicks of several clickworker’s crater-marking inputs are combined into a single consensus.
One science product that can be (re)produced directly from all your clicks is an age mapof Mars, based on crater counts. See details.
In March 2001, we informed the planetary science comminty of our results, so they can consider using it in the future. We presented a poster at the Lunar and Planetary Science Conference, March 2001. Here is a copy of our abstract, written in January 2001. [121k PDF-format document]

Full results compiled as of June 2001

The inputs of all clickworkers through early June 2001 have been compiled. They are being made available here for use in planetary science and for any other type of research. Note: The files are all huge tables of numbers, suitable for downloading and importing into a database and not at all interesting to browse.

Craters observed in Mars Orbiter Camera images [8300k text document]
Columns:
1. MOC image id
2. crater center x coordinate within image (pixels)
3. crater center y coordinate within image (pixels)
4. crater diameter (km)
Index of above file by image [592k text document]
Columns:
1. MOC image id
2. consensus number of craters (if 0, the image will not appear in the crater table)
3. number of clickworkers marking this image (use as a confidence measure)
Crater classifications in Mars Digital Image Mosaics from Viking Orbiter images (the PDS dataset used was VO1/VO2-M-VIS-5-DIM-V1.0) [548k text document]
Columns:
1. Crater id: internal clickworkers unique id
2. Crater id: Map Subquadrangle in which it appears
3. Crater id: internal id within Barlow catalog
4. Crater id: latitude
5. Crater id: longitude
6. Crater id: diameter
7. Crater classification votes: Number of clickworkers classifying crater as FRESH
8. Crater classification votes: Number of clickworkers classifying crater as DEGRADED
9. Crater classification votes: Number of clickworkers classifying crater as GHOST

FAQ | Privacy | Contact us