Some questions about AMC and a modification I would like to make
After my first trial of AMC with ipads (and other tablets), I have some questions moving forward. If you remember my first post, I am trying to find something that will work with large multisection exams (500-1000). I also have a specific format that I like to create exams with. So, here are my questions.
1. In my last trial, I put a grading grid next to each question number. Thus, when the instructor graded, they checked the score they wanted to assign, and then the software (eventually - see my other post) picked up the scores. However, this was a junior level class, and each question needed a specific amount of space. Thus, I created empty pages (except for the barcode and alignment circles) to use for the back. The reason for doing this is that I wanted students to be able to continue work on the back side, and so the back side needed to be in the scan. My question is: what would happen if I did not do this? That is, if the back side did not have barcodes or alignment marks? I suspect that the code would mark each one of these as unreadable, or unrecognizable (or whatever it says). Is that what happens? When the student's annotated papers are created, will these "back sides" be put in the pdf in the correct place? Is there a way I can say, for example, "don't scan the even pages" or something like that?
2. In my last trial, since I had a grading grid for each problem and one problem per page, I created a unique exam for each student (which really had the same info, but the barcodes were unique). As I understand it, this is necessary in order for AMC to work correctly. I think I understand why, but if I am going to use AMC for 1000 exams, this could be an onerous restriction. Usually we only have two forms that we use. I realize I cannot use AMC in this way normally. However, when I scan exams in, I have a way to tie exam number to students directly. In fact, I can pass one file per student to the system, if needed. Is there a way to modify AMC to get it to do what I want? How difficult would this be? I am more comfortable in python than perl, but can get around if need be.
I think I have other questions, but will wait for the responses for these questions before I proceed.
RE: Some questions about AMC and a modification I would like to make - Added by Alexis Bienvenüe about 8 years ago
- You can omit barcodes and alignment marks on pages where there is nothing for AMC to read (no boxes for students, no boxes from codes, no student name zone). When you feed AMC (automatic data capture) with pages without barcodes and alignment marks, the page is not recognized by AMC, and AMC will tell you about it (these scans won't be considered by AMC). There will be no problem with the marks. These scans won't be included in the annotated papers, however.
- You are right. In fact, AMC splits all the multi-page scans files to get one page per file, and then processes all these pages with as many processes as cores/cpus, in parallel. That's why AMC does not know which pages can be considered as being together. Perhaps one solution is to make a project per page (per problem), and so ask for the student ID (and/or name) on each page. Not very satisfactory but this does not need to modify AMC.
To modify AMC for your needs, we have first to process all pages sequentially (forcing a single parallel process in the queue), and then use a new copy number only when reading the first page of an exam. This could be implemented, but this is very sensitive to the process quality: if a first page scan is not recognized, this can break all the association of pages from a single exam. There is also a problem when needing to scan another time a given page (when there has been problems the first time). In fact, I couldn't imagine a safe workflow for AMC to work with multi-page photocopied exams, and that's why this is not (yet) allowed in AMC.
RE: Some questions about AMC and a modification I would like to make - Added by Davide G.M. Salvetti about 8 years ago
What about allowing the user to add not recognized pages to a manually selected exam copy with a manually selected page id?
As AMC would not be able to automatically recognize the page if scanned again this is perhaps suboptimal, but it would allow not recognized scans which don't need to be recognized by AMC to be manually added to annotated copies nevertheless.
It might not be useful for exams with a large number of copies, but I do have a case for it with open answer questions, where sometimes students need to add additional free sheets if and when they fail to make good use of the alloted space on the exam paper.
RE: Some questions about AMC and a modification I would like to make - Added by S. McKay about 8 years ago
Thanks for the replies. I guess I did not know about the reason for the unique exam copy. Being able to process exams in parallel might be useful. I"ll play around with that.
If I find I don't need that, however, I am inclined to change AMC so that it processes one exam at a time. You are correct in that if it fails to process the first page correctly, then it will cause problems. However, (I think) it should not cause the problem you are thinking about. If each exam is processed before starting the next, then the pages of that exam are known to be with each other, and perhaps we could assign some unknown id to keep track of them until the front page is fixed manually? I am just thinking about possibilites here.
Another thought I had would be to input the files with exam numbers as part of the name (that is easy for me to do, currently), and then use that to track student exams when the files are split up. This would allow parallel processing, but give a "fail safe" if the front page is not read correctly. I realize this is different from other ways people do this. I am wondering how hard this would be to modify.
As for the blank pages, this seems to be problematic. It seems as though the blank pages showing up as errors would make it harder to find the actual scanning errors. For now, I will "create" blank pages that AMC understands, but I wonder if there is an easy way to mark all "even" pages as not needing processing while keeping them in the annotated exam files.
Thanks for helping me think this through,