On Oct 30, 2014, at 8:47 AM, P.G. wrote: > Thanks for the early replies, here are more questions as some of you wanted > to get more details. > > What tools are available to create metadata?
define 'create metadata'. We generally call it 'cataloging' or 'annotation' in my field, as we don't 'create' anything -- we only document what already exists. > What tools are available to check the structure of metadata? It's a function of the metadata standard. If they're in XML, they might have an XML schema that you can test against. For FITS (the image file format, not the software for archiving) there's a website that you can give them a file, and they'll validate it for you. > What tools are available to check the accuracy of metadata? hehehehe. Oh, wait, you were probably serious. There are some things that you can validate easily (eg, if we know the size of an image, we can do simple consistency checks to make sure the size matches the height * width * bits/pixel) ... and there might be some statistical stuff that you can re-compute to validated (eg, mean, std.deviation, min, max, 10% and 90% thresholds). And we might be able to ensure that things are consistent (within tolerance) if they're specified in more than one manner (eg, if we know the pixel scale, and the width in pixels, and the image width in arcseconds, we can verify that (pixel width * pscale == angular width). But for the rest of it ... you're likely on your own. We were serving browse images from a telescope where person who created the JPEGs forgot to check a flag to see if the spacecraft was upside down. We didn't notice for months until I put together CGI for our website that would make a slideshow of images ... and the sun was spinning backwards. It was years before someone realized that another spacecraft's clock was drifting rather significantly, and that all of the times were bad. We've got other images where the pointing information is known to be bad and you have to use an external catalog to correct them (because they don't want to re-process all of the files, and risk other corruption). > Can't I just buy software that conforms to the Standard? You can ... but standards change, tools implement standards wrong, and there are lots of standards. > What is the standard metadata for images, audios and videos? Do you mean 'what are the standards' or 'what concepts do we record about the items in our collection'? There are a LOT of general categories of metadata (content, structure, archiving, context, etc.). Just for images alone, there are two major competing standards for content (EXIF and ITPC). Then there's Adobe's XMP, which allows you to put any metadata onto almost any type of file. For astronomy images, you also want VAMP. Structural metadata depends on the file format being used. Archival and administrative metadata is typically not attached to the files themselves, but is a function of the archival system being used. I don't deal with audio or videos, but I'm guessing that there are similar issues there. -Joe
