101 things I could/would do with all the images from the Granite Mountain Vault
16 March 2006 01:17

This actually comes later in the sequence of Thursday's conference, but I want to get it out so I can point some folks to it. Shane Hathaway gave a presentation about his work on a long-term storage system for all the images that will be generated from the microfilms in the Granite Mountain Records Vault. More details from that in a later post.

There are endless 'cool' (not to mention useful) things that could be done with the images in the vault if the delivery system built by the Church includes a web services API. I can imagine a number of 'meta-databases' growing up around this collection of images (a la del.icio.us and the web).

As one idea, a simple tagging database could be built that could associate any given set of tags with any set of images by user. If each image has its own permanent URL (I surely hope so!), then this kind of database would be trivial to implement, and almost wouldn't even require an API. Taking the idea a step further, I imagine a meta database that would allow me to tag and link to regions of images, as specified by X,Y coordinates. This would let me link to a specific line in, say, a census record.

I'm wrestling with what specific value a web services API for the image delivery system would provide, if each image had its own specific URL. There's a lot of power in simply being able to reliably link to any given resource on the web. Perhaps the real value would be in a web service that provided metadata details about each image (which microfilm it came from, the sequence number within the microfilm, which geographic locations are covered in the image, which time period is covered in the image, all the publication and citation information from the associated Family History Library Catalog entry, and so forth).

Some ideas I posted on the LDSOSS mailing list a while back included the following:

"one could build a service around this hypothetical API such that users could create RDF 'semantic-web' data associated with [or extracted from] each image, such as 'John Doe was born in Sussex, England in 1815', which would then be machine-readable, and much in the same way proofs can be deduced in a relational database based on the basic information stored in relations. Imagine being able to query 'show me all images for a John Doe born in Sussex, England within 5 years of 1815'."

But really, such could be done by simply linking to a permanent URL. Imagine if in parallel with the FamilySearch Indexing work going on to extract the literal text from these images, someone (like me?) built a database system/service that allowed a community to extract and associate semantic data (stored as RDF) with any image in the system. The church wouldn't need to develop this, all that would be needed would be permanent URLs for the images (but an API to grab all the metadata would be very helpful, too).

What kinds of things would you like to be able to do with an API for this system?

As a final thought, I really hope an RSS/Atom feed of new images coming online is made available. A customizable feed that lets me indicate I'm interested in these specific geographic regions and these time periods, and then gives me a custom link to a feed for images matching these criteria would be utopia over the next few years.