Special thanks to Julie over at strangelibrarian.org for prompting this discussion of how Reference Extract fits into the participatory librarianship umbrella.
For more information on Scapes: https://davidlankes.org/rdlankes/blog/?p=460
8 Replies to “Reference Extract and Participatory Librarianship”
Comments are closed.
I think this is excellent. On the surface, RefEx (as i’m now calling it) can easily be misconstrued how a lot of the media has picked it up- librarians taking on google. bc when ppl think of search, they think of google. and anyone who tries to do something different w/ search is automatically assumed to be trying to do it better than google. Like we can’t exist in peace simultaneously.
But it’s not about that. It’s about RefEx being the first building block to the new world. The world where libraries are at the center of all learning conversations, everywhere. It’s a new world where i’m certainly excited to be. 🙂
Thanks for taking the time to make the video and offer that explanation of how RefEx links to participatory librarianship and Scapes. What I am wondering about is what sort of systems we have or we are capable of building that will structure the data in Scapes. With the example you offered for Scapes, there are songs, albums, movies, and chat transcripts that have to be linked in some way that is based on standards. The data about those objects has to be based on standards, the conversations about those objects would need to be standardized, the connections between the objects and conversations would need to be standardized. But how? In a world where metadata schemas about and clash, is there a system that can be applied here? Will a new system need to be devised?
Great questions, and one that we are churning on a bit here. So far this is where we are. Scapes needs to take advantage of existing metadata standards, absolutely. When you drag an object into a Scape you should be able to know a lot about that object already. Maybe it has some metadata schema attached (Like MARC, or Dublin Core, or even a more file format’ish one like ID3 tags for MP3 files, or XIF for images). Where possible Scapes should be able to take advantage of structures it finds.
The real question comes in the standard way of capturing the relationships that the user builds (including when not capturing traditional artifacts and document-like things, but ideas and agreements…but I digress). Here we want to be very flexible to both avoid asking too much of the user, and to get at the true point which is meaning comes not from metadata describing objects, but the relationships of these objects to each other as defined by the user. Right now, that is probably RDF. You have an object, what it is related to, and how they are related. In the current Scapes example there are three types of relations present:
1. a simple link connecting objects,
2. “Is a Version Of” and
3. “Another Broadway Remake”
Thee obviously come from the user describing the relationship. One way to do this would be to try and define every possible relationship type there is. Folks in the ontology and OWL world have tried that. It seems to me that this is the wrong path. Simple linking, with the user adding tags to these links (useful for humans, but of limited computational use), may be plenty. After all, look what Google has done with an amazingly simple model of WEB PAGE -points to -> WEB PAGE. There is really only one relationship “points to.” My hypothesis is that keeping the relationship types simple, and allowing for aggregation and the equivalent of link tagging, will be amazingly powerful. This is, however, an empirical question. I would like to start simple and add complexity as necessary, rather than trying to think through every possible use/ontology/process. That is what killed the NISO Virtual Reference Exchange protocol. Let’s not take out our metadata and standards hammer out too early on this one.
Agreed: a deep debate over or investigation into metadata at this point would likely be a black hole. I was trying to envision what Scapes would look like to the user when hundreds of thousands or millions of “conversations” or scapes were in the system and what would be the joins between data. With the Gwen Stefani example, what would it look like if you build off it to include thousands of related conversations: Gwen Stefani was originally in a sorta 3rd wave (and the number of waves is highly debatable) ska band whose range of musical styles it employed and drew on expanded. So maybe you might find someone wanting to establish links to all those styles and traditions, which then connects up to other bands and songs, etc., etc. Once you’ve got this rich, yet tangled network of connections, what is the ideal way to present nodes to the user? Will displays that are essential visual browsers be the ideal way for users to navigate interconnected nodes? Is the node/hub/spoke display enough or would you also want a way to explore via textual/faceted displays? I don’t know the answers to any of these questions but I think they are interesting ones that when explored might point the project along different paths.
Another question: would it help to envision Scapes as some sort of personal information management system that allows a user to share or interconnect his/her Scapes (or just selected nodes in his/her Scapes) with Scapes that others have created?
YES YES! To be honest the only part of the sort of “meta-scapes” interface we’ve thought about is Reference Extract to find them. However, you are absolutely right, that we need a way to sort through these (maybe Flickr-like?) and the personal information management system is great.