Tuesday, June 14, 2005

first thoughts on building a Flickr-like service at UC Berkeley #

On September 9, 2004, I posted my first picture on Flickr. Over the last nine months, working with my images on Flickr has become a central part of my online life. I went from being rather blasé about uploading my photos to changing the whole way I create, process, share, and think about digital imagery because of interacting with Flickr. I can definitely see how having a service like Flickr can also change a researcher's use of digital images. What would it take to Flickr or Flickr-like services available to the UC Berkeley community at a economically and intellectually sustainable service? Let me list some of the issues involved in figuring out the answers to this question: Does Flickr indeed offer a service that is more or less what researchers/teachers would use at UC Berkeley? What features would make it especially useful to our audience? Technically, how can one create a system with Flickr-like functionality? Buy the service from Flickr? Build a new software system? Buy/adapt existing software? (DSpace, Fedora, something else?) What are Flickr's competitors doing (like smugmug and Zoto) What services are available? (We can compare Flickr Services with zoto's api.) How to deal with providing reliable hosting? Flickr gives me 2GB/month for $25/year with as much storage I can get in with that bandwidth. This amount is significantly greater than I would expect to get for basic Linux hosting for that amount of money. The amount that flickr provides feels very comfortable to me as a power user: comfortably more than I am using. How much storage would our users want and need? How much can we provide and at what rate? There are a lot of questions and I will come back to them over time.   Posted by Raymond Yee on 6/14/05; 6:18:03 PM from the Educational Technology dept. Discuss

First thoughts and notes on DSpace #

Colleagues at the Museum Informatics Project and I are evaluating DSpace for possible use as a destination and source repository. We have an instantation of version 1.2.2 running and are trying first to understand what DSpace can and cannot do. I've uploaded one image into our test DSpace repository and have been poking through the documentation (including the DSpace Wiki and DSpace mailing lists. I will admit that my initial prejudices are towards wanting DSpace to do what Flickr is providing me. That expectation has made me rather grumpy about having to entire quite a bit of metadata just to upload one image. Now, I don't know whether I had to enter the metadata; perhaps I could have left it all blank. Maybe what is obligatory or optional is configurable by the system administrators to fit the intended use and policy behind the repository. The following are some interesting links that I have found so far. some introductory materials and guidelines for manager-types: DSpace conceptual diagram and service model questionnaire a tidbit to keep in mind from LessonsLearned - DSpace Wiki: "The persistent identifier for content is the single best selling point for DSpace when talking with faculty." there has been interest in integrating RefWorks with DSpace: SourceForge.net: dspace-tech After uploading one image manually (and finding the process to be tedious), I started wondering how to automate my interactions with DSpace. It seems that thought has been given to implementing WebServices with DSpace but that the work is very early stage: WebServiceForDSpace - DSpace Wiki and LightweightNetworkInterface - DSpace Wiki In fact, as I was sorting out what programmatic interfaces to DSpace exists, I found the following distillation helpful: NetworkInterfaces - DSpace Wiki records: Currently Available Interfaces OAI-PMH support already built in. This allows metadata harvesting currently of simple Dublin Core only. OAICat has a plug-in framework where additional 'crosswalks' can be added for qualified DC, METS etc. etc. SRW/U Web-services and RESTful interfaces for remote search (based on Z39.50, but simplified.) Currently a separate piece of software. Ideally, should be integrated into DSpace; the DSpace 'bridge' part should be part of the DSpace code, and then the SRW/U code could be included as a JAR. Ralph LeVan at OCLC has agreed that this is a good route forward; all we need now is someone with the time to do it! SSH command-line interface - mentioned for completeness Web UI - mentioned for completeness I did a cursory search for more information on the SRW/U interface but came up more or less empty-handed. I will take a closer look at the mention on the mailing list of the SRU/W interface and SRW/U (OCLC - Software) Dublin Core is the base schema in DSpace. There might be more support for multiple schemas forthcoming: MetadataSupport - DSpace Wiki When we are ready to set up handle for real in DSpace, we can come back to Frequently Asked Questions About the Handle System: 1.16 I'm in the process of setting up an instance of DSpace at my university and need a handle prefix for the DSpace handle server. What do I need to do to obtain one? The handle server is included in the DSpace distribution. Please follow the steps in the DSpace documentation which instruct you to run the SimpleSetup program and send the generated file to hdladmin@cnri.reston.va.us to receive your prefix (naming authority). In the email, along with the file please indicate that your request is for a DSpace implementation and include your full name and organization. Finally, it might be instructive to look at SourceForge.net: Detail: 1160997 - RSS Add-on and the list of other DspaceProjects - DSpace Wiki.   Posted by Raymond Yee on 6/14/05; 6:15:18 PM from the Web Technology dept. Discuss