The Social Side of Reflections proposal lays out the framework for our pilot effort to learn how much users of Minnesota Reflections will take advantage of a user commentary system. This page tracks the evolution of that proposal as we move toward a real breathing project. Please feel free to leave comments below.

Meetings

On 071105 JTB and PS met with EFC and discussed the technical side of the project. See Server Site and Registration below.

On 071121 PS and MB met with EFC to discuss Server Site and maintenance issues.

Server Site and Software Settled

On 071105 the server tempest.minitex.umn.edu was made available by MINITEX. It will be plenty powerful enough to run the PmWiki site that we agreed will form the engine for this project.

On 071121 the last kinks were worked out in our new domain name: http://views.mndigital.org. The single URL on the Reflections side will now be used to reach PmWiki which will determine the record in question via the referrer.

Registration

An extended discussion on 071105 led to the conclusion that simple is best. When considering the array of social software out in the real world we realized that most (blogs, slashdot, wikipedia) do not require registration to participate as a commenter. If we require any sort of registration we are probably limiting our pool of participants. Furthermore, these other tools (with the prominent exception of wikipedia) do not allow editing of comments already posted. Typically in a "comment" system, once a comment is posted the only alternative for the user to correct prior posting is to make a new posting. We thought that simple system would be good enough for this project. Remember, our goal is to encourage participation and keep the barriers low.

This amounts to a departure from the Project Description in the proposal. Given the way this simplifies both the implementation task and the user interaction, we felt the departure was warranted.

The result is that the SSR will not require any user registration and will not allow regular users to edit any prior postings. Note that we will still make a (post-posting) attempt to gather some contact information (email address) from users and that system administrators (and this can be anyone we designate, not just the consultant) will be able to edit any and all comments left on the site.

Progress Report 1

I had a wonderful conversation with Paul and Michael on 071121. We worked out a few (very minor) kinks in the setup of the server at MINITEX and I introduced them to PmWiki and the potential maintenance tasks in their future. We agreed to meet again, probably next month before the SSR goes into production.

The DLDL team made one change critical for progress when they added a small invisible button to the item view pages in Reflections. When the project goes live this invisible button will be made visible to users.

An "alpha" version of the Social Side of Reflections is now available (071126) if you click on this invisible button. It is a 10x10 pixel area just below the "Loc" in "Local Identifier" at the bottom of the description on any item view page. While this is not ready for production yet (in particular, we are still missing a link to the MDL id number on comment records), it is plenty ready for playful use. I'd be interested in any feedback you or others in the MDL effort might have before moving this too much further along.

Even though the contracting took a few weeks longer than we had hoped, I'm feeling pretty good about the timeline as it stands in the contract. That give us December to refine the SSR and aims for a public release in January.

MDL ID not reliable

Email with JR and JB (071126-27) reveals that the MDL id has not been assigned reliably. Some identifiers may not be unique and others do not conform to the structure we allowed for the identifier.

The identifier was to be a unique string composed of a lowercase alpha prefix (representing scanning center or site) plus a set of digits (representing, more or less, each item) plus an optional alpha suffix (representing multiple scans required to digitize some items, back and front of postcards, for example). Some examples: umn16595a, umn16617, irr00517, shm001558, mhs02721.

JR is concerned that some scanning sites may not have kept id's unique. This would make the MDL id pretty unusable as a key to the dataset. JB also points out that some MDL ids in the existing dataset do not conform at all to the structure outlined above (see "ms.gn00001.1.d1" as one example). This means that the SSR cannot count on their structure to create proper links.

As a result, I think we will probably have to use the CONTENTdm collection alias and pointer (such as "swede-50" with the dash being used to separate the two) as the key to reference between systems. We will keep the MDL id somewhere in the SSR data for future reference, but I can't see how we could rely on it as a key given the current state of affairs.

Note, to use the collection alias and pointer as a key means we have to enforce some rules on the collection aliases we use. JR states that "no slashes, dashes, dots or caps, now or ever" would be enforceable. That sounds like it will work.

Compound Documents

Email with SH revealed that we were not yet dealing with compound documents correctly. (Note, "compound document" is CONTENTdm's way of referring to these, you may also see "complex documents" or just "documents" as opposed to "items". A compound document is a document made up of many individual items.) Some back and forth with JB and JR arrived at a goal: lets make discussions about compound docs refer to the whole document, not just to an item within it.

We learned that CONTENTdm does assign an identifier (a collection and pointer) to the whole document that is separate from those assigned to the constituent parts. Unfortunately, the MDL has not been assigning MDL identifiers to the document as a whole. This makes using the MDL id as the key problematic (we already had reasons for thinking this, see above).

This also sparked a conversation about the correct position for the "discuss item" link. We agreed to try it in the "green header" next to "cite this item".

Zoom Bug

IVB and KE uncovered a bug in the course of doing their MDL SSR homework. It appears that after a bit of zooming CONTENTdm changes the "/" preceding the collection name into its hex equivalent "%2F". I've modified the parser of the referrer on 071210 so that it can now cope with either case.

Assessment

I've created an export script in PHP that produces XML of all comments. You can try it out. The names and text of the comments are encoded as HTML entities.

This is the XSL stylesheet which makes tab delimited data from the XML.

This is a spreadsheet which analyzes comments made through 3/29.