Today, we are going from 12,598 records to more than 94,000 in our collection online and this increase represents a substantial change in the way we are releasing information on the web.
With the launch of the collection online in July 2008, we began with a policy to release only records that had been properly vetted. The curatorial staff felt that the data released needed to be both accurate and best reflective of the various collections, so in our original spec we created a multi-layered approval process to publish an object. The vetting process would start in curatorial, allowing them to vet certain parts or an entire record, then the final release would come from our collections staff after an additional once over. Even though the vetting process worked very well, we were finding that it was taking an awfully long time to publish objects to the web. For instance, in July 2008 we started with 5168 objects, but eighteen months later we had only grown to 12,598. While that represents a substantial amount of good data going online, behind the scenes we were seeing long queues of objects ready to release, but hung up somewhere along the way for little bits of final approvals. In the end, keeping up with the demand of approving records was causing us a great deal of work and getting information out the door was problematic, so we’ve changed course.
At this point, records get published by default and information is added or corrected as we go. While there are still tons of records that are restricted for various reasons, what we are seeing at this point is much more representative of our holdings. Of course, the release of more records means we had to rethink some aspects of the user experience, so I’m going to run through some of these changes now.
Record Completeness Meter:
One of the most important changes is the visual meter that indicates the completeness of the record. We want to give our users a very visual way to understand where a record may stand in terms of the overall picture of our data. As you move your pointer over the meter, the ratings are explained in a tool tip.
Release of Study Images:
Given that we have a lot more records going online, one thing we are lacking are photographs for many of them. Often our curatorial staff will take study shots with point and shoot cameras just to have something on file for reference. Internally, we find these shots very helpful and figure that our web visitors may feel the same, so many of these images are being to be released with an explanation regarding the quality.
Comments Overhaul:
We’ve had comments enabled on our object records since the beginning and have received an entire range of responses. One of the things we noticed is many of the same questions pop up over and over again, so we’ve overhauled the comments area to include a FAQ. We are hoping this will cut down on some of the more routine questions by putting the information right where people are asking for answers. Here’s a great example showing why we enable comments on collection objects—you just never know what will come by allowing this kind participation, but we hope the added FAQ will provide better communication all-around.
Browsing, Searching, Sorting:
When browsing, quality records are being pushed to the top of the pile. When searching, we default to showing relevant records, but users can re-sort the results on demand by relevant or complete.
Author RSS 
RSS 
Eastern Parkway/Brooklyn Museum
Can you please discuss the “Record Completeness Meter” more and how the rating is made?
The timing of this post is perfect – I’m dealing with these issues for a small teaching collection of historic fashion at Vassar College. I’m trying to get the collection online for student research, but there’s no time for the “vetting” process.
I’m imagining this kind of visual meter would provide a wonderful compromise – to share what we have, but make it clear that it’s incomplete.
Hi Arden,
Basically, on the back end, we assign a point score based on how much information we have and that determines where a record sits on the meter. If you run your mouse over each rung in the meter, it will show you roughly what we mean by each bucket, but here they are in a list:
Poor: Objects lack photos and have very little data.
Meh: Some data is approved, but records are far from fully vetted. Objects rarely have photos or descriptions.
Adequate: Much of the data is approved, but records have not been fully vetted. Objects may have low-quality photos.
Good: Most data is approved, but records have not been fully vetted. Many objects have quality photos and descriptions.
Best: All data has been vetted. Objects generally have quality photos and descriptions.
This kind of thing was inspired by what we’ve seen on Wikipedia with regard to stub records – we just wanted ours to be a bit more in keeping with our own graphics.
Real interesting work. I’m all for it.
Meh is a great category. I think many of our projects fall under it.
This is fantastic! I’m a big fan of the idea that some information is better than no information, especially if you’re able to give the user some idea of the quality of what they’re seeing. Great work getting this all online!
Congratulations and thank you!
Hi!
I find this very interesting! I work at the Maritime Museum in Sweden and we are currently working on “opening our floodgates”. I have a question for you: We are debating whether we should publish all of our database posts, even the ones where we have very littleinforamtion, or if we should just open the floodgates and display everything. Was this a difficult question for you? What made you decide on publishing everything?
Best regards
Johan Löfgren
Hi Johan,
I had mentioned in the post the reason for the change of policy, so I won’t reiterate that here. In terms of the current policy, we are holding back the record if there are fewer than three fields populated. With just two fields, the information seemed frustratingly too little to do much good. However, over time, our thoughts on that may change as well as we see how people use the database.
Hi,
I’ve been trying to create a similar quality of records scale for our own internal use and I think that publishing this information is a really interesting and cool idea. Especially in terms of prioritizing data cleanup projects this kind of tool would be really helpful. Unfortunately we don’t have the ability at the moment to evaluate approved/vetted for individual fields, only for the entire record.
Was this scale a new creation as a part of your ‘open the floodgates’ project or was something similar already in place that was for internal use only?
Best,
Scott
Hi Scott,
The scale was created specifically for this project – it’s not part of our internal systems.