Playing with WorldCat's Long Tail
Most people don’t get too excited about search result pages. We type in some terms; we get some results back. We expect the results to be helpful, and we expect all this to happen in a few seconds. We don't ask for much more than that.
At OCLC we take search a little more seriously than most folks. But even so, I was really surprised when I started experimenting with the new digital image collections that have been integrated into WorldCat.org. Just adding 'cntnt' to a search term will limit your search to the new collections.
A fellow Tribe fan sent me a link to a search on "Cleveland Indians" which pulled up about 60 digital images mostly from the Cleveland Memory Project. I love this shot of Satchel Paige in his Cleveland uniform.
That was cool. But a quick cut-and-paste, and I had a handful of images of diners, and then a great result set for "automobiles".
But my favorite searches so far have been on "political cartoons" which unearthed around 30 great cartoons from the 1960's and earlier, and a search I did for my mom, who collects postcards, which brought up several thousand wonderful old postcards.
So try some searches using the limiter 'cntnt' and see what you find. And share your discoveries with the rest of us by posting the links in the comments below or on your own blog.
0 TrackBacks
Listed below are links to blogs that reference this entry: Playing with WorldCat's Long Tail.
TrackBack URL for this entry: http://worldcat.org/blogs/mt-tb.cgi/44
On April 15, 2008 at 7:31 AM K.G. Schneider said:
Search results pages are indeed fascinating.
I tossed in the term *lesbian*:
http://worldcat.org/search?q=lesbian+cntnt&=Search&qt=results_page
Hmm. Why is OCLC 62943300 in those results?
(Fascinating as that item sounds!)
Also, of the six results, why is 163926908 the first one? That one seems a step away from several of the items, such as pictures of (gay and) lesbian marches. ??
On April 15, 2008 at 8:56 AM Alice said:
Hey Bob--I assume 'cntnt' is short for content?...but I guess we can't use the spelled out word because the search engine would think I'm searching for topics on "content" rather than content itself! Funny how images are termed "content." Are there plans to have a tab for image results?
And sometimes the image shows up in the brief results page, but sometimes I have to click around and it takes me (eventually) to the photo collection Web site itself. Should I be using the search limiter differently?
On April 15, 2008 at 2:18 PM Bob Schulz said:
Regarding the "lesbian cntnt" search, after talking with our engine developers, we've discovered that all of those results are coming back with the same relevancy. The search isn't finding both of those terms in the same field anywhere in those records. As such, the items are being returned in more or less random order, in spite of the fact that some items have at least one of the terms in the title, while others have one of the terms in the Summary/Abstract (which can be found under the Details tab in the record).
This, of course, is not ideal behavior. I'll be speaking with the developers to improve this. Thanks for bringing it to our attention.
On April 15, 2008 at 2:25 PM Bob Schulz said:
CNTNT is for CONTENTdm, and that code appears in one of the MARC fields for most CONTENTdm records. As such, it can be used to surface a sizable chunk of them. CONTENTdm is more than just images; It represents a wide variety of digitized collections, including "local history archives, newspapers, books, letters, maps, slide libraries or audio/video materials."
We're actually looking at a broader scope than just images (although an image search would be excellent, once we had a critical mass of images to search). We're planning on providing a "digital only/online only" type of search which would return only those items that could be reached online, no matter their type.
On April 15, 2008 at 8:18 PM Daniel Cornwall said:
Thanks for sharing this tip. I was impressed with the number of non-Alaskan institutions with interesting photographs from Alaska. I could see using this in some of our history searches.
On April 16, 2008 at 7:49 AM K.G. Schneider said:
Thanks -- I will indeed keep poking around. I assume that under those conditions you really mean to give titles a little more weight??? Also, WC exploring using actual data for relevancy... e.g. "a lot of people looked at this picture"?
There are two reasons I suspect CNTNT is not an ideal limiter, and only one is work-safe: do you want to suggest WorldCat will only work with OCLC's DAM?
On April 16, 2008 at 8:47 AM bobrobboy
said:
Daniel, Care to share any of the searches you conducted to turn up the Alaskan photographs?
On April 17, 2008 at 7:32 AM K.G. Schneider said:
Can't speak for Daniel but I threw in "alaska cntnt" and he's right, it's a fascinating range of institutions. Something fun about digging up a 1964 photo of an Alaskan caribou (or in the singular is that "caribee") from a digital archive in Baton Rouge. Makes me want to see if an archive in Alaska has pictures of Gulf shrimp. (Another WorldCat game??)
*Is* this limited to ContentDM collections?
On April 17, 2008 at 8:35 AM bobrobboy
said:
Karen, It is fun! These searches remind me of a time as an undergrad at Ohio State when I got a few precious minutes in the University Archivist's office and was able to flip through a box of photos.
And to answer your question, yes. These searches are limited to ContentDM collections.
On April 21, 2008 at 11:56 AM Jennie Stapp said:
I'm glad to here that there are plans to improve how users can limit searches for digital content. While the current limiter provides interesting results, it is not intuitive or obvious for users.
Jennie
On April 21, 2008 at 10:25 PM bobrobboy
said:
Right. I debated a bit on whether the cntnt search would be useful for readers because it is so unintuitive. But in the end I decided it was "interesting" and "fun." But we need to make some changes if this type of a search is going to be useful...or used.
On April 23, 2008 at 11:50 AM K.G. Schneider said:
The G-Man does it with links at the top of the page which switches the search to images, plus of course a few top images show in many searches. Not that you don't know that, just pondering out loud. :-) (I know: let's write a blue sheet! sorry... Dialog joke)