What two changes we made to ArchiveGrid mean for users

When Morristown National Historic Park in New Jersey was dedicated on July 4, 1933 as the first national historic park in the United States, there was a parade. Its theme was Spirit of ’76, referring to the Revolutionary War when a cluster of three historic military posts, Jockey Hollow, the Ford Mansion, and Fort Nonsense, served as George Washington’s army headquarters.

Imagine you’re in Google looking for photographs of the parade. You click on something and go to a page in ArchiveGrid, and this digitized black-and-white photograph from the dedication is discovered:

If you have visited ArchiveGrid before, you’ll know you’re looking at an archival collection description. You’ll be able to find out who holds the collection (this one’s at the Morristown and Morris Township Public Library) and how to access it.

Google Analytics shows that around 80 percent of ArchiveGrid visitors arrive this way, by clicking on a collection description or item they found in Google. These descriptions act as a kind of “home page” for most new visitors to our site, and this might be our best opportunity to show them what ArchiveGrid is all about and provide them maximum value.

With our new “More like this” feature, we’re hoping to do exactly that. Located in a box on the right-hand side of an individual record display, “More like this” uses connections made in ArchiveGrid’s Solr index to offer extra contextual information and links to related materials – without disrupting the flow for those who just want contact information and to learn more about access to the resource.

Success of “More like this” depends on how rich the collection description is and the extent to which related people and topics can be found in other descriptions.

Right away we noticed this feature seems to work well for items from digitized collections, such as in the example above. It provides a way to view other items in the collection without searching:

For other collections, it can suggest closely-related resources at other institutions:

Now imagine you’re doing a search in ArchiveGrid. A relevance ranking algorithm that mostly paid attention to keyword matches, compared to the number of times they occur in a description, and the description’s overall length, used to generate your search results.

We made some adjustments so now matches in certain metadata elements (title, author, scope and content) get emphasized over other fields in the keyword index. Behind the scenes, we’re grouping descriptions by their extent into small, medium, and large. This allows us to give greater weight to collection descriptions over sub-series and items.

As a result, we’re doing a better job of making “key collections” appear near the top of a related search result in ArchiveGrid.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>