I’ve delayed updating my project blog for quite some time while the DH Press team has worked to prioritize the enhancements and added features for the beta 2.0 version. These features have been determined in large part by the needs and requirements of many of our ongoing projects. We have several DH Press projects that will be launching in the next several months; these projects must necessarily determine much of our work for the time being. Specifically, the Digital Portobelo project (a DIL/IAH Faculty Fellow Project; learn more at the project blog) and the Lebanese Migration to NC Project have firm launch dates. Both projects require enhancements to the tool that, we hope, will make for a more robust platform for other users. I’ll blog about these two projects shortly.
For now, I want to share what we’ve been working on lately, and the directions we expect to be taking in the next three months or so.
For starters, Joe Hope (RENCI) has been working diligently to clean up the beta plugin. He’s been streamlining the code to make future programming easier. He’s also removing all legacy references to “diPH” (the original and ill-fitted name of the toolkit). In the process, he has significantly revised some of the existing functionality, and our DH Press development team is now testing the revised plugin (version 1.5?) to help debug it. Joe will be updating the plugin on GitHub soon.
Joe has also created a new multisite WordPress environment for live projects, so that we can get them out of our Sandbox. This will help us clean the Sandbox, and reserve it as a space entirely for testing and playing. Right now, we have 39 sites and a total of 87 users in the Sandbox. Many of the sites are inactive, and I am currently assessing which of our inactive sites can be deleted to free up server space for our more active users (if you have a Sandbox account, look for an email from me in the coming weeks).
We are also updating the default WordPress theme for DH Press from Twenty Twelve to Twenty Thirteen. This will improve the usability of DH Press projects by moving the navigation bar below the header image (we hope that will be more intuitive for most users). We are currently exploring how we can enhance the customizability of DH Press’s interface, either through customized child themes or through the adoption of existing widgets and other plugins. Since many of our display pages require PHP (because they are dynamic pages which aggregate content based on categories), the trick will be to find lightweight solutions to configuring the look of a DH Press site that do not require much beyond CSS and widget/plugin experimentation. Jade Davis is spearheading this effort for us.
Here’s a closer look at what we’ve done so far to improve the beta plugin:
1. Bundled DH Press with the other required plugins
Previously, in order to use DH Press, users needed to install three other plugins (CSV Importer, which allows users to bulk load their project data; Term Menu Order; and Taxonomy Metadata). Joe has incorporated the CSV Importer and Term Menu Order into the current plugin, and is working on adding Taxonomy Metadata to allow for an easier, one-click install of DH Press in any WordPress site.
2. Eliminate category conflict
Over the summer, we began noticing odd bugs as a result of creating multiple similar projects in a single DH Press environment. While conducting trainings, we would typically use the same training data set. But when participants would go to configure their map legends, we’d see empty duplicate values. Joe has been working hard to fix this, in the event that someone wants to host multiple DH Press projects in a single site, rather than keeping projects in separate silos. Now, if your DH Press site has projects with similar or overlapping data, DH Press will create “alias” values that will improve legend set-up.
3. Parsing multiple values in a custom field
While we had always envisioned supporting motes / custom fields containing multiple unique values, we had not yet implemented this. Now, DH Press allows you to create fields containing multiple values. DH Press will parse those values properly. If you’re using that field / mote to create a legend, the first listed value in the field will determine the marker’s appearance when all markers are selected. The marker will change appearance as the filters are applied on the map, and the marker will show up any time its associated values are turned on in the map legend. The default delimiter is a comma; if you plan to use a comma you will not need to specify the delimiter when creating the mote. If you use something else (e.g. a semicolon), you can specify that so that DH Press will know how to parse your data.
For example, if a datum from the Charlotte 1911 project represents a space that is both residential and commercial, you would be able to list it as “residential,commercial” (rather than having to create a new category, such as “residential and commercial,” or “both”). When activating the “residential” markers in the building use legend, the marker would show up. But it would also show up when clicking the “commercial” value. If both values were turned on, the marker would appear as a “residential” marker.
In addition to these improvements, we are very close to supporting two additional functions:
4. Faceted search
DH Press currently allows you to create parent-child relationships in legends. You can even create new parent categories that had not been included in your original data set. This was a really important feature for “Mapping the Long Women’s Movement,” as it allowed us to create large groupings of concepts and spaces. Rather than create a unique marker for each of the 100+ concepts in our data, we grouped them into eleven parent categories (following the principle that the human eye can only process about 10-12 colors quickly). But we do not have the capacity to display all of the unique child values in the marker legend, such that a user could click on individual child categories to filter the map. Joe is very close to finishing this, so that users can drill down into the visualization with more precision in a faceted search approach.
5. Filtering the map on multiple map legends
Faceted search will be enhanced even further once users will be able to apply filters from multiple legends. Currently, we can only show one legend at a time, but with this enhancement, we’ll be able to show the intersection (“AND” / “OR”) of two (possibly more) legends. So, users exploring the Long Women’s Movement project would be able to look for markers that include a value from the “concepts” legend, and one from the “spaces” legend, for instance: oral history segments that discuss feminism (primary concept) in educational spaces.
Once the updated plugin is debugged, we’ll begin working on the new set of features required for our current projects:
We hope to add additional base maps, such as Google satellite view. But more importantly, we are working hard to extend the current map library beyond historic NC maps. Currently we are experimenting with TMS maps (an OpenLayers protocol) to pull in maps the CDLA processed for Driving Through Time, using a different protocol than those done for the Sanborn Fire Insurance Maps. If we can pull these maps in, we expect that other TMS-styled maps will work, too!
Finally, we are working hard to create an interface that supports multiple unique map views. While I’d like to have up to four views, this may prove too taxing for load time. So right now we’re working on two map views. This would mean that you’d be able to see, for instance, a map of Charlotte in 1911 and a map of Winston-Salem in 1912. Or maybe you want to see two different years for one place. This will be a critical feature for our Lebanese Migration to NC project, which will be part of an exhibit at the NC Museum of Art in February, 2013.
We are also working to extend the audio/transcript tool in two significant ways:
First, we are expanding to include video files that are streaming from YouTube. This would function exactly like our SoundCloud audio files, but it would work with an embedded YouTube media player instead.
Second, we are working to extend the capabilities beyond English interviews. While we cannot automate the transcript timestamping process for non-English files (our version of Docsoft:AV only supports English and is cost-prohibitive to update), preliminary experimentation with manual timestamping has been encouraging. We are developing a process for handling parallel English and Spanish transcripts in a single instance, such that users may be able to view one or the other (or both) transcripts while listening to the audio. Stay tuned for our progress on that front.
We’re also hoping to create a more dynamic transcript view that scrolls along with the playing audio, to help users find their way in the transcripts better.
Additional Entry Points
Once we’ve stabilized the new version of the plugin, we’ll also begin adding additional entry points, or visualizations. We’ll start with a timeline view. Preliminary discussions suggest that the first implementation will be integrated with the map, so that site visitors would be able to see the display of markers in a chronology. We have not yet begun testing this, so our implementation will likely change.
Secondly, we’ll be working on what we’ve been calling the “topic card” view, which is a gallery entry point into the data (check out this early demo using jScroll for infinite scrolling). This will be a nice visualization for image-heavy projects, such as the Digital Portobelo project.
Ultimately, users will be able to create multiple entry points into a single project. We think this will require configuring modals that are unique to each entry point. This will greatly enhance the power of DH Press, and move it beyond a spatial-mapping tool.
I’ll report back on our progress over the coming months. In the meantime, got an idea for a feature? Email me!