iSchools and the Digital Humanities

This week, I spent much of my time working on modifying the logic in the watracking project to match up with the new database structure. Information about say, object TYPE classification (audio, video, text, image), is linked in an entirely different way, requiring some rewriting. CodeIgniter restructuring is the ultimate goal, and I have had some uncertainties about the best way to approach the problem. I could spend some time in the design phase now, and see if that informed the current code additions and subtractions that I’m making to get the application working. OR, I can wait and get the application working, with logic that reflects the new, generalized database design and then spend some time in design phase before implementing the restructuring.

Searching forums online, I haven’t found too much, (nothing, really), about others’ approaches to changing an existing CodeIgniter structure and challenges that seem to common across these attempts. However, I have found sample PHP instantiations of MVC and I think I have a fairly good grasp of what goes in your average model relative to the controller and the view.

The model should contain all the data logic (easily said, not as easily understood). One example that was particularly helpful in making sense of it for me asked readers to imagine a video-on-demand site. Just to keep it simple, you have two models – user.php and movie.php. In user.php, you create a user class and then you had methods such as “getUserName()” and “logIn().” In movie.php, you have a movie class with methods such as “getTitle()” and “getPrice().” After reading this simplified example, I thought, “How does this translate to our tracking app?” Well, I’m still working that out…

At first, I thought for each major application directory, (controller, view, model), you would have subdirectories themed by UI function: add, edit, delete, search.  However, after talking to Keith – it seems that might be alright for the controllers but coupling the models to the controllers so closely like that does not seem like the best approach.

I understand why currently there is a model to correspond to each table in the database – each represents a kind of data and it takes certain methods to use that data. I have some concerns about code repetition but perhaps you can’t avoid it all together. I’m continuing to work on it.

Whitman Camp is coming up here at CDRH, and Matt Cohen, from UT, is expected to be in town. I am looking forward to seeing how CDRH organizes these kinds of events – what gets done, what kind of plans get made, etc. And I am also eager to speak with the digital humanists from home to see what kinds of thoughts they have about the future of the field at UT – a center, or an embedded kind of approach where each humanities department develops its own dh working group. I’m sure there are lots of contributing factors in determining what sort of configuration works best.

Next weekend is Fourth of July and I’ve heard from more than a few people that it is a pretty big deal here. The celebration has begun a little early in my neighborhood – several loud fireworks went off over the weekend. The fireflies are equally enthusiastic – as I’m seeing them in larger numbers every evening around 8:30.

Week 3 was all about database redesign. After talking to Keith earlier in the week, I had an idea about how to proceed. Before speaking with Keith, I was experiencing some project focus confusion between taking his code and getting it to work for Cody – an approach that would produce a usable application for the project team in a shorter amount of time VERSUS doing something totally different, something  more generic that would include a database redesign and some code rewrites -an approach that would produce something more modular that could be easily implemented as a tracking system for most types of CDRH projects but would be more time intensive. We determined the latter was the best use of my time here and I set out on the redesign.
I think one tricky part of extending what someone has already done, is that it is easy to become married to that idea/structure/organization in the time it takes for you to understand how it produces the outcome it is producing.

…And you don’t do database redesign in a vacuum. For every change, you have to have some idea for how it might play out – either in a query or somewhere in the application logic. Another potential trap is trying to design and conceptualize your way out of trouble without ever getting your hands dirty. It’s impossible. You are bound to get stuck in a purgatory of MySQLWorkbench EER modeling, or covering your desk in notebook paper prototypes, (which is a likely outcome even if you do get your hands dirty).

I’m currently in the midst of trying out my latest design. My approach is to:
1) change whatever needs to be changed in the project directory to get the application working with my new database design.
2) make changes as needed
3) reorganize the ci project so that the bulk of the logic is contained in the models and the add and jsrequest controllers are broken up into more task specific chunks that can act as controllers themselves.

We’ll see how it goes…But, in space between design, testing my design’s feasibility, and producing informative output I’m learning a ton.

After my first week’s self-imposed rush to get comfortable with a range of new models, desktop work environments, and the CDRH itself – this week I felt more at ease with all three.  I dove head first into Keith’s code. When you look at someone’s source code for their application, you are looking at their problem solving process – the bulk setup (the design), the quick fixes(constraints on your time during the process), the places they intended to go back and revisit. Code is a fairly revealing form of authorship.

You reveal in it your desire for elegance, for modularity.

However, reading code that is so thoroughly abstracted can be difficult. The MVC model that is instantiated by the CodeIgniter framework is straightforward in the abstract. However, in practice and as you might imagine, it takes on the idiosyncrasies of the programmer and the specific project needs. Therefore, the business logic sometimes shows up in unexpected places.

We met with Laura, CDRH’s metadata encoding specialist, who has worked the most with the Cody EMS. She walked us through the parts of the EMS that are particularly problematic. Keith’s tracking DB for the Whitman already solves most of the problems that Laura pointed out, which means that modifying the Whitman tracking DB is probably the best plan. Also, CDRH faculty and staff are already familiar with the user interactions and the location of specific functionality such as editing and deleting an entry.

There are some differences between Whitman and Cody tracking systems that we are already aware of: Cody will include video and image object types; the site will require a changelog so that users can see the last person that made a modification to the object entries; naming scheme will have to be revised; additional object type-specific, dynamically generated forms; and a few more issues that I have documented elsewhere. The Whitman currently produces additional forms based on the kind of object that the user is adding (Text, etc.). Text is currently the only object type that has additional forms, (these provide places for user to add in info about whether the object they are adding is ‘recto’ or ‘verso’, etc), however, if the Cody project members decide on elements of video they would like to track, the current AJAX enabled logic that produces the additional forms without reloading the page, will need to be copied and modified accordingly.

Also, any additional object types correspond to changes in the table design. One current DB issue with Whitman is the way that object type and object ID are currently linked. The Object Type table consists of an ObjectType PK, the ObjectID, the ObjectType(text, audio, ephemera, etc), and something else. Why not simply create foreign keys for all objecttypes (AudioObjectID FK, VideoObjectTypeID FK, etc)? Well, if you did that each object type would need to be a column, and you would have 3 out of 4 cells in a given row with NULL values. In its current configuration, as Keith explained, “It sacrifices referential integrity in order to reinforce the idea that each object can only have one type.” Tasked with rebuilding the DB for Cody, we are given the opportunity to rethink the current scheme.
Another current Whitman DB issue is that ImageObjects (which would include photographs, poster, and other image content – not scans of text content) were never used. So, we get the opportunity to explore what kinds of elements need to be tracked there as well. Originally Keith had planned for ImageCreationMethods and one other element. After speaking with Laura it seems that tracking elements for video and images in particular, haven’t yet been determined. This presents an obvious but common problem in DB design, which is planning for likely additions and trying to make the schema as flexible a possible. I’ve built an EER model in MySQL Workbench to play around with alternatives.

I’ve been writing notes all week about specific files that will or won’t have to be changed, which lines and so on…attempting to make my job a little easier as I begin the new CodeIgniter project for Cody, which I’ll be doing now….

My summer project consists of understanding the structure and logic of the Walt Whitman object tracking database, perform a needs assessment on users of the current tracking system used by the William F. Cody project and then, determine if/how/what to abstract from the Whitman tracking tool to improve on the Cody system OR determine other alternatives, and finally to come up with a proof of concept on how that might work

This week was a whirlwind introduction to: the convenience of working on projects in development environments like NetBeans and Eclipse; PHP frameworks, CodeIgniter in particular; using Boolean notation in PHP; AJAX and JQuery integrated into a PHP application; object oriented principles such as classes, inheritance and abstraction; pass by reference; and some other things that I’ve documented but are not immediately coming to mind.
Professor Walter introduced me to many of the CDRH and Library faculty and staff members, including the designer and creator of the Whitman tracking database, Keith Nickum. In terms of CDRH staff, I’ll probably be working with Keith most often for the duration of the internship.
He is a wealth of knowledge, having created the application himself, and having worked on all aspects of the project for the past year and a half.

My first day, I found myself at Keith’s desk more often than I would have liked but by the second day, I had a laundry list of items that I could research and learn on my own. Although I can’t say that I feel comfortable with all these new concepts already, my anxiety level has gone from information overload to manageable thanks to a systematic approach to Keith’s code. I make notes on everything I don’t understand, look those items up, and run through practice exercises whenever possible. In the process more questions arise and I repeat those steps.

I am grateful for the opportunity to be here, at CDRH, working with an amazing group of people who are looking for new and exciting methods to apply to working with humanities materials and humanities-oriented users.

Well my time has come to an end at the CDRH and I have had the opportunity to dip my toes in to a number of digital humanities related fields.

The main focus of my time in Nebraska was working with Dr. Peter Bleed in the creation of an archaeological data web tool that would enable researchers to add, access and edit archaeological data and historical data. The idea was to create a resource that would allow for the integration of historical documentation (photographs and maps) with modern views (Lidar data, orhto-rectified maps) displaying archaeological data (GPS data, site maps).The hope was to create a tool that would aid in the interpretation of archaeological findings and assist in further archaeological exploration. To some extent this was successful. I was able to use ArcGIS to Geo-reference old Sanborn maps to the excavation site so that the historical evolution of the site could be seen. Using this to predict where other structures might be was a bit more problematic. Sanborns are accurate but they are not precise. This means that while you have a general idea of the location of a structure concrete prediction is still iffy. Alongside of this visualization I created a collection of digitized primary sources from the site. All in all, this project was more of a proof of concept, an exploration of what sort of tools could be created and how to best go about implementing them.

As a supplementary to the creation of the visualization tool I also wrote up a set of recommendations for the creation of a permanent repository for archaeological materials. It should come as no surprise that archaeological data is increasingly “born digital”. Unfortunately stewardship of this data is undertaken in a piecemeal fashion. The purpose of this archive would be to create a resource for preservation that encourages use and discovery.

In addition to do all of that, I had the opportunity to do some TEI work on a project about the Mountain Meadows Massacre. The scholar had
located a number of contemporary accounts written about the massacre using Google books. I worked on encoding OCR’d XML of the texts to help make them searchable.

I was also able to work with the archive at UNL adding photos to collections using the Archon records software. Again, very interesting work. Archon is a very robust software package and it was interesting to see how the process of resource creation functioned, and just how versatile of a software package it is.

I also had the opportunity to explore Lincoln and Nebraska in general. The folks there are generally lovely and the summer was wonderfully
mild compared to my native Austin. If you are every in Lincoln I recommended taking a couple of hours and seeing the Capitol, it is an amazing art deco structure, then go get some pizza at Yai Yai’s.

-Stephen

Posted by: Stephen Pipkin