It doesn't seem like it's been over four years since I joined MITH and started working with Project Bamboo. Just because I've moved on to a startup and the project's been mothballed doesn't mean we can't mine what was done.
The problems with Project Bamboo are numerious and documented in severalplaces. One of the fundamental mistakes made early on was the waterfall approach in designing and developing an enterprise style workspace that would encompass all humanities research activities rather than produce an agile environment that leveraged existing standards and tools. Top down rather than bottom up.
However, the idea that digital humanities projects share some common issues and could take advantage of shared solutions is important. This is part of the reporting aspect of research: when we learn something new, we not only report the new knowledge, but how we got there to help someone else do similar work with different data. If we discover a way to distinguish between two authors in a text, we not only publish what we think each author wrote, but the method by which we made that determination. Someone else can apply that same method to a different text.
As I start thinking about what should go into the core of a linked open code schema, I'm tempted to put a lot of high-level operations into the core so they run faster. History tells us that's the wrong way to go.
I've been working off and on over the last six months on a programming language that sits on top of linked open data. Think of it as linked open code.
von Neumann Architecture
Before von Neumann made his observations about code and data, computers typically had some memory dedicated to code, and other memory dedicated to data. The processing unit might have a bus for each, so code and data didn't have to compete for processor attention.
This was great if you were able to dedicate your machine to particular types of problems and knew how much data or code you would typically need.
Von Neumann questioned this assumption. Why should memory treat code and data as different things when they're all just sets of bits?
I'm building a language designed to be natural with linked data just as programs today are natural with local memory. The result is highly functional and data-relative in nature and reminds me of how XSLT works relative to XML nodes (e.g., a current node, child nodes, ancestor nodes). I have quite a few tests passing for the parser and core engine, so now I'm branching out into libraries and seeing what I can do with all of the pieces.
A few months ago, I accepted a job outside the academy. This doesn't mean that I'm abandoning digital humanities. In this post, I lay out what I want to do in DH going forward. The common thread through all this is that I believe linked open data is the best way to break down the silo walls that keep digital humanities projects from sharing and build on existing data. Continue Reading My DH Agenda
Today, let's build a set of tools that will help us create a concordance of a text. We'll have to make a lot of assumptions so that we can see the core pieces, so keep in mind that any real implementation will probably have different details.
We'll assume for now that we have a stream of characters representing the text. We haven't discussed where we get data or where we store it yet. That's for another time. For now, we're focused on what we do with the data between getting and storing it. If we can wrap our minds around what to do with the data, then we can plug-in any data retrieval or storage we want onto our processing later.
While I am trying to round out the content management aspects of OokOok this year, I'm starting to think ahead to next year's work on databases and processing. Part of the goal is to offer a platform that lets you take advantage of parallel processing without requiring that you be aware that you're doing so. Of course, any such platform will be less powerful than hand-coding parallel code in C or using your favorite Hadoop library. Less powerful is better than not available. I want OokOok to make available capabilities that would otherwise be hidden away.
Map/reduce seem like the simplest way to think about parallel processing. We have two kinds of operations: those that look at one item at a time (mappings), or those that have to see everything before they can finish their calculation (reductions). Reductions can get by seeing one item at a time if they can keep notes on a scratch pad. We could put operations then into two slightly different camps: those that need a scratch pad (reductions) and those that don't (mappings).
I'm making rapid progress in getting OokOok to a stable programming scheme. I haven't made a lot of changes in its capabilities, though I did add the ability to archive themes and projects as Bagit files yesterday, I've been working on making the important stuff declarative. By hiding all the details behind a veneer of relationships, I can fiddle with how I manage those relationships without having to touch every relationship every time I make a change in the underlying schemas (and schemes).
For those used to an older style of Perl programming, this might come as a surprise. For those who have dealt with things like MooseX::Declare and CatalystX::Declare, you'll be shaking your head at my foolhardiness in jumping into making an OokOok:Declare that hides the details of how to construct certain types of classes.
Behind the scenes, OokOok consists of controllers, models, views, REST collections/resources, SQL result objects, a template engine, and tag libraries for the templates. Almost two hundred classes in all.
If I built all of these the usual Perl way, there'd be a lot of boilerplate code around. By moving to a declarative approach, I can isolate all the boilerplate in a few core meta-classes. When the boilerplate has to change, I only have to touch one place. Everything else comes along for the ride.
For the rest of this post, I want to walk through how I use some of these declarative constructions. I won't get into the scary details of how to make declarative constructions in Perl (at least, not in this post).