Saturday, December 11, 2010
Wednesday, August 25, 2010
Friday, April 30, 2010
There are times as a Research Assistant that I’m very thankful for my ipod. I’ve been spending a lot of time in the Hillman Library lately, researching the imports and exports of opium from the British Parliamentary Papers. I’m not going to lie. It can get a little tedious, and I’m glad I have the music and podcasts coming from my headphones to keep me company. Sometimes I wonder what research assistants did before they had this helpful piece of technology to get them through times like this.
I am getting quite a view of some the technology they had to deal with though. There’s this machine way in the back recesses of the microfiche room called a Readex Microprint reader. I’ve been using it to view the vast amounts provided by the British Parliamentary Papers. The process is painstaking. After consulting an index and locating the right microprint card out of tens of thousands, I scroll through the statistical tables of imports/exports for a particular colony for a particular year until I locate opium. Then, after squinting for quite some time at the screen, I make out the quantity and value of opium and note copy it into an excel document. I then double check the numbers by adding the total for each country being imported from or exported to and making sure it matches the total listed in the original document. You can view the initial results here. (Be sure there is more to come. A lot more.)
I’ve been in the library enough in the past month that the staff members in the Microform room know me by sight. They smile and say hi. Eventually one of them came up to me to ask me how many cards I had used so far. That’s when I realized why I was so recognizable. I was the only person that ever used the Microprint Reader. In fact, they needed to know how many cards I had used so they could tell their administrators that people still used it.
The Readex Microprint Reader is somewhat old and cumbersome. In fact, I came across an article in The Journal of Documentation where D.T. Richnell noted that “In academic libraries, in particular, it has been felt that the advantage of possessing in compact form such material as the Microprint edition of the Three Centuries of Drama has been outweighed by the fact that members of the academic staff were unwilling to subject themselves to the strain of prolonged reading on these reading machines.” That article was written in 1957.
Sometimes I think about what this kind of research will be like years from now. Will humble research assistants still rely on their ipods to get them through the more monotonous parts of the day? Or will there be some other form of technology they’ll be thanking? Our hope with the Dataverse is that with the help of many contributors there will be a massive depository of data freely available in an easy to use format, allowing scholars and researchers to get on with the important work of testing hypotheses about large-scale historical patterns.
I also wonder how long that Readex Microprint Reader will be in the Hillman Library. And what will happen to all the information that can be read from it if we don’t get it online fast enough…
Thursday, April 22, 2010
Thursday, April 15, 2010
Archiving data is straightforward. These data are gathered from contributors and placed online for public consumption. But you may ask, “Why? Aren’t there a million places that already do this?” Yes, there are (well, maybe not a million), but it is our end goal that separates our project from other archives.
So far we only have a handful of these data on display, but they serve as a starting point for a larger project. We hope to one day transform these data to be able to link them together (a.k.a. data federation). It is a very ambitious project, but I believe it is feasible, and could be a game changer for researchers. In my opinion two primary things have to take place in order for this project to be successful:
- We need to identify common denominators for our collection of data. In order to tie these data sets together, we need a common denominator that we can use in a programming algorithm. For this purpose, I feel that GIS coordinates make the most sense, because almost all data are tied to a location. We’ve also thought about tying data temporally and by discipline, and are hard at work at gathering these metadata.
Of course tying data to any common denominator will not be a simple task, but this is what separates our project from most other archives and keeps us motivated to push ahead with the project. That said, we are already in the process of coding our data with GIS attributes (along with adding temporal and thematic metadata). I hope to cover these side elements in a separate post.
- We have to start small. We should approach the programming task using a select group of datasets that are similar. Doing so helps:
- The project to focus on programming.
If we use a set of data that is too diverse at the onset, we might not be able to obtain a sensible common denominator. For example if we looked to combine silver data in the 18th century with poverty data in the 20th century, we might face a common denominator issue: Even though these data are tied to a place, the names of these places are different in the two centuries.
On the other hand, if we focus on a handful of similar data sets, we should be able to give priority to programming. Admittedly, this time-place issue would have to be addressed one day, but to really kick start the project we have to start with a small and similar set of data.
- Promote our project to other contributors.
Researchers may be more inclined to donate their datasets when they see a niche project that fits their work. (One of the biggest issues with data collection is to gain contributions from others, I may write about the issue of incentives in a separate post.)
We have already begun to address the issue of this niche dataset. Our commodity in focus side project has begun collecting opium data sets. Next year, we hope to focus on silver.
- The project to focus on programming.
There are many trial and errors in the work that we’ve done so far. And perhaps we’re wrong to use GIS as a common denominator. But this won’t deter us from working on this ambitious project. We will adapt to new challenges and continue to press on with the creation of a World-Historical Dataverse.
Friday, April 9, 2010
The interactive visualization can be found here. Information on the project can be found here. A newer (though less dramatic) version of the visualization can be found here.
One way we could understand the popularity of data visualization is by tracing its history. I find the work of Michael Friendly and Daniel J. Denis to be the most comprehensive in cataloging milestones in data visualization. Their website is called Milestones in the History of Thematic Cartography, Statistical Graphics, and Data Visualization. Their work involves breaking down each milestone in an interactive timeline. Friendly also has a paper by the same name, which is worth the read if you are interested in this topic.
There are some interesting stuff on the site, such as the first star chart (pictured below) and the hysterical best/worst data visualizations.
“By the mid-1800s, all the conditions for the rapid growth of visualization had been established. Official state statistical offices were established [throughout] Europe, in recognition of the growing importance of numerical information for social planning, industrialization, commerce, and transportation. Statistical theory, initiated by Gauss and Laplace, and extended to the social realm by Guerry  and Quetelet , provided the means to make sense of large bodies of data.” (Friendly 18)
Conversely, the mid-1930s is described as the “modern dark ages” of data visualization:
“There were few graphical innovations, and, by the mid-1930s, the enthusiasm for visualization which characterized the late 1800s had been supplanted by the rise of quantification and formal, often statistical, models in the social sciences. Numbers, parameter estimates, and, especially, standard errors were precise. Pictures were—well, just pictures: pretty or evocative, perhaps, but incapable of stating a “fact” to three or more decimals. Or so it seemed to statisticians.” (Friendly 27)
What about today? According to Friendly, the period that we’re living in now is known as the “High-D data visualization” era. This period is characterized by accelerated and varied visualization processes mainly due to technological advancement. We should note that there have not been many innovations from the 2000s.
When we place data in context of time, we can see why data visualization remains to be the most popular type of data site out there: We are in a period of awakening from the data dark ages. This movement reflects our own mission to improve historical data access to the public and our use of tools such as GIS data mapping to push the boundaries of visualization.
Thursday, April 8, 2010
We make this distinction because data set sites are sites that houses data itself, while data portals point toward other sites that house the data (in a way then our Dataverse project is a portal to other portals).
Below are some frequently asked questions (and their corresponding answers) that I get when I talk to people about this project:
What’s the point?
The main goal of the Dataverse Project is a creation of a comprehensive, long term, world historical database. By gathering external data sources:
- We are identifying potential collaborators that we hope would contribute relevant data to this objective.
- We are informally gauging the availability of long term historical data.
- We are providing an auxiliary resource of data for people who visit our website.
- We are analyzing new innovations from data collectors.
There were some discussions going back and forth as to what we should name these sites that we were collecting. At the end of the day we used our own logic, instead of naming conventions to name the data we have collected. We did this because there were no naming conventions to speak of. For example, what we call data portal is also known as data warehouse, gateway, clearinghouse, etc. Though someone could technically describe the difference between these naming conventions, the difference between them is small and so calling something as encompassing as a portal would meet our need.
We divided them into two categories of data, but in all honesty, there were sites that do not conform to each individual description. For example, there are sites that provide data sets and data portals (similar to our Dataverse Project), but they were few and far between and we found no justification to create a new category for these sites (for now).
How do we display them?
Currently, it’s set up as a bunch of links on two web pages.
How would we like to display them?
What we hope to do going forward includes separation of portal/sets by categories/discipline, region and timeline. Adding metadata/tags to individual data sites would be the subsequent step. This tagging capability gives users the ability to extract data sites based on naming conventions that go beyond categories, region and timeline. Tagging could be done by users (tags) and/or added by staff (metadata).
There are many sites that already implement our proposed categorization techniques. We will be featuring them on our blog in the future.
What are some of the issues we face?
Besides naming conventions and lack of data display options, another issue that we ran into is attribution. We did not want to give an impression that these sites belong to us, so we explicitly make users agree that the information they are obtaining are from a third-party site, that we are merely a conduit of these data. There are also not one universal set of rules for data usage for these sites, thus we made it clear that usage is preconditioned upon the terms that each website has set.
There are also issues of viruses and expired links. We make the best effort in avoiding sites that do pose these problems. We run virus scans on a sample of files in the sites and we routinely check for broken links, to give users the best browsing experience. That said, we still included a disclaimer that we cannot guarantee virus-free links.
We are also having issues naming the external sites. At the time of writing, they are alphabetically arranged based on our description of the site. We are addressing this issue by providing a search link, but a naming convention has to be established to improve the functionability of our website.
What else can we do to improve the external links section?
As of now, I can’t think of any beyond what has already been said. I sincerely hope that you can provide some feed back on our website. We’re new and we’re willing to listen.
Conclusion: what we’re attempting is nothing new. However, compared to a search of “News Blog” (20.9 million hits) or “Sports Blog” (8.6 million hits), data blogging has yet to take flight (assuming that it will).
So data blogging is actually in a strange place where only a few have taken the lead yet the level is nothing compared to other forms of blogging. The first order of business then, is to see what these select few are up to.
Through simple Google research here are some key findings for data blogs:
- Data blogs are predominantly blogs on data visualizations. To me, this makes sense because people are trying to find new ways to display data and go beyond the ubiquitous scatter plots and pie charts. Also, since people are visual animals, this topic is able to attract more traffic, than say, the folks who write about data mining, making it more profitable and pervasive.
- These data display sites come in three different forms:
(a) Sites that provide tools to display data
(b) Sites that display interesting data
(c) Sites that discuss data visualization
- Many of the data blogs that features interesting data are concerned with showcasing contemporary data. The Guardian has an excellent one called the Data Blog.
- There are virtually no sites that are blogging exclusively about historical data and its implications (aka what we’re doing). Hooray! There are, however, historical blogs and data blogs that occasionally write about history, though none are completely devoted to historical data. Note: there are also insanely interesting blogs about the history of data.
- Then you have really esoteric sites writing about data mining, data storage, legality of data, etc.
- In general, there are two types of sites out there: sites that archive and sites that point to these archives. The Dataverse Project does both.
- The Dataverse site is far from perfect. There are sites that are the models for categorization (spatial, temporal, by discipline), sites that are models for archiving, and sites that are models for data attribution that we can learn from.
- There are also sites that provide space to archive. One famous one is the Dataverse Network Project (not to be confused with our World-Historical Dataverse), run by the folks at Harvard.
- Finally, you have sites that are in my dreams: sites that do not exist[s], that I wish had existed. These include blogs on the nature of data, on data software, on open source data, etc.
The Dataverse Blog is a space where we will discuss data ideas and data issues. These include, but are not limited to, data mining, data visualization, data archiving, data attribution, and data storage. An emphasis would be given to historical data, although this does not preclude us from showcasing contemporary ideas that may benefit historians.
We want this blog to be a companion site to our Dataverse Project, so that users can see what goes behind the scenes of our day-to-day activities. More importantly it’s a good way to show you how we reach certain decisions on data issues we’ve encountered along the way. The blog also serves as a place where we will feature best practice sites, from a myriad of disciplines, for the benefit of anyone working on similar projects.