HistoryWired was an absolute letdown. I was really excited to check it out when I saw it was operated by the Smithsonian. However, I was disappointed to learn that the site had been discontinued in 2016. When you go to the link for HistoryWired it takes you to what is essentially a memorial page which gives background and context for what it used to be.
From what I gathered HistoryWired was an experimental website (after doing proper website optimization) was launched in 2001. It’s main purpose was to explore a “sampling” of the National Museum of American History’s museum collection including 450 objects. These objects were diverse ranging from the “everyday” to the “famous” and they were selected because of the interesting stories they had to tell.
It used a mapping strategy developed by Martin Wattenberg at SmartMoney, Inc. and featured a map that was designed to help users find objects of particular interest to them. Apparently, the map face had 10 categories, each one representing different types of objects from the collection.
The best I have to go on for determining how this site worked is a screenshot featured on the memorial page. It shows a number of blocks, each corresponding to one of the 450 featured objects. These blocks varied in size based on the number of views an object had received during site visits by other users. Evidently, it had many customizable features including timeline sliders, theme buttons, and text search.
Since shutting down this site in 2016, they simply rely on their collections search. I checked out their current collections search and it is pretty typical of most museums with online exhibits and a collections database. You can search an object and it gives you the information. One neat feature is the”featured object groups” section. Here you can click on a link like “The Antibody Initiative” or “Women in World War I.” These links take you to a page where they have curated mini-exhibits that you can click through. It is a cool tool even though it eliminates the really awesome visual aspect of HistoryWired. You can’t set parameters and there is no block size based on previous users to analyze. Below is a link for the “Women in World War I” exhibit.
While this site is a letdown, I think it provides an important lesson on the pitfalls of digital media. There are many experimental digital media platforms out there and it is crucial to remember that they require constant updating and monitoring. Institutions running these platforms must continually assess whether the platforms are working and engaging site users. HistoryWired is hardly the first or the last platform to fall by the wayside or be discontinued. The Smithsonian seems confident in their choice and believes that there current collection search technology rivals their HistoryWired platform. At the bottom of the HistoryWired memorial page they state the following:
HistoryWired was the Museum’s first effort to provide a way for online visitors to search and explore a sampling of our collection. A more standard database search was added in 2005. Today, our collections search provides access to hundreds of thousands of digital records documenting artifacts and archival materials held by the Museum.
A WORKING WEBSITE!!! I was excited that my second website is in use and functioning. This site is used for text reading and analysis and is meant to “facilitate reading and interpretive practices for digital humanities students and scholars as well as for the general public.” The key to using Voyant Tools is that you have to use web-based texts!
To start with Voyant, you take a bunch of online texts and put them all together. For my example, I took the blog posts from this class and put them into Voyant. Now, this was not the first example I attempted. I found the actual input of text to be incredibly challenging. I first thought that I would input all of our online readings for this week. However, when I tried to take the text and copy and paste it into the add text portion, it was very difficult to highlight the parts I wanted to. I also tried to use the URL and insert the text that way, since that was described as an option. This too was a failure as it did not actually input the text. Using the blog posts ended up being easier to input as they were much shorter.
Even though it was easier with the blog posts it was still really tough. I tried for a long time but honestly could not figure out how to make the separate posts into individual entries. After reading another blogger who used this site to analyze the 30,000 Hillary Clinton emails, I decided I would follow their lead and enter all the text on the same entry. Check out this Hillary Clinton project though. It is pretty neat. http://www.maxkemman.nl/2016/11/a-republic-of-emails-what-are-the-contents/
Once I entered my text the website created my “corpus.” The site took all the words that I entered and churned out a number of visual forms that analyzed my words. There is a huge list of ways that the site can organize your data but it gives you a default page with a few main options. The first is “cirrus” which is basically just a word cloud. It highlights the most repeated words in your text. For our class, the biggest word was “history” closely followed in size by “digital” (shocker right?!).
The second default tool is reader. This is self explanatory. It is just the text you originally input.
The third default is trends. This is a graph that allows you to visualize the distribution of the most popular words across the document.
The next tool is Summary. As the name suggests it provides an overview of the document. This includes the number of words, longest and shortest documents, the density of vocabulary, and the average number of words per sentence. This function is much more interesting in other corpuses (corpi????) which actually input documents separately. In class, I will show you the Jane Austen Corpus which does this quite nicely.
The final default is Contexts. This is a really neat feature. It shows you the frequently used words surrounded by just a little bit of the adjacent text to give you the idea of how the term is used.
In addition to the defaults, there is a whole list of other modes of analysis that can be applied to a corpus. One of my favorites is Bubblelines. This feature allows you to visualize both frequency and distribution of keywords.
I would say that our blog posts did not serve as a particularly useful or interesting topic for analysis but I honestly got so fed up in attempting to input digital texts that this was the most realistic option. The cool thing about this site is that you can actually access what other people have done. This is really awesome because of how difficult I found the data entry. This site is incredibly time intensive and very hard to use in my opinion.
BUT! This website can potentially tell you a lot and allows you to visualize. You can take a broad overview of the most used words or you can go in for close analysis like using the context tool to see how these words are being used. For comparing a large body of texts this site could be really fascinating IF they were online. I cannot imagine trying to input an entire book or anything larger than an article for this site. I do see the advantage if you are doing analysis of short texts like emails. The Hillary Clinton email project is pretty cool. They got to see how many times certain words popped up over the course of the emails and they could use all of that to make connections and determine significance.
Below is the link to my attempt at using Voyant Tools with our class blog posts.
For your consideration:
How can you see this data visualization impacting your own research? Is it useful to you?
Do you think the current search function on the Smithsonian website is a suitable substitute for the HistoryWired site?