To test out some of the visualization tools available, I first started playing with Voyant. This dataset represents every full text Apple Ad that I currently have for personal computers (Apple II, Macintosh, iMac G3, and iMac G4).
Pros of this tool:
- Extremely useful way to visualize word frequency and total word count, as well as distinct words and phrases across different documents.
- Can search for word frequency of specific words and phrases, which helps respond to page of descriptive language that we built.
- Seems to be embeddable in the WordPress site.
Cons of this tool:
- I can’t find a way to import metadata along with the full text, and Voyant doesn’t recognize the full text transcripts in the format that we have them. I had to copy and paste each block of text and save it as a .txt file, which stripped it of all of the useful metadata that we’ve been collecting and organizing.
- Individual visualizations (the graph, the table of most frequently used words) seem to embed fine, but the full interactive site does not.
- I am not sure how to link between the full text in this tool and the full text column in our Google Sheet, which renders this visualization static rather than interactive.
The second toolset I played with was Palladio. I think that Palladio will be an extremely useful tool.
Pros of Palladio:
- It allows us to keep all of our metadata and visualize connections by nodes.
- It also allows for geographic visualization, which is something I find particularly interesting.
- Finally, it has a “timeline” filter, which is particularly useful.
- It seems embeddable as a .svg and a .json file.
Cons of Palladio:
- Palladio seemed to recognize the entire “key images” column as one individual item rather than a list of items. It would be nice to find a way to recognize each item in my list, so that we could visualize connections between key images.
Unfortunately, I cannot get the .json or the .svg file to upload and display correctly on this page, so here is a screenshot:
I also decided to start playing around with some of the geospatial tools that Andrew Battista had suggested looking into. The first one I started playing with was CartoDB. I looked through all the ads in the database and created a separate related table of specific geographic mentions (either in the text or as an image shown on screen) and the associated keywords/phrases/text. I then plotted them using some of the metadata that I have for the ads in the main table.
The first map is an intensity map of mentions:
CartoDB can generate some really interesting maps based on the metadata that we have, such as sorting by specific categories (company, here):
Or by object, here:
We can also show change over time:
Finally, we can also filter by the metadata categories we import, so here, for example, I have filtered to only show products over time made by Apple:
Cons of using CartoDB:
- The user experience is limited by the options we provide: on this filtered map, I don’t think that the user can change the settings we have selected to display.
- These are strictly geographic data (latitude, longitude, correlated with our metadata), so we are completely missing out on the ability to navigate between ads and see connections. I am also unsure of how to link the ads back to another section of our site.