Public News Network D3 Sentiment research was performed during the Summer of 2015. Since 2003 I have been recording the major network evening news television broadcast. Depending on availability, networks such as ABC, CBS, NBC, PBS and FOX were recorded and a custom application I wrote in Java was utilized to extract FCC required transcript data.
My goal was to see if there was a tool that would allow me to visualize the tenor/tone of the broadcasts I had recorded and correlate this with what we know about the various networks and the historical events that have occurred during this period. I decided to use the popular D3, or Data-Driven Documents as primary visualization tool, combined with an existing Drupal database I have maintained, as well as the Alchemy API as a natural language processing tool for the extraction of “sentiment” from the transcripts I have captured.
6107 transcripts from 2003 - 2014 were analyzed using Alchemy API. Each was graded with a sentiment score from -1 to 1, measuring the general negative to positive tone of the broadcast. These values are stored in a database and can be viewed in raw form or in a table structure on the http://emeragency.electracy.org website. The D3 visualization makes calls into the database to populate a time-based, scaleable plot of the data. Buttons for each of the networks, allow the user to filter the information and make comparisons. Clickable icons pop up links to the transcripts on the website, along with specific broadcast information, keywords, and recorded video. All of the transcripts were automatically tagged with keywords allowing the user to filter broadcasts based on topic.
Unfortunately, at about this time, cable networks moved from over-the-air broadcasts which required embedded closed caption data, to encrypted transmissions that no longer allow the tactical-media artist or citizen journalist the ability to work in this way. I could no longer afford to maintain the streaming video server that was used in the project, so that component is no longer working. In the end, however, the research has provided valuable insight that will continue to inform my work.
Medium: networked art, data visualisation