Inaugural Address

The midterm election is approaching, and the mass media is flooded with all kinds of political messages we can’t unsee. Election is an interesting activity. When we choose a policy or another, it seems that we tend to be motivated by the belief that we have built up throughout our life or even others’ opinions.

For the Data Art text analysis project, James, Heng and I decide to go beyond the illogical emotions and analyze the inaugural addresses of all the U.S. presidents, which represent the sum of people’s hope. The inaugural addresses not only bear people’s expectation but also represent the presidents’ vision of America and their goals for the nation. Some of the most powerful speeches are still quoted today. Thus, we think it would be fun to make a search engine to let people search and compare the frequency of a specific word mentioned across all inaugural addresses.

The design of the search engine is simple. It has a dominant text input box waiting for you to type in any word. The search result is made up of vertical bars which represents how many times the word is mentioned in each inaugural address, and the height of the bars is based on the usage frequency of the word. The text-to-speech is integrated with the result. We choose to use the Russian female voice to pronounce the word along with the vertical bars, and the pitch of the voice changes according to the bar height.

Also, inspired by the One Angry Bird project, we want to give each word an emotion. One Angry Bird uses artificial intelligence to detect facial expressions of the presidents while speaking in inaugurations. As for ours, we also use AI but in a much smaller scope. We predefine five different emotions, which are happy, surprise, angry, sad and hope, and then use Word2Vec with a trained dataset to calculate and see which distance between the searched word and emotions is the closest. Once we get the emotion of the word, the color will be applied to the whole background accordingly. As for how we decide the color, we reference it with an online project called Cymbolism.

In the end, I find it’s fascinating to see that there are words used constantly, used only once, or in an echo of a specific issue in a specific era. Here you can find more technical details of how this project is made: https://github.com/pondjames007/DataArt/tree/master/hw2_InaugurationAddress.