If a picture is worth a thousand words, how many is a GIF worth? We may be about to find out.
Two graduate students at the MIT Media Lab aim to create a text-to-GIF translator, as part of an ongoing project, one which could also reveal cultural differences regarding emotions.
[seealso slug="3d-gifs"]
Kevin Hu and Travis Rich created a website called GIFGIF, which presents users with two GIFs at a time and asks them to choose which better expresses one of 17 emotions (such as contempt, pleasure or surprise). There are currently approximately 4,000 GIFs in the GIFGIF system.
This process creates a collection of definitions that other programs can use to translate short bursts of emotional text, such as a Shakespeare sonnet, into GIFs. It doesn't quite give computers the ability to read emotions, but it comes close.
The GIF translator is just one potential outcome of the project. According to Rich, once enough data is stored, GIFGIF could help recommend a movie that corresponds to a specific mood based on a GIF. The two cofounders plan to create an open API so other researchers can access the findings.
The endeavor was in part inspired by the Place Pulse project, also an MIT Media Lab enterprise, which aimed to quantify people's perceptions of different cities. It also holds a resemblance to the now-defunct Google Image Labeler that asked users to describe pictures in order to improve the quality of the company's image search.