Yahoo's Developing a Map Algorithm to Find the Most Beautiful Route

Sometimes you want to get from A to B as quickly as possible—but what if you want to savor the journey? Well, Yahoo has been working on an algorithm that can plot the most beautiful route, for those days when time is less of a concern.

In a research paper published to arXiv, Daniele Quercia from Yahoo Labs explains how a new algorithm could optimize directions for beauty. "The goal of this work is to automatically suggest routes that are not only short but also emotionally pleasant," he writes.

So, how does it work? Well, first Quercia and his team created a database of images, taken from Google Street View and Geograph, of the center of London. Then, they crowdsourced opinions about how pleasant the parts of town were relative to each other via users see two images and have to choose which one is nicest. It's Hot or Not for cities.

Armed with that data, the team plotted heat maps, showing how the beauty of London varied with geographic location. From there, it was fairly straightforward to create an algorithm that found the most attractive route rather than the fastest: it searches through the possible routes typically suggested, but adds attractiveness along each one and then chooses the one with the highest score.

The results tend to be about 12 percent longer than the shortest route. In a slightly-less-than-scientific test, Quercia had 30 Londoners follow her suggested paths, and they agreed that the routes were indeed more attractive than the shortest.

The team has gone a step further, though, developing an automated beauty analysis instead of crowdsourcing it—the idea being that the whole system could be automated and computerized. So, they took five million images from Flickr, taken in the same place as the original images they used, and mined their tags to understand what counted as beautiful.

They then applied their insight to images captured in Boston, creating a new, computer-generated beauty heatmap for the city. Then, they searched for beautiful routes through the city and once more had 54 volunteers try them out. Again, slightly unscientifically, the participants agreed the routes were more beautiful than the shortest ones.

Sure, there are issues with this work. First, beauty is subjective, and testing the algorithms isn't particularly rigorous. More than than that, beauty can vary by the day or even hour: some of these location may looks wonderful at sunset but suck during the morning rush hour, say. But small issues aside, it's a neat idea—and one that Quercia is soon to turn into an app. Get ready for your travels to get a little prettier. [arXiv via Technology Review]