Professor quantifies how 'one thing leads to another'

Aug 15, 2014 by Anne Ju

(Phys.org) —"One thing led to another," people often say. Events, discoveries and relationships are triggered by something previous. The iPhone case was designed only because the iPhone was invented first. A song became popular only after someone liked it.

Mathematicians always want to quantify things, and Steven Strogatz, Cornell's Jacob Gould Schurman Professor of Applied Mathematics, is trying to quantify the commonsense concept of "correlated novelties" – that one new thing sometimes (but not always) triggers another. He and three researchers from Italy's Sapienza University, where he spent three months on sabbatical leave, published a paper in Scientific Reports July 31 on their first crack at it.

"This concept of correlated novelties had only been phrased in verbal or conceptual terms," Strogatz said. "It seemed like a really intriguing idea, with room for creativity to make it testable."

Correlated novelties are everywhere: in economics, or perhaps most obviously, in , where traits evolve over time from a cascade of tiny events.

In this era of big data and supercomputing, going back to find patterns in outcomes is easier than ever. Strogatz and his colleagues used four data sets as test cases, each consisting of a sequence of elements ordered in time: texts, where the elements are words, and the novelty occurs when a word appears for the first time; online music, where the novelty is the first time a listener hears a song or artist; Wikipedia, where novelties are the first edit of a given wikipage; and social annotation in tagging sites, where novelties are the introduction of a new tag.

The researchers quantified the rate at which novelties occurred by focusing on the growth of distinct elements in an ordered sequence. They discovered that the rate at which novelties occur decreases over time.

They proposed a simple mathematical model that mimics the process of exploring a physical, biological or conceptual space that enlarges whenever a novelty occurs, according to the paper. The model is based on Polya's Urn, familiar to mathematicians. An urn filled with black and white balls churn like lottery balls; whenever one color is picked, more of the same color are added back. That enhances the likelihood of that original color being picked at the next turn.

The concept has an echo of the familiar, Strogatz says. For example, the urn could be considered a simplification of the "rich get richer" concept – that socioeconomic stratification isn't random, but strongly correlated with previous events.

Such insights hint at the possibility of a sophisticated building on the one proposed to address an endless list of human problems in economics, biology and epidemiology, to name a few. For example, applying the model to 100 years of influenza data, epidemiologists could perhaps learn to predict how the flu will spread going forward, Strogatz said.

Explore further: UMSL scholar examines evolution of learning

More information: "The Dynamics of Correlated Novelties," Scientific Reports, 2014.

add to favorites email to friend print save as pdf

Related Stories

UMSL scholar examines evolution of learning

Aug 14, 2014

Why do monkeys learn to be afraid of snakes and not flowers? Is this knowledge the result of evolution by natural selection? Did the monkeys that couldn't learn that association quickly die and not reproduce?

A vexing math problem finds an elegant solution

Nov 14, 2013

(Phys.org) —A famous math problem that has vexed mathematicians for decades has met an elegant solution by Cornell researchers. Graduate student Yash Lodha, working with Justin Moore, professor of mathematics, has described ...

Recommended for you

Entrepreneurs aren't overconfident gamblers

2 hours ago

Leaving one's job to become an entrepreneur is inarguably risky. But it may not be the fear of risk that makes entrepreneurs more determined to succeed. A new study finds entrepreneurs are also concerned about what they might ...

New branch added to European family tree

4 hours ago

The setting: Europe, about 7,500 years ago. Agriculture was sweeping in from the Near East, bringing early farmers into contact with hunter-gatherers who had already been living in Europe for tens of thousands ...

User comments : 1

Adjust slider to filter visible comments by rank

Display comments: newest first

antialias_physorg
not rated yet Aug 15, 2014
Such insights hint at the possibility of a sophisticated mathematical model building on the one proposed to address an endless list of human problems in economics, biology and epidemiology, to name a few.


I don't get it. It sounds like they are reinventing the wheel (statistics).