Motivation

The age of big data has taught us the “network perspective”: that the connections between things are often as interesting as the things themselves. Network analysis involves a broad range of tools that take the network perspective, siphoning streams of meaning from what is otherwise a firehose of information. The tools we are building on Osiris deal with areas like the following:

● Social network analysis and visualization, where graph algorithms from mathematics are used to describe the shape of a network as a whole and properties of the parts like centrality. This allows us to see, for example, how well network members communicate with each other, and to infer who the most influential communicators are.

● Probabilistic graphical models, where algorithms from probability and statistics are used to describe causal relationships, so we can tell, for example, the webpage that people would be most likely to want to visit or the best way to treat a patient’s illness given all the facts we know about the patient.

● Network evolution, where the dynamic unfolding of network relations over time is studied using evolutionary computation and neural networks that both generate and predict network outcomes. Particular attention is paid to what causes growth and decay in networks, so we can predict, for instance, what goods and services will be in demand next year in a particularly competitive market.

● Networked artificial intelligence, the study of cooperative and competitive connections between distributed artificial intelligence programs and the processes by which these algorithms self-organize into better solutions. We use the principles of distributed AI not only to design the Osiris dynamics, but also as a tool to save human labor and make our AI programs serve our customers more effectively.

● Agent-based simulation of complex adaptive systems, the emulation of the virtuous and vicious feedback cycles in real-world systems to find the best policies to achieve goals. For example, we may want to explore ways to break the vicious feedback cycles of corruption in our society, develop an alert that the housing market is in a bubble, or emulate symbolic interactionist social feedback in Sophia the robot.