Social-Behavioral Modeling for Complex Systems

20 Social Media Signal Processing

Prasanna Giridhar and Tarek Abdelzaher

Computer Science Department, University of Illinois at Urbana–Champaign, Champaign, IL, 61801, USA

Social Media as a Signal Modality

Posts on social networks collectively comprise a new type of indexing into physical reality, social beliefs, concepts, biases, and ideas (Levy 2013). A logical question becomes: can one develop an instrument to browse this reality, a new macroscope into world state? One purpose of such a device would be to reliably observe physical and social phenomena at scale, as interpreted by the collective intelligence of social media users.

This chapter describes computational insights that give rise to information processing algorithms for such an instrument. Engineers have long investigated how signals propagate through noisy channels. For example, engineers study how AM/FM radio transmissions propagate through air, walls, and metal, how vibrations travel through terrain, and how acoustic waves travel through physical matter. Signal loss and distortions are introduced. Such loss and distortions change or bias the received signal. An understanding of signal emission and propagation properties can simultaneously help reconstruct both a good approximation of the original transmitted signal as well as a model of the introduced perturbation or bias. Can one apply the same wisdom to social sensing to reconstruct both physical reality and human biases from posts propagating on ...