Special Sessions


Emergent Phenomena and Big Data*
sponsored by the Intelligence Advanced Research Projects Activity (IARPA)

Challenges to the security, health, and sustainable growth of our society keep escalating asymmetrically due to the growing pace of global change. The increasing velocity and volume of information sharing, social networking, economic forces, and environmental change has seemingly resulted in an increase in the number and frequency of "game-changing moments" that a community can face. Now more than ever, we need anticipatory reasoning technologies to forecast and analyze potential change points in order to secure and improve our way of life and the environment we inhabit. Examples of relevant big data include social media, financial transactions, internet traffic, high throughput biomedical repositories, publication databases, extensive climate data, geophysical data, and geospatial and space data collections, all of which provide a unique resource to develop such technologies. Early detection of emergent phenomena of national security importance may require the detection of complex interactions of independent parts that are not analyzable simply as the sum of their effects. The ability to leverage interdependencies across massive datasets to capture the complexity of global information may enable a paradigm shift in information analytics by helping discover emergent phenomena before they escalate into serious predicaments.

The goal of this session is to explore how big data is impacting the way we use information to reason about change. Research papers and talks are invited that discuss detecting and characterizing emergent signals from regularly updated, very large scale information sources (big data). Research papers from a wide variety of systems and domains are sought and are asked to address any combination of the following aspects of emergent phenomena detection:

  • Theoretical or empirical definitions of emergence that enable early warning detection of relevant phenomena
  • Real time detection of emergent signals from big data streams
  • Detection of emergent signals, events, or features that are only apparent through fusion of multiple data streams
  • Measurement of the quality of "emergent phenomena detection" algorithm performance, e.g., precision/recall, ROC analysis, or other more innovative approaches
  • Human computer interaction that enables emergent phenomena detection

* Paper submissions intended for this session should clearly state so by including the session title "Emerging Phenomena and Big Data" as a subtitle.