APPLICATION TYPES - DISCOVERY

The hypothesis formation and evaluation process in EFAR is strong aid for discovering complex behaviours in large data. Whether the analysis is started via exploration or initiated by a set of observations, the technology allows analysts to incrementally and steadily conjecture and verify phenomena and behaviours which are too complex to uncover through one-shot data-mining techniques and without the strong support of domain expertise.

EFAR-based discovery techniques have been or are being experimented in two aplication domains:

1. Discovery of behaviours in complex technical systems: Analyses of recorded data from complex technical systems often prompt deeper forays into the data repositories to better understand observations and singularities. Where such observations are considered highly relevant, experts like to develop wider and consistent characterisations of the behaviours that surround them.
   
   EFAR technology is able to support this goal by allowing experts to develop knowledge-based suppositions which are then confirmed or refuted in data. More specifically, EFAR technology has proven its advantages in characterising combinations of conditions that lead to system/component malfunctioning. By being able to single out, extract and correlate technical parameter behaviours which recur in a consistent manner, experts were able to map them back into the system knowledge domain and to develop meaningful explanations for new types of system behaviours.
   
2. Discovery of phenomena in economic and financial data: Economic and financial processes are constantly scrutinised to gain a better understanding and to enable predictions. EFAR technology provides opportunities for a deeper and more detailed insight into data, as it enables the widening of initial suppositions through knowledge-driven searches and assessments in large data volumes.
   
   EFAR was tested in the analysis of the quality of decision-making of financial decision-making actors against a backdrop of economic data. EFAR allowed to develop observations supported by reliable associations with events gleaned from economic data and to detect and describe specific decision-making behaviours of the economic actors. The technology proved to be a powerful tool in detecting and explaining actors' decisions in a bottom-up manner, i.e., from conditions to decisions, as opposed to a top-down justification approach, from decision to presumed causes.