Distinguished Lecture, October 9, 2006

Human-Centric Constructs of

Granular Computing and Fuzzy Logic

Witold Pedrycz

Department of Electrical & Computer Engineering

University of Alberta, Edmonton, AB

Systems Research Institute, Polish Academy of Sciences

Warsaw, Poland

Abstract

Human-centric systems and human-centric computing (HC²) that constitute an innovative and visible trend in modern information technology are inherently associated with the usage of heterogeneous and highly distributed data. The data usually come from a broad range of sources including users, designers, networks of sensors and distributed databases. Various pursuits along the line of e-society including intelligent housing, semantic web and web intelligence, e-health, e-commerce, intelligent data analysis, and wearable hardware are examples of the tendency of the highly collaborative processing present in the development of HC² systems.

In contrast to numeric computing, human-centricity is directly associated with the processing of information granules (and fuzzy sets and their higher-order constructs, in particular) – semantically meaningful information constructs being used in representing and communicating knowledge about the problem at hand. Constructing information granules on a basis of distributed data sources (data nodes) which could be accessed only on a local basis constitutes a genuine challenge yet opens up new and highly promising avenues of knowledge-based processing.

We discuss various aspects of human-centricity in the setting of the design of granular models and demonstrate how the topology of such models hinges upon collections of information granules and associations between them. In this talk, we also propose a collaborative scheme of forming information granules whose crux lies in the formation of interaction linkages between the data nodes that are established at the granular level. A general taxonomy of the interaction mechanisms (emerging under the umbrella of so-called C³ paradigm) is introduced and studied. In the sequel, we look at a suite of algorithmic developments that support the realization of this paradigm.