Banerjee: Adaptive Fuzzy Cognitive Maps Vs Neutrosophic Cognitive Maps

Introduction Knowledge management (KM) and artificial intelligence (AI) are interconnected disciplines to discern information for information management systems. Researchers have raised issues of knowledge that are living and active. Decisions based on real life knowledge bases are subjective judgments in nature. AI has well-developed cognitive tools that can process qualitative information of knowledge domains (universities, educational bodies, research laboratories, business enterprises and bureaucracy). Artificial Neural Network (ANN) is a simulation of human brain consisting of billions of neurons interconnected by network of synapses. Due to uncertainties involved in relationships, system cannot model human expert’s behaviour as the number of rules increases. Decision support system (DSS) tools should be equipped to model dynamically evolving knowledge through feedback mechanisms. This paper evolves a decision-making system using Fuzzy cognitive maps (FCM) for knowledge-based institutions. The paper showing FCM of a research institution encoded with symbolic input knowledge learns through selective interconnected alternatives and evolves its strategic decisions. Learning is finally generalized using Neutrosophic cognitive maps (NCM). Methodology Fuzzy Cognitive Map (FCM) Cognitive maps are a collection of causal nodes linked by arcs or edges. Nodes drawn as circle (Fig. 1) represent concepts (C I i=1,...., N), which are variables of problem domain. FCM is a fuzzy version of Axelrod’s cognitive maps. FCM combines ideas of fuzzy logic and neural network (NN) in a hybrid mode, wherein an organization(s) can be interconnected. In Axelrod’s cognitive maps, interconnections are crisp values [+1, 1]. In fuzzy version, connection weights are obtained from either fuzzy membership functions or fuzzified from crisp values. These weights (edge values) are posted along digraph arrows in the map. Causal influence between concepts can be negative, positive or none. Influences are expressed in fuzzy terms as weak, medium, strong, very strong etc. Concepts can assume any of three values: –1 (moderately on); 0 (off); or +1 (on). Inclusion of real values assigned to concepts has recently been made possible. FCM applications in knowledge organizations belong to business, stock investment and finanace disciplines in supervised mode. Present study describes FCM in unsupervised mode, which has relatively limited applications.