Editorial: Hybrid Techniques in AI

To enhance knowledge-base capability in terms of handling knowledge with inconsistency, uncertainty, incompleteness and irregularity, traditional techniques from Artificial Intelligence (AI) for knowledge representation, reasoning and learning have conspicuous difficulties in different areas of AI research. Researchers in AI are striving towards combining techniques from other domains to combat these difficulties. In general, AI techniques can be grouped into two broad classes—traditional or hard AI and non-traditional or soft AI techniques. Hard AI refers to the more traditional AI techniques, i.e. classical methods such as knowledge representation, rule-base systems, tree search methods and pattern matching. On the other hand, soft AI techniques deal with approaches to problem-solving which are based on metaphors from nature, including neural networks, genetic algorithms, simulated annealing and fuzzy logic. Knowledge representation and reasoning are multidisciplinary subjects that apply theories and techniques from many other different fields but they all follow three basic steps: (i) Logic provides the formal structure and rules of inference, (ii) Ontology defines the kinds of elements that exist in the application domain, and (iii) Computation supports the applications that distinguish knowledge representation from pure philosophy. A combination of hard and soft techniques can enhance these three fundamental steps. Hybrid techniques will thus be able to improve performance in terms of knowledge representation, fast reasoning, and handling uncertainty in a significant way. The purpose of this special issue is to explore hybrid techniques in knowledge representation, reasoning and learning in AI and for enhancing