The interplay of symbolic and adaptive techniques: two case studies

Two well known goals of artificial intelligence (AI) are those of computer vision and natural language processing. This paper discusses methods used in the application of adaptive and symbolic techniques to these problems. The first method discussed is applied to shape representation. The initial problem is one of AI and in order to find a solution we turn to maths. This leads to new problems that can be tackled via the use of AI techniques. These AI techniques can then be further refined via maths. This recursive use of one discipline to help solve problems in another has proved successful. The second method is used in natural language engineering. LOLITA is a large-scale domain-independent natural language processor that has been developed at the University of Durham over the last seven years. A largely symbolic approach has been adopted throughout development, and the system is now at the stage where it is to be optimised for particular domains. Adaptive technique are to be used for this optimisation