A Real--Time and Any--Time Approach for a Dialog System

Abstract

In this paper we present a knowledge based approach to a speech understanding and dialog system which possesses any--time and real--time capabilities. Knowledge is represented by means of a semantic--network formalism; the concept--centered knowledge base is automatically compiled into a fine--grained network which allows an efficient exploitation of parallelism. The interpretation problem is defined as a combinatorial optimization problem and solved by means of iterative optimization methods. By using iterative methods, any--time capabilities are provided since after each iteration step a (sub--)optimal solution is always available. At the moment, the real--time factor for interpreting an initial user's utterance is 0.7. 1 INTRODUCTION Automatic recognition of complex patterns, specifically motion sequences and spontaneous speech,become increasingly relevant for a great number of applications, such as service robots to aid handicapped individuals, multi--modal telecooperation tasks, ...