An approach to natural language understanding in rule-based systems

Aproved for pbli c iJ i 1 AFIT/GOR/ENS/88D- 12

Transformation units and transformation modules are structuring concepts for rule-based systems. Their semantics is described by binary relations on configurations which are obtained by the interleaving of rule applications and calls of imported transformation units. The composition

This paper describes a method of analysing rule-based systems, which models the procedural semantics of interpretation', the program, AbsPS, derives a description of the mapping between a rule base's inputs and outputs. In contrast to earlier approaches, AbsPS can analyse the effects of

Abstract—The exclusion rule-based system, proposed in the paper, is an alternative method of designing a rule-based system for an expert or a control system. The starting point of the approach is the set of all possible decisions the considered system can generate. Then, on the basis

Rule-based systems that are easily testable are re-quired for high reliability applications. However, as a rule base evolves, developers prefer incremen-tal evaluation owing to the high cost of regression testing. For, quality aatd reliability improvement, researchers advocate that the evaluation

In this paper, we propose new structuring concepts for rule-based systems that are independent of the type of rules and of the type of configurations to which rules are applied. Hence the concepts are applicable in various rule-based approaches allowing one to build up large systems from small

mathematical data from some special regions of the face. The extracted mathematical data are then fed to a fuzzy rule-based system. Fuzzification operation issues triangular membership functions for both input and output. The method is implemented on MATLAB. An Algorithm is developed which gives 6 facial

Set-oriented constructs for forward chaining rule-based systems are presented in this paper. These constructs allow arbitrary amounts of data to be matched and changed within the execution of a single rule. Second order tests on the data can be included in the match. The ability of a single rule

provides for systems better presentations of the knowledge, which is often expressed in terms of the natural language. Fuzzy rule based systems constitute an extension of classical Rule-Based Systems, because they deal with fuzzy rules instead of classical logic rules. Since bio-inspired optimization