Uso da lógica paraconsistente bivalorada no tratamento de respostas de usuário em um sistema especialista

Abstract

The Expert Systems (ES) assist in decision-making tasks and problems solution in a specific area. They can be implemented in environments (shells) having an inference engine for forward chaining or bacward chaining . In forward chaining the process starts by the conditions of a rule, unlike the backward chaining unlikewhere you begin to prove from its conclusion. When using the environment with the SE rules codified, the user will answer a set of questions to the shell search your knowledge base one or more true answers and display as a research tree. Not always the user has the absolute certainty of what responds and use a classical logic can answer only yes or no. There are some studies in the literature for the treatment of uncertainty in expert systems using non-classical logics such as: Fuzzy and Paraconsistent, but they do not address the issue of uncertainty in user response. This paper applied the paraconsistent logic with two values for the treatment of uncertainty informed by the user to answer a set of questions derived from a SE. A system to receive as input was developed questions, rules and search tree, is able to calculate the degree of belief and unbelief: a rule or final answer brought by SE. The SE system used as the experiment was the selling price of training that aims to inform the user which of the pricing methods are most suitable for your business, taking into account your answer on the questions of questions. The SE used was implemented in the shell Expert Experience. Analysed the difficulties in applying the fuzzy logic that have more than two variables in the composition of the rules and the use of Expert Experience to incorporate uncertainty calculations based on non-classical logic too.