Uma interface para consultas a sistemas de suporte à decisão baseada em linguagem natural, casamento de padrões e metadados.

Abstract

The widespread use of the Word Wide Web and Internet has been accompanied by the development of technologies that enabled the emergence of web-based decision support systems. This new paradigm has reduced technological barriers and made it easier and less costly to make decision-relevant information available to all types of users in a company. However, traditional user interfaces specially designed for professional users will not work well for casual and non-specialist users. This dissertation presents a new interface to webbased decision support systems which is based on restricted natural language, pattern matching and metadata. This new architecture was designed with the aim of assisting users without technical instruction with the task of retrieving decision-relevant information. The proposed solution shows that an alternative and simpler approach to the problem of natural language interfaces to databases can achieve useful results. The architecture relies on the multidimensional model and dimensional modeling to provide a transparent and natural database access for non-specialist end-users, and address the domain portability problem.