Abstract
Innovations for a database thesaurus must be vigorously pursued in the next decade. Our project was intially intended to develop and artificial intelligence in forward inference for database query systems, but later it evolved to a thesaurus system which is equipped with a search engine composed of a framed knowledge base. This objective has been accomplished for a database of FEM (Finite-Element Method) literature accompanied by their mesh data, of which graphical output is viewable at a human interface. When we were analyzing maneuvers of the search engine among the network of keyword thesauri, the findings led us to to try applying this engine for a very similar tracking of a network of FEM meshes in view of its potential ability for mesh pattern recognition. Furthermore, the algorithm used to infer keywords in the thesaurus is adapted to recognize shapes of FEM meshes used in the analysis. These meshes are kept in a general-purpose RIQS database, where one might be able to find a particular mesh, the configuration of which might be similar to his own, or a mesh belonging to a specific classification. By replacing the thesaurus topology with a network of FEM meshes, the inference engine can retrieve elements or mesh contours by comparing them based on similarities or dissimilarities. The efficiency in terms of retrieving mesh objects has been measured by applying real FEM data in RIQS database.
