The botanical identification of species native of the Amazonian is an integral part of any forest inventory, mandatory for forest management plan and, therefore, essential for the scientific community to know more and better the Amazonian forest. However, the usual process of botanical identification is based on the empirical knowledge of native people from the forest (Bushmen) that use vernacular names to identify species, which in turn do not match the scientific names cataloged by taxonomists. Having this problem as the research scenario, this work proposes a conceptual model based on a semantic referential to support the process of identification of botanical species in the Amazon, helping reducing the knowledge mismatch between taxonomists and foresters, and consequently increase the accuracy of the current identification method. Semantic resources (e. g. ontology and semantic vector) are used in the formalization of captured knowledge. Two application scenarios are used to assess this work, namely: (i) the Forest Inventory that uses as an evaluation tool the specialist system for botanical identification by characteristics; (ii) the Image Timber that uses as an evaluation tool the expert system for image classification of wood. As part of the results, these scenarios use the pattern recognition to support decision making by providing computational tools to aid the process of identification of forest species marketed in the Amazon, with 65% accuracy rates in wood images. Final conclusion is that the semantic reference proposed in this work brings a relevant contribution regarding the production of knowledge about the Amazon area.
|
