Computing a concept lattice from structured and fuzzy data

The problem of acquiring knowledge from a set of data is fundamental for humans to understand their experiences and to classify new observations. In this paper, we are concerned with acquiring concepts from data that may not be precisely or perfectly defined. We show the utility of lattice fuzzy sets to modelize uncertainty and imprecision for complex and structured data in rich domains such as Ecology or Botany. We propose an incremental algorithm to organize a given set of observations into a concept lattice. An application on data sponge illustrates the interest of the learning system proposed.