From Ontology Learning
Empirically Grounded Emergent Knowledge
Vít NOVÁCEK and Siegfried HANDSCHUH
Abstract. This chapter introduces a framework for representation and processing of emergent knowledge. By emergent knowledge we mean most primarily semantic patterns discovered within simple statements automatically extracted from textual resources. Empirical grounding of the emergent knowledge is achieved by applying the principles of distributional semantics in order to anchor the discovered semantic patterns in the textual data in a well-founded and non-arbitrary manner. We also propose a method for seamless combination of the distributional (bottom-up) and symbolic (top-down) aspects of the semantics of emergent knowledge. Broad applicability of our approach is showcased within several practical scenarios in the domain of life sciences.