Нечіткі дані у семантичних wiki-ресурсах: моделі, джерела та методи обробки

Abstract

We analyze main types of dirty data processed by intelligente information systems, criteria of data classification and means of detection non-classical properties of data. Results of this analysis are represented by ontological model that contains taxonomy of classical and nonclassical data and knowledge-oriented methods of their transformation. Special attention is paid to semantically incorrect data that corresponds to vague knowledge. This ontological model intended to provide more effectively methods for transforming raw data into smart data suitable for automatic analysis, knowledge acquisition and reuse in other information systems. The ontological approach provides integration of the proposed model with other external ontologies that formalize characteristics of various methods and software tools that can be used fo data analysis (data mining, inductive inference, semantic queries, and instrimental tools for testing various aspects of the ontology quality, etc.). The work uses the experience of knowledge base developing of the portal version of the Great Ukrainian Encyclopedia e-VUE. This information resource is based on the semantic Wiki technology, it has a large volume, a complex structure and contains a large number of various heterogeneous information objects. Wiki resources are interesting from the point of view of collaborative processing the fuzzy data that describe heterogeneous information objects and knowledge structures. Due to the fact that the creation of this information resource involves a large number of specialists of various scientific fields, who have different areas of expertise and qualifications in use of knowledge-oriented technologies, there are many differences in the understanding of the rules for presenting and structuring data, and therefore a significant part of the Encyclopedia content needs additional verification of its correctness. Therefore, we need in formalized and scalable solutions for detection and processing various types of inconsistence, incompleteness and semantic incorrectness of data. The proposed approach can be useful for the creation of other large-scale resources based on both the semantic Wiki technology and other technological platforms for collaborative processing of distributed data and knowledge.