Application of machine learning methods for filling and updating nuclear knowledge bases

The paper deals with issues of designing and creating knowledge bases in the field nuclear science technology. authors present results searching for testing optimal classification semantic annotation algorithms applied to textual network content convenience computer-aided filling updating scalable repositories (knowledge bases) physics power engineering and, future, other subject areas, both Russian English. proposed will provide a methodological technological basis problem-oriented as artificial intelligence systems, well prerequisites development technologies acquiring new on Internet without direct human participation. Testing studied machine learning is carried out by cross-validation method using corpora specialized texts. novelty presented study lies application Pareto optimality principle multi-criteria evaluation ranking absence priori information about comparative significance criteria. project implemented accordance Semantic Web standards (RDF, OWL, SPARQL, etc.). There are no restrictions integrating created third-party data metasearch, library, reference or question-answer systems. software solutions based cloud computing DBaaS PaaS service models ensure scalability warehouses services. public domain can be freely replicated.