Many large-scale Knowledge Bases (KB), such as Yago, Wikidata, Freebase, and Google’s Knowledge Graph, have been build by extracting facts fro structural Wikipedia and/or natural language Web documents.

The main observation of using knowledge base is that not all facts are useful and have enough information. To tackle this problem I will introduce how we build various data sources to help facts and keywords selection. We will also discuss important questions of KB applications including, - architecture of a KB processing and extraction system using Wikipedia and two public available KB including Wikidata and Yago; - method for calculating contextual relevance between facts. - how to present different facts to users.