The Hummingbird Charm system is built on three systems: 1. an automatically generated knowledge base that is continuously updated. 2.  hierarchical ,  querysets semantically expanded over the knowledgebase and 3. Content tagged automatically and qualitatively ranked by crowdsourcing.

Reference librarians provide a human query interface to the information they curate.  Search engines replaced part of that human query interface with keyword search of online documents.  Keyword search has allowed the ability to search through vast quantities of information and provide a PageRank, added by Google.  The term frequency-inverted document frequency algorithm then provides a relevance ranking for a oline resource given keyword query terms.  In 2013 Google announced the Hummingbird Algorithm, which expands the keyword query against the Google Knowledge Graph to provide a semantic meaning to terms.  This increases the ability to find additional relevant content.

Sowa:  Ontology, Concept Graph

Etzioni: Machine Reading 1) web 2) academic papers

TAC: Evaluation Frameworks

Automated Knowledge Base Construction (AKBC) has become tractable in recent years.  Pre-processing data, extracting information (features), factor graph generation, and then statistical inference.  Knowledge bases are the