Early search engines answered queries by listing the titles of
documents related to the query. The obvious limitation was that a
user had to open each document to assess its relevance. The situation
was greatly improved with introduction of snippets. In the quest for
semantic search we need to address the same problem in different
settings. Semantic search engine NAGA can answer queries such as '$x
isCitizenOf Switzerland', but the list of answers contains only names.
Usefulness of NAGA would be greatly enhanced by semantic snippets,
which could express information like: "Felix Bloch was a physicist. He
investigated nuclear magnetic resonance. In 1952 he was awarded Nobel
Prize in Physics." Since the knowledge stored in ontologies such as
YAGO is very different from collections of documents, we need to
develop new techniques for summarization.