"I believe AI has an opportunity to achieve a true breakthrough over the coming decade by at last solving the problem of reading natural language text to extract its factual content. In fact, I hereby offer to bet anyone a lobster dinner that by 2015 we will have a computer program capable of automatically reading at least 80% of the factual content across the entire English-speaking web, and placing those facts in a structured knowledge base."
--http://www.ai.rutgers.edu/aaai25/mitchell.htm
I saw an on-going project recently that was attempting just that. They were using texts from the web to infer relationships from recognizable patterns. Like (the head of X, Y) implies X is an organization, Y is a person, and Y is the leader of X.
It would also try to learn new patterns from facts it already knew, like if it notices Jobs and Apple in a sentence it can hypothesize that sentence pattern is about a leader of a company.
For the life of me I cannot remember the name of the project or even the university, hopefully someone else has heard of this.
