Never-Ending Learning for Deep Understanding of Natural Language
Tom M. Mitchell · 2017
Abstract : This research has explored the thesis that very significant amounts of background knowledge can lead to very substantial improvements in the accuracy of deep text analysis and understanding. To explore this thesis we have built on our earlier research on the Never Ending Language Learning (NELL) computer system, which has been running non-stop since January, 2010, learning to read the web, and automatically constructing a large knowledge base (aka knowledge graph) by extracting structured factual assertions from unstructured text on the web.