UMBC ebiquity
Presentation
Sixty years of knowledge graphs for language understanding
September 19, 2019
1780201 bytes
PDF Document - Need a reader? Get one here
There is a long history of using structured knowledge of one kind or another to support AI tasks, especially ones involving natural language understanding. Over the years, the names and details have changed, from semantic networks to frames to logic programs to databases to expert systems to knowledge bases to the semantic web and currently to knowledge graphs. However, a common thread is that an organized representation of knowledge that can be queried and evolved is a core component of many intelligent systems. Current research is exploiting today’s two significant technology trends: the ready availability of enormous amounts of content in the form of text, data, images, and videos, as well as powerful new machine learning technologies that can take advantage of it. After briefly reviewing this history, I’ll give examples of combining KR and ML from our recent work.
Keynote presentation at the Knowledge Representation & Reasoning Meets Machine Learning workshop held at MIT on September 19, 2019
70 downloads
Public