Toward an Architecture for Never-Ending Language Learning

• DOI: 10.1609/aaai.v24i1.7519
• Publication Date: 2022-09-13T05:19:39Z
• Authors: Andrew Carlson; Justin Betteridge; Bryan Kisiel; Burr Settles; Estevam Hruschka; Tom Mitchell
• Keywords: Base (topology)

We consider here the problem of building a never-ending language learner; that is, an intelligent computer agent runs forever and each day must (1) extract, or read, information from web to populate growing structured knowledge base, (2) learn perform this task better than on previous day. In particular, we propose approach set design principles for such agent, describe partial implementation system has already learned extract base containing over 242,000 beliefs with estimated precision 74% after running 67 days, discuss lessons preliminary attempt build learning agent.