Rocket: Robust Knowledge and Concept Extraction from Text


Rooms 301 and 302 August 22, 2016 12:15 pm - 12:45 pm

Bookmark and Share

Shi Yu, American Family

Shi Yu, American Family

Download Presentation

In this talk I will describe a machine learning, natural language processing framework
that allows the automatic extraction of key relevant information from unstructured text. The
approach combines several state­of­the­art text representations and  uses an active learning/
crowdsourcing  paradigm to interact with users to optimally label/tag the concepts or assertions
of interest to be learned. The resulting system can be used to learn concepts form any domain
with few labeler domain expertise/ interaction. We will show some use case relevant the
insurance domain.