Universal Schema for Slot Filling and Cold Start: UMass IESL at TACKBP 2013.
Sameer Kumar Singh, Limin Yao, David B. Belanger, Ari Kobren, Sam Anzaroot, Mike Wick, Alexandre Passos, Harshal Pandya, Jinho D. Choi, Brian Martin, Andrew McCallum · 2013
We employ universal schema for the TAC KBP slot filling and cold start tasks. The technique enlarges the set of relations in an ontology, e.g., TACKBP slots, to con-tain all surface patterns between pairs of entities in a large corpus. By factoriz-ing the matrix of co-occurrences between entity pairs and universal schema rela-tions, we are able to predict new target slots. This differs fundamentally from tra-ditional relation extraction approaches be-cause an entire knowledge base is con-structed jointly over train and test data. To produce submissions for the slot filling and cold start tasks, we simply query this knowledge base. We describe universal schema, our data preprocessing pipeline, and additional techniques we employ for predicting entities ’ attributes. 1