Stanford's Distantly-Supervised Slot-Filling System

摘要

This paper describes the design and implementation of the slot-filling system prepared by Stanford's natural language processing group for the 2011 Knowledge-Base Population track at the Text Analysis Conference. Our system relies on a simple distant supervision approach, using mainly resources furnished by the track's organizers: we used slot examples from the provided knowledge base, which we mapped to documents from several corpora-those distributed by the organizers, Wikipedia, and web snippets. This system is a descendant of Stanford's entry into last year's evaluation, with several improvements: an inference process that allows for multi-label predictions and uses world-knowledge to validate outputs; model combination; and a tighter integration of entity coreference and web snippets in the training process. Our submission scored 16 F_1 points using web snippets and 13.5 F_1 without web snippets (both scores are higher than the median score of 12.7 F_1). We also describe our temporal slot-filling system, which achieved 62 F_1 points in the diagnostic temporal task on the test queries.