Temporal Processing
Inderjeet Mani · 2014
This chapter introduces approaches to the processing of time-related constructs in natural language. It covers information extraction systems that can interpret time expressions and create a chronology of events in a narrative, as well as systems that can generate such expressions and render an ordering of events in a fluent discourse. Annotation schemes for marking up time expressions and temporal orderings are introduced, along with automatic methods for computing such annotations. The representation of annotated temporal relations in the interval calculus is described, as is the problem of inferring a consistent set of temporal relations for a narrative. The accuracy of humans and machine annotation is examined, and a question-answering application is discussed.