Send us a comment.
9.1 Question answering 9.2 Evaluating DBQ formulae 9.3 Standard logical inference 9.4 Implementing forwards inference in Prolog 9.5 The pathological nature of logical inference 9.6 Primitives and canonical forms 9.7 Classes and inheritance 9.8 Plausible inference and defaults
Inference mechanisms can play a number of useful roles in language processing systems. For instance, they can help a system to:
Answer questions whose answers are distinct in form
from any single piece of information that the system already possesses,
Make predictions about possible future natural language inputs.
Make decisions about assimilating new information into existing
beliefs.
Solve problems about how the world does or might behave.
Detect that two statements are semantically equivalent or that a
statement is anomalous.
In this chapter, we concentrate mainly on the kinds of inferences that a language processing system might be required to make, generally illustrating the ideas by referring to question-answering systems. It is with such systems that our discussion begins. In Chapter 10, we will consider other uses of inference for the processing of natural language in context.
Send us a comment.