Grammar-free approaches to natural language processing work without the need for a formal grammar. Techniques for doing this include template matching often used in ChatBots, and keyword matching. For example, a telephone help system might ask the caller to describe theor problem and then route it to an appropriate operator depending on certain keywords being detected. Statistical techniques such as Markov modles may also be used to create predictive models for disambiguation or to create POS taggers. Lareg-language mdoels are an extreme version of this.
Used in Chap. 13: page 216