Natural Language Generation
Natural language generation is a relatively new field in computer science. The two main questions to be solved are "what to say" and "how to say it." What to say includes answering questions not specifically asked by the user, and remembering previous statements made by the user and by the program itself. How to say it involves construction of logical sentences, use of pronouns, and sentence fragments. When speech synthesis is used, it also involves producing intelligible word rhythm, emphasis, and pauses. A new, exciting extension of speech synthesis is the inclusion of "emotion" in speech.
Men are all alike
In what way?
They’re always bugging us about something or other
Can you think of a specific example?
Well, my boyfriend made me come here.
Your boyfriend made you come here.
He says I’m depressed much of the time.
I am sorry to hear you are depressed.
It’s true, I am unhappy.
Do you think coming here will help you not be unhappy?
I need some help; that much seems certain.
What would it mean to you if you got some help?
Perhaps I could get along with my mother.
Tell me more about your family.
My mother takes care of me.
Who else in your family takes care of you?
You are like my father in some ways.
What resemblance do you see?
Natural language generation is the generation of printed text as well as synthesized speech. The transcript above is from a system that generates "psychoanalytic" responses to user input. This system is called ELIZA, and was developed at MIT in the 1960s. The program illustrates some of the issues of natural language that will be discussed. Other uses of natural language are for customer services and e-mail that are provided by telephone. Natural language generation can also be used in some expert systems. Another application is in translation programs. It is probable that many new applications will be developed as natural language programs are refined. The two main issues in natural language generation are determining what to say, and determining how to say it.
The issue of determining "what to say" quickly spills over into other areas of Artificial Intelligence, however, a few issues are especially interesting. The basis of these issues is that natural language is so "natural" to users that they expect, or at least desire the computer to "think" like a human. This causes users to use imprecise language when communicating with natural language systems. For example, the following questions were typed to PLANES, which is a question and answer system delivering information regarding aircraft flights and maintenance:
1. I request the number of flight hours…
2. I need to know the number of flight hours…
3. Want number of flight hours…
4. Find the number of…
5. Was any work performed on Plane 3…