Template Based Natural Language Generation

From Cohen Courses
Jump to navigationJump to search

This is a method for the problem of natural language generation.

In general, template-based systems are natural language generating (NLG) systems that map their non-linguistic input directly to linguistic structure. This structure may contain "gaps" which are filled in during output.

For example a simple template-based NLG system may take in some semantic representation like:

  • Departure(Delta 2047, Gate B37, 15:00)

Which will in turn be associated directly with a given template, whose "gaps" are filled in with data from the semantic representation:

  • [flight] will depart from [gate] at [time].

The following are some modern template-based systems:

  • TG/2
  • D2S
  • EXEMPLARS
  • YAG
  • XTRAGEN

More sophisticated models of template-based NLG have been proposed to explore the following:

  • Automatic methods of inducing templates
    • To address maintainability issues of template-based methods
  • Making templates more domain independent.
    • To address the perception that template based methods are always domain specific
  • Output quality & variability


References

  • Real vs. template-based natural language generation: a false opposition? Deemter et al. CL 2003.

Relevant Papers