Automatic Natural Language Generation (NLG) Performance Measure
Jump to navigation
Jump to search
An Automatic Natural Language Generation (NLG) Performance Measure is a NLG performance measure that uses computational techniques to evaluate the quality of generated text.
- Context:
- It can (typically) objective, reproducible, and scalable.
- It can be critical for the iterative development and refinement of NLG models.
- It can provide efficiency in evaluation, it may sometimes miss nuanced aspects of language quality.
- It can be used to compare different NLG systems or tracking improvements in a single system over time.
- It can range from being a Automated Objective NLG Performance Measure to being a Automated Heuristic NLG Performance Measure.
- ...
- Example(s):
- Measuring the perplexity of sentences generated by a language model.
- Evaluating machine translation outputs using BLEU scores.
- ...
- Counter-Example(s):
- See: Natural Language Generation, Computational Linguistics, Automatic Evaluation, NLG System Evaluation.