NLG Performance Evaluation Task
(Redirected from NLG Evaluation Task)
Jump to navigation
Jump to search
A NLG Performance Evaluation Task is a NLP evaluation task that involves assessing the effectiveness and quality of outputs produced by NLG systems.
- Context:
- It can (typically) utilize an NLG Evaluation Metric, such as: Text Fluency, Text Coherence, Factual Accuracy, and Text Relevance.
- It can (often) involve human judges to provide qualitative analysis and feedback, complementing automated evaluation metrics such as BLEU, ROUGE, and METEOR.
- It can range from being an Intrinsic Evaluation Task assessing the text internally to an Extrinsic Evaluation Task that evaluates how well the text performs in achieving specific external goals.
- ...
- Example(s):
- Counter-Example(s):
- Text Generation Tasks, which focus solely on the creation of text rather than its evaluation.
- Data Analysis Tasks, where the main goal is to extract insights from structured data rather than generate or evaluate text.
- Machine Learning Model Training Task, which involves training algorithms without the specific focus on evaluating generated text.
- ...
- See: Natural Language Generation, Text Summarization Task, Machine Translation Evaluation Task, Automated Content Generation, Evaluation Metric