Evaluating the Faithfulness of Importance Measures in NLP by Recursively Masking Allegedly Important Tokens and Retraining.
Andreas MadsenNicholas MeadeVaibhav AdlakhaSiva ReddyPublished in: CoRR (2021)
Keyphrases
- natural language processing
- expressive power
- quantitative measures
- information extraction
- wordnet
- question answering
- neural network
- genetic algorithm
- natural language
- text mining
- pointwise
- field of natural language processing
- empirically derived
- computational linguistics
- text processing
- evaluation measures
- machine translation
- line segments
- case study
- information systems
- machine learning