Text Simplification with Reinforcement Learning Using Supervised Rewards on Grammaticality, Meaning Preservation, and Simplicity.
Akifumi NakamachiTomoyuki KajiwaraYuki ArasePublished in: AACL/IJCNLP (Student Research Workshop) (2020)
Keyphrases
- reinforcement learning
- learning algorithm
- supervised learning
- markov decision processes
- function approximation
- state space
- unsupervised learning
- machine learning
- natural language
- text retrieval
- information retrieval
- semantic representations
- model free
- temporal difference
- database
- free text
- optimal policy
- keywords
- reward function
- reinforcement learning algorithms
- semantic representation
- reward shaping
- text documents
- policy iteration
- multiresolution
- semantic content
- dynamic programming
- supervised classification
- semantic information
- text mining
- textual data
- key concepts
- partially observable
- learning process
- learning problems
- feature selection
- hidden state
- lexical semantics
- neural network