Dependency Parsing with Backtracking using Deep Reinforcement Learning
This addresses error propagation in NLP tasks like parsing, though it is incremental as it builds on existing transition-based methods.
The paper tackles error propagation in greedy NLP algorithms by introducing backtracking via reinforcement learning, showing it effectively reduces errors in POS tagging and dependency parsing.
Greedy algorithms for NLP such as transition based parsing are prone to error propagation. One way to overcome this problem is to allow the algorithm to backtrack and explore an alternative solution in cases where new evidence contradicts the solution explored so far. In order to implement such a behavior, we use reinforcement learning and let the algorithm backtrack in cases where such an action gets a better reward than continuing to explore the current solution. We test this idea on both POS tagging and dependency parsing and show that backtracking is an effective means to fight against error propagation.