Dependency Parsing with Backtracking using Deep Reinforcement Learning

Abstract Greedy algorithms for NLP such as transition-based parsing are prone to error propagation. One way overcome this problem is allow the algorithm backtrack and explore an alternative solution in cases where new evidence contradicts explored so far. In order implement a behavior, we use reinforcement learning let action gets better reward than continuing current solution. We test idea on both POS tagging dependency show that backtracking effective means fight against