UCL+Sheffield at SemEval-2016 Task 8: Imitation learning for AMR parsing with an alpha-bound

Proceedings of SemEval-2016, pages 1167–1172, San Diego, California, June 16-17, 2016. c⃝2016 Association for Computational Linguistics UCL+Sheffield at SemEval-2016 Task 8: Imitation learning for AMR parsing with an α-bound James Goodman∗ Andreas Vlachos# Jason Naradowsky∗ ∗ Computer Science Department, University College London james@janigo.co.uk, jason.narad@gmail.com # Department of Computer Science, University of Sheffield a.vlachos@sheffield.ac.uk Abstract We develop a novel transition-based parsing algorithm for the abstract meaning representa- tion parsing task using exact imitation learn- ing, in which the parser learns a statistical model by imitating the actions of an expert on the training data. We then use the imita- tion learning algorithm DAGGER to improve the performance, and apply an α-bound as a simple noise reduction technique. Our perfor- mance on the test set was 60% in F-score, and the performance gains on the development set due to DAGGER was up to 1.1 points of F- score. The α-bound improved performance by up to 1.8 points. 1 Introduction In abstract meaning representation parsing (Ba- narescu et al., 2013), the goal is to parse natural lan- guage in a domain-independent graph-based mean- ing representation (AMR). In the first AMR parsing work, Flanigan et al. (2014) split the task into two sub-tasks; concept identification and graph creation. The sub-tasks are learned independently, and exact inference is used to find highest-scoring maximum spanning connected acyclic graph that contains all the concepts identified in the first stage. Later work by Wang et al. (2015b) adopted a different strat- egy based on the similarity between the dependency parse of a sentence and the semantic AMR graph. They start from the dependency parse and learn a transition-based parser that converts it into an AMR graph. To learn the parser, Wang et al. (2015b) de- fine an algorithm that for each instance in the train- ing data infers the action sequence that convert the input dependency tree into the corresponding AMR graph and train a classifier to predict the actions to be taken during testing. This strategy is also referred to as exact imitation learning, while the algorithm that infers the action sequence in the training instances is commonly referred to as the expert policy. In our submission to SemEval Task 8 on AMR parsing, we follow the transition-based paradigm of Wang et al. (2015b) with modifications to the pars- ing algorithm, and also use the DAGGER imitation learning algorithm (Ross et al., 2011) to generalise better to unseen data. The central idea of DAGGER is that the distribution of states encountered by the expert policy during training may not be a good ap- proximation to those seen in testing by the trained policy. Previous work by Rao et al. (2015) used SEARN, a similar imitation learning algorithm, on the AMR problem, with an algorithm that constructs the AMR graph directly from the sentence tokens. Imitation learning has also been used successfully in other semantic parsing tasks (Vlachos and Clark, 2014; Berant and Liang, 2015). In imitation learning approaches such as DAG- GER the previous actions become features …

Content truncated to 500 words