PTB Graph Parsing with Tree Approximation

Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 5344–5349 Florence, Italy, July 28 - August 2, 2019. c⃝2019 Association for Computational Linguistics 5344 PTB Graph Parsing with Tree Approximation Yoshihide Kato and Shigeki Matsubara Information & Communications, Nagoya University Furo-cho, Chikusa-ku, Nagoya, 464-8601 Japan yoshihide@icts.nagoya-u.ac.jp Abstract The Penn Treebank (PTB) represents syntactic structures as graphs due to nonlocal dependen- cies. This paper proposes a method that ap- proximates PTB graph-structured representa- tions by trees. By our approximation method, we can reduce nonlocal dependency identifica- tion and constituency parsing into single tree- based parsing. An experimental result demon- strates that our approximation method with an off-the-shelf tree-based constituency parser significantly outperforms the previous meth- ods in nonlocal dependency identification. 1 Introduction In the Penn Treebank (PTB) (Marcus et al., 1993), syntactic structures are represented as graphs due to nonlocal dependencies, which capture syntac- tic discontinuities. This paper proposes a method that approximates PTB graph-structured represen- tations by trees. By our approximation method, we can reduce nonlocal dependency identification and constituency parsing into single tree-based parsing. The information loss of our approxima- tion method is slight, and we can easily recover original PTB graphs from the output of a parser trained using the approximated ones. An experi- mental result demonstrates that our approximation method with an off-the-shelf tree-based parser sig- nificantly outperforms the previous nonlocal de- pendency identification methods. 2 Nonlocal Dependency Identification This section explains nonlocal dependencies in the PTB, and summarizes previous work on nonlocal dependency identification. 2.1 Nonlocal dependency in PTB In the PTB, a nonlocal dependency is represented as an edge. One node is called an empty element, which is a covert element in the syntactic repre- sentation. The other is called a filler. PTB’s syn- tactic representations are graph-structured, while its constituency structures are represented by trees. Below, a syntactic representation in the PTB is called a PTB graph. The left graph in Figure 1 is an example of PTB graph. The empty elements are labelled with -NONE-. The terminal symbols such as 0 and ∗T∗ designate their types of the empty ele- ments. 0 and ∗T∗ represent a zero relative pronoun and a trace of wh-movement, respectively. If a ter- minal symbol of an empty element is indexed with a number, its corresponding filler exists in the PTB graph, and is indexed with the same number. For example, the empty element of type ∗T∗ is indexed with 1 and it has the corresponding filler WHNP-1. For more details about PTB nonlocal dependen- cies, we refer readers to (Bies et al., 1995). 2.2 Previous Work Most PTB-based parsers deal with the trees ob- tained by removing nonlocal dependencies and empty elements (we call such trees PTB trees). While such parsers are simple, efficient and ac- curate, they cannot handle nonlocal dependencies. To fill this gap, several methods have been pro- posed so far. They can be classified into the fol- lowing two categories: the methods that introduce special operations handling nonlocal dependen- …

Content truncated to 500 words