1
Enhancing Shift-Reduce Constituent Parsing with Action N-Gram
Model
HAO ZHOU, Nanjing University
SHUJIAN HUANG, Nanjing University
JUNSHENG ZHOU, Nanjing Normal University
YUE ZHANG, Singapore University of Technology and Design
HUADONG CHEN, Nanjing University
XINYU DAI, Nanjing University
CHUAN CHENG, Nanjing University
JIAJUN CHEN, Nanjing University
Current data-driven shift-reduce parsers ’understand’ the context of parser actions by embodying large num-
bers of binary indicator features with a discriminative model. In this paper, we propose the action n-gram
model, which utilizes the action sequence for parsing disambiguation. The action n-gram model is trained on
action sequences with the n-gram estimation method, which gives a smoothed maximum likelihood estima-
tion of the action probability for a specific action history. We show that incorporating action n-gram models
into a state-of-the-art parsing framework could achieve parsing accuracy improvements on three data sets
across two languages.
Categories and Subject Descriptors: I.2.7 [Artificial Intelligence]: Natural Language Processing—Syntax
Parsing
General Terms: Languages, Experiments
Additional Key Words and Phrases: Shift-Reduce Constituent Parsing, Action History, Action N-gram Model
ACM Reference Format:
Hao Zhou, Shujian Huang, Junsheng Zhou, Yue Zhang, Huadong Chen, Xinyu Dai, Chuan Cheng and Jiajun
Chen, 2014. Enhancing Shift-Reduce Constituent Parsing with Action N-Gram Model. ACM Trans. Asian
Lang. Inform. Process. 9, 4, Article 1 (September YYYY), 17 pages.
DOI:http://dx.doi.org/10.1145/0000000.0000000
1. INTRODUCTION
Modern data-driven transition-based parsers parse a sentence by performing a se-
quence of shift-reduce actions. Most of these transition-based parsers run in linear
time, which is faster than traditional chart-based parsers [Eisner 1996; Collins 1997;
Charniak 2000; McDonald et al. 2005]. The linear parsers achieve state-of-the-art
This work was supported by National Natural Science Foundation of China (61300158, 61170181, 61472191)
and Natural Science Foundation of Jiangsu Province, China (BK20130580).
Author’s addresses: Hao Zhou, Shujian Huang
+
(corresponding author), Huadong Chen, Xinyu Dai,
Chuan Cheng and Jiajun Chen, State Key Laboratory for Novel Software Technology, Nanjing Univer-
sity, Nanjing, China; Junsheng Zhou, Department of Computer Science and Technology, Nanjing Nor-
mal University, Nanjing, China; Yue Zhang, Singapore University of Technology and Design, Singa-
pore; Email: zhouh@nlp.nju.edu.cn, huangsj@nlp.nju.edu.cn, zhoujs@njnu.edu.cn, yue zhang@sutd.edu.sg,
chenhd@nlp.nju.edu.cn, daixy@nlp.nju.edu.cn, chengc@nlp.nju.edu.cn, chenjj@nlp.nju.edu.cn.
Permission to make digital or hard copies of part or all of this work for personal or classroom use is granted
without fee provided that copies are not made or distributed for profit or commercial advantage and that
copies show this notice on the first page or initial screen of a display along with the full citation. Copyrights
for components of this work owned by others than ACM must be honored. Abstracting with credit is per-
mitted. To copy otherwise, to republish, to post on servers, to redistribute to lists, or to use any component
of this work in other works requires prior specific permission and/or a fee. Permissions may be requested
from Publications Dept., ACM, Inc., 2 Penn Plaza, Suite 701, New York, NY 10121-0701 USA, fax +1 (212)
869-0481, or permissions@acm.org.
c
YYYY ACM 1530-0226/YYYY/09-ART1 $15.00
DOI:http://dx.doi.org/10.1145/0000000.0000000
ACM Transactions on Asian Language Information Processing, Vol. 9, No. 4, Article 1, Publication date: September YYYY.