From text to graph: a general transition-based AMR parsing using neural network
- PDF / 2,987,789 Bytes
- 17 Pages / 595.276 x 790.866 pts Page_size
- 92 Downloads / 188 Views
(0123456789().,-volV)(0123456789(). ,- volV)
ORIGINAL ARTICLE
From text to graph: a general transition-based AMR parsing using neural network Min Gu1 · Yanhui Gu1 Weiguang Qu1
· Weilan Luo2 · Guandong Xu3 · Zhenglu Yang4 · Junsheng Zhou1 ·
Received: 12 March 2020 / Accepted: 18 September 2020 © Springer-Verlag London Ltd., part of Springer Nature 2020
Abstract Semantic understanding is an essential research issue for many applications, such as social network analysis, collective intelligence and content computing, which tells the inner meaning of language form. Recently, Abstract Meaning Representation (AMR) is attracted by many researchers for its semantic representation ability on an entire sentence. However, due to the non-projectivity and reentrancy properties of AMR graphs, they lose some important semantic information in parsing from sentences. In this paper, we propose a general AMR parsing model which utilizes a two-stack-based transition algorithm for both Chinese and English datasets. It can incrementally parse sentences to AMR graphs in linear time. Experimental results demonstrate that it is superior in recovering reentrancy and handling arcs while is competitive with other transition-based neural network models on both English and Chinese datasets. Keywords Semantic analysis · AMR parsing · Two-stack-based transition algorithm · Neural network
1 Introduction Meaning Representation of natural language is an important issue for massive data in the real world. How to achieve complete semantic understanding of natural language sentences in real-world data has been attracted by researchers [1–6]. When applying natural language processing technologies to mine semantic information and understand the real meaning of the data, semantic representation is the carrier of semantic information.
& Yanhui Gu [email protected] & Zhenglu Yang [email protected] 1
School of Computer Science and Technology, Nanjing Normal University, Nanjing, China
2
School of Computer Science, Nanjing University of Posts and Telecommunications, Nanjing, China
3
Advanced Analytics Institute, University of Technology Sydney, Sydney, Australia
4
Institute of Big Data, College of Computer Science, Nankai University, Tianjin, China
In recent years, graph representation-based strategies show their excited performance in expressing information in complex condition, such as social network representation [3, 7–10], chemical molecule representation [11–15] and so forth. Because of the characteristics of real-world data and the properties of graphs which can handle much more complex information, we intend to harness the wisdom of graphs to represent the semantic information in sentences. Because of the ambiguity and polysemy of semantics, it is a great challenge to realize the semantic understanding of natural language sentences. Traditional sentence semantic understanding research usually designs a formal meaning representation form for a specific domain [16, 17]. However, the real world is a multi-domain hybrid environment where peo
Data Loading...
