A Neural Network Architecture for Syntax Analysis







Deposit Papers 


Chen, Chun-Hsien and Honavar, Vasant (1995) A Neural Network Architecture for Syntax Analysis. Technical Report TR95-18, Department of Computer Science, Iowa State University.

Full text available as:Postscript
Adobe PDF


A Neural Network Architecture for Syntax Analysis
Chun-Hsien Chen  & Vasant Honavar
Artificial Intelligence Research Group
Department of Computer Science
226 Atanasoff Hall
Iowa State University
Ames, IA 50011. U.S.A.,
ISU CS-TR 95-18, July 1995
Artificial neural networks (ANNs), due to their inherent parallelism
and potential fault tolerance, offer an attractive paradigm
for robust and efficient implementations of syntax analyzers.
This paper proposes a modular neural network architecture for
syntax analysis
on continuous input stream of characters. The components of the
proposed architecture include
neural network designs for a stack, a lexical analyzer,
a grammar parser and a parse tree construction module.
The proposed NN stack allows simulation of a stack of large depth,
needs no training, and hence is not application-specific.
The proposed NN lexical analyzer provides a relatively efficient and
high performance
alternative to current computer systems for lexical analysis
especially in natural language processing applications.
The proposed NN parser generates parse trees by parsing strings from
widely used subsets of deterministic context-free languages (generated by
LR grammars).
The estimated performance of the proposed  neural network architecture
(based on current CMOS VLSI technology) for syntax analysis is
compared with that of commonly used approaches to syntax analysis
in current computer systems.
The results of this performance comparison suggest that the proposed
neural network architecture offers
an attractive approach for syntax analysis in a wide range of practical
applications such as programming language compilation and natural language
Vasant Honavar's research was partially supported by the National
Science Foundation through the grant IRI-9409580.

Subjects:All uncategorized technical reports
ID code:00000105
Deposited by:Staff Account on 17 August 1995

Contact site administrator at: