An Empirical Evaluation of Rule-Based and Statistical Models for Natural Language Parsing

International Journal of Computer Science and Technology Vol. 3, No. 1, January 2001 | pp. 1–8

DOI: 10.46882/2001/IJCST/000025

Research Article

Title: An Empirical Evaluation of Rule-Based and Statistical Models for Natural Language Parsing

Names of Authors: Arthur Pendelton¹, Yuki Sato²

Authors’ Affiliations: ¹Department of Computer Architecture, University of Texas at Austin, USA; ²Graduate School of Engineering, Osaka University, Osaka, Japan.

Abstract:
The growth of textual processing tools requires efficient semantic analyzers capable of parsing structured sentences with minimal ambiguity. Rule-based parsers struggle with colloquial phrases, while pure statistical models exhibit high memory profiles and require massive training corpora. This paper details a hybrid parsing model that merges a deterministic rule engine with a probabilistic contextual grammar matrix. The framework applies logical deduction patterns to resolve base sentence components before executing statistical likelihood metrics to evaluate ambiguous grammatical nodes. We tested the system across an text repository consisting of 200,000 multi-source sentences. The performance analytics show that our hybrid model boosts parsing accuracy by 8.4% and drops computational execution cycles by 23.5% relative to standalone statistical classifiers. The dictionary memory footprint dropped by 18.2%, validating that our structural approach can optimize language-processing tools running on restricted computing environments.

Keywords: Natural Language Processing, Computational Linguistics, Parsing Algorithms, Hybrid Models, Statistical Classification, Grammar Matrices.

Manuscript Timeline: Received: October 05, 2000; Revised: November 12, 2000; Accepted: December 02, 2000; Published: January 18, 2001.

Citation: Pendelton, A., & Sato, Y. (2001). An Empirical Evaluation of Rule-Based and Statistical Models for Natural Language Parsing. International Journal of Computer Science and Technology, 3(1), 1–8.

Subscribe to read the full article: https://internationalscholarsjournals.org/subscribe-to-read