An Approach to Modelling English Sentence Structures Based on the Combination of Graph Theory and Probability

Abstract

This paper studies the uses of graph theory and probability theory to model English sentence structures. Through a representation of sentences as directed graphs (parsing into dependency trees) and analysis of syntactic patterns using probability models, we also illustrate, how mathematical parametric modelling can increase our understanding of linguistic data. We take a look at the dependency grammar, as well as at the Markov models and the probabilistic contextfree grammars (PCFG) to investigate how mathematical tools can forecast and explain sentence construction. These results indicate that graph-theoretical and probabilistic tools are very strong tools for NLP, syntax-parsing and machine-translation.

Keywords: Graph Theory, Probability Theory, English Syntax, Dependency Structure, Sentence Structure, Natural Language Processing, Probabilistic Context-Free Grammars, Markov Models, Computational Linguistics, Machine Learning in Linguistics, Syntactic Parsing, Neural Language Models, BERT, GPT-4, Statistical Language Modelling, Deep Learning for Syntax