Modelling Grammaticality-grading in Natural Language Systems Using a Vector Space Approach
Moses Kehinde Aregbesola *
Department of Mathematics and Computer Science, Elizade University, Ilara-Mokin, Nigeria.
Rafiu Adesina Ganiyu
Department of Computer Science and Engineering, Ladoke Akintola University of Technology, Ogbomoso, Nigeria.
Stephen Olatunde Olabiyisi
Department of Computer Science and Engineering, Ladoke Akintola University of Technology, Ogbomoso, Nigeria.
Elijah Olusayo Omidiora
Department of Computer Science and Engineering, Ladoke Akintola University of Technology, Ogbomoso, Nigeria.
Oluwaseun Olubisi Alo
Department of Computer Science and Engineering, Ladoke Akintola University of Technology, Ogbomoso, Nigeria.
*Author to whom correspondence should be addressed.
Abstract
There exist several natural language processing systems that focus on checking the grammaticality (grammatical correctness or incorrectness) of natural language texts. Studies however showed that most existing systems do not assign specific scores to the grammaticality of the analysed text. Such scores would for instance prove very useful to second language learners and tutors, for judging the progress made in the learning process and assigning performance scores respectively. The current study was embarked upon to address this problem. A grammaticality grading model which comprised of 6 equations was developed using a vector space approach. The model was implemented in a natural language processing system. Correlation analysis showed that the grading (in %) performed using the developed model correlated at a coefficient of determination (R2) value of 0.9985 with the percentage of grammatical sentences in evaluated texts. The developed model is therefore deemed suitable for grammaticality grading in natural language texts. The developed model would readily find use in computer aided language learning and automated essay scoring.
Keywords: Computational linguistics, automated evaluation, grammar, grammaticality, grammaticality grading, natural language processing, Vector Space Model, Mosesean Vector Space.