Hybrid Model to Word Sense Disambiguation for Hadiyyisa Language Using Supervised Machine Learning Models

Abstract

The process of determining the most appropriate interpretation of an ambiguous word means Word Sense Disambiguation. WSD has been numerous key applications compared to other NLP applications such as Text Summarization, Text Categorization, and Question Answering. The basic task of this study has been to design the quality and technical word sense disambiguation for Hadiyyisa text.  The data collection and data set preparation from different licensed areas was been induced. From the collected data, pre-processing techniques like TF-IDF were been used to convert text into the vector form. In this study Hybrid model and Supervised ML Models have been introduced. These Supervised ML models such as Naive Bayes, Neural Network, Support Vector Machine, and Hybrid Model were been used. In addition to this, to validate this model, an appropriate method adopted like Monte Carlo Cross-Validation (Shuffle Split) was been verified. The result was been found in two models such as SVM, and Hybrid Model. The accuracy of the SVM and Hybrid model for the selected ambiguous words such as Anga, Diinate, Hagara, Hurbaata, Misha and Seera respectively for both models were been investigated. The average accuracy of the model for SVM and Hybrid Model were been verified with their result of 79% and 82% respectively.