Word Sense Disambiguation using Neural Networks

Natural Language Processing

Word Sense Disambiguation (WSD) is the task of removing ambiguity in different senses of words. It is a core research field in computational linguistics dealing with the automatic assignment of senses to words occurring in a given context [11]. Humans are inherently good at WSD and distinguish senses used in words through spoken language. Computers on the other hand have difficulties identifying correct senses of words. Various advancements have been made in the task of disambiguation using mainly four approaches: Knowledge-based, Supervised, Semi-Supervised, and Unsupervised. Better understanding of the human language will help computer’s performance in various applications such as search engine optimization, information retrieval, information extraction, software assistants, and voice command interpretation. The objective of this work is to present a supervised neural network machine learning model using various algorithms dedicated to the task of maximizing accuracy of sense detection. The input layer of the neural network will consist of nodes having binary values depending on the presence or absence of frequently occurring context words related to the ambiguous words. The output layer will consist of nodes equal to the number of senses the ambiguous word has. Training and testing of the model will be done using lexical resources such as SemCor or OMSTI. Accuracy will be calculated based on All- Word tasks from SemEval International Workshops

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.