← Back to Explore

The Impact of Part-of-Speech Tagging on the Accuracy of Word Sense Disambiguation Models

JSHS · 2022

Overview

Billions of people use search engines, like Google, daily. These systems identify meanings of words using machine learning classifiers (MLCs) through a task called word sense disambiguation (WSD). MLCs are algorithms that identify patterns in data to infer the definitions of words. A task relevant to WSD is part-of- speech tagging, which identifies the part-of-speech of a word using algorithms called part-of-speech taggers (POSTs). Past research on WSD and part-of-speech tagging indicates that using POSTs while training MLCs increases the MLCs’ accuracy. However, the specific changes in accuracy that different POSTs cause in MLCs is not up-to-date with the MLCs and POSTs that are currently widely used. I aimed to study the relationship between popular MLCs and POSTs to provide information that is current. I explored the effect of two different POSTs, the Natural Language Toolkit Tagger and the Stanford Tagger, on the accuracy of three different MLCs: the naive Bayes, random forest, and k-NN classifiers. I used the SemCor 3.0 dataset, and all of my objectives were met. My findings supported conclusions from Moreno-Monteagudo et al. (2006), which stated that POSTs increase the accuracy of MLCs by a marginal amount, and certain combinations of POSTs and MLCs, such as the Stanford Tagger and the k-NN classifier, are more accurate than others. My study also found that the Stanford Tagger increased the accuracy of MLCs the most, and the k-NN classifiers outperformed all other MLCs tested. There were also instances where the accuracy decreased, which may be due to overfitting.

Competition history

  • JSHS 2022 Category not listed

Resources

Related projects

Closest projects by meaning, across every fair and year in the corpus.

Source: Junior Science and Humanities Symposium

Save projects to your library

Sign in with Google to keep track of projects you find interesting, organized into folders. An account also raises your daily allowance for “Has this been done?”, and lets you create a key for the MCP server with a much higher limit than anonymous use. Browsing stays public.

Continue with Google