Supervisor
Muhammad Iqbal
Programme
MSc in Data Analytics
Subject
Computer Science
Abstract
Sentiment analysis within customer queries stems from its critical role in shaping the perception of a company’s brand. Poor handling of customer queries may lead to adverse consequences. This paper explored and compared the performances of NLP models, including NLTK, spaCy, BERT and DistilBERT on a dataset comprising of customer queries and feedback. The study aimed to evaluate the accuracy and effectiveness of these diverse NLP models in analysing sentiment within customer communications.
The findings reveal distinct patterns among the models. BERT and DistilBERT exhibit greater similarity in their results, as do NLTK and spaCy. Notably, BERT and DistilBERT demonstrate a tendency to categorize queries as predominantly neutral, suggesting potential strengths in handling diverse customer sentiments. This analysis contributes valuable insights into the strengths and weaknesses of various NLP models.
Date of Award
1-2022
Full Publication Date
1-2022
Access Rights
open access
Document Type
Dissertation
Recommended Citation
De Camillis, Patrizia, "Analysing Natural Language Processing Techniques: A Comparative Study of NLTK, spaCy, BERT, and DistilBERT on Customer Query Datasets" (2022). MSc in Data Analytics. 1.
https://arc.cct.ie/msc_da/1