Is this project an undergraduate, graduate, or faculty project?
Graduate
Project Type
group
Campus
Daytona Beach
Authors' Class Standing
Ross Dickinson, Graduate Student Diogo Cobos, Graduate Student Elif M Cankaya, Graduate Student Jungwon Zang, Graduate Student Michael Fornito, Graduate Student
Lead Presenter's Name
Elif Cankaya
Lead Presenter's College
DB College of Arts and Sciences
Faculty Mentor Name
Hong Liu
Abstract
Previous customers profoundly influence the purchasing or booking decision of potential customers, and, therefore, it is imperative for businesses to assess and evaluate the social media reviews on their services or products. The aim of this project was to introduce deep learning as a tool for analyzing and evaluating customer reviews in the hotel industry. More specifically, we aimed to design several recurrent neural networks (RNN) models to analyze hotel reviews and reactions gathered from a popular travel accommodation website, TripAdvisor. To analyze the reviews, four LSTM RNN models with different labels and a convolutional neural network (CNN) model were developed. All models were evaluated to determine the most suitable model for hyperparameter tuning and distinguish the best performing model multi-class text classification. Based on the performance metrics, LSTM RNN was reported to be promising for sentiment analysis and the LSTM RNN with three classes and five classes achieved the best performance outcomes compared to the other models.
Did this research project receive funding support (Spark, SURF, Research Abroad, Student Internal Grants, Collaborative, Climbing, or Ignite Grants) from the Office of Undergraduate Research?
No
Trip Advisor Hotel Reviews: Text Classification Model
Previous customers profoundly influence the purchasing or booking decision of potential customers, and, therefore, it is imperative for businesses to assess and evaluate the social media reviews on their services or products. The aim of this project was to introduce deep learning as a tool for analyzing and evaluating customer reviews in the hotel industry. More specifically, we aimed to design several recurrent neural networks (RNN) models to analyze hotel reviews and reactions gathered from a popular travel accommodation website, TripAdvisor. To analyze the reviews, four LSTM RNN models with different labels and a convolutional neural network (CNN) model were developed. All models were evaluated to determine the most suitable model for hyperparameter tuning and distinguish the best performing model multi-class text classification. Based on the performance metrics, LSTM RNN was reported to be promising for sentiment analysis and the LSTM RNN with three classes and five classes achieved the best performance outcomes compared to the other models.