Sentiment Detection of Comment Titles in Booking.com Using Probabilistic Latent Semantic Analysis

Abstract

In a competitive and dynamic environment, many hotels compete to provide the best quality for customers. Hotel quality affects the brand image of a hotel marked from whether or not customers are satisfied. Booking.com website is chosen as the ideal data source as being able to utilize User Generated Content (UGC). UGC, serves as a data source ensuring the authenticity of data for customer reviews in Ponorogo, Indonesia. This experiment uses English text classification, to determine customer satisfaction and dissatisfaction based on the text they write on the title of customer testimony. The classification method used is PLSA and python programming language is used for preprocessing data. The data sampling is performed by data crawling using WebHarvy. The test results show that PLSA slightly outperforms previous research methods, namely LSA.