Sentiment analysis in online customer reviews: the Feels Like Home case

• Autor(a) principal: Almeida, Duarte Rodrigues dos Santos Farinas de
• Data de Publicação: 2019
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10071/22965
• topic: Data mining -- Data mining; CRISP-DM; Sentiment analysis; Polarity; Rev; Mineração de dados; Análise de sentimentos; Polaridade; Comentários

Portugal has been, for many years, an attractive destination for tourists from all over the world. This continuous flow of people opens opportunities for companies to explore and for some new other companies to emerge. All the data generated from the interaction of these companies with tourists can be submitted to data mining techniques to extract useful information and, therefore, create knowledge.Thiscasestudy uses those data mining techniques to try to explain the polarity of sentiments found in the online reviews of the properties that Feels Like Home, a local accommodation rental platform, manages. Out of the Feels Like Home’s portfolio, information regarding negative and positive mentions for each house (monthly) was retrieved from ReviewPro’s API, allowing for the final data set to have 1131 entries containing important information to be targeted bydata mining.Through the usage of descriptive analysis and predictive models (CART decision trees), the relationship between the properties and reservations’ characteristics and the sentiment polarity found in the reviews is described, as well as the main factors that can help predict those sentiments are revealed. Additionally, the relationship between the monthly occupancy rates and the sentiments’ polarity is also described.This way, this study generates useful knowledge for Feels Like Home and possibly for the rest of the industry to use and adapt to their business needs