숙박공유서비스의 온라인 리뷰를 활용한 전반적 만족도 결정 요인 분석

Abstract

With the Fourth Industrial Revolution and the rapid development of new technologies, the environment is changing rapidly, and the online market is growing. Consumers use the Internet and mobile platforms to find products or services they need. The culture of exchanging surplus resources with neighbors in the past has been expanded to exchange resources and services with strangers due to the development of information technology. It created economic benefits by utilizing resources by sharing unused or less-used products or services with others. As these new forms of trading increased, the sharing economy began to gain attention. Airbnb is the most representative business of P2P sharing economy in the area of accommodation between individuals. The Airbnb is a lodging platform company that provides a peer-to-peer (P2P) service that connects hosts and tenants directly so that hosts can share or rent their own empty rooms or houses. The service method features a form of community platform that allows people to share and book accommodation around the world on online, mobile and tablet. According to the Airbnb, the economic effects of Korean Airbnb market have resulted in 1.37 trillion won in 2018. In 2017, 1.888 million guests visited, of which 1.23 million Koreans were found to be using the Airbnb, 65% of the total. As the use of Koreans increases, it is time to study Koreans. Also, the subject of the study is timely at a time when COVID-19 is now turning to domestic travel rather than overseas travel. Recent prior studies targeting the Airbnb locals can be seen, and there are limitations in research by following existing survey research methods or by conducting research on specific topics. The overall lack of exploratory approaches to Koreans makes this study absolutely necessary. Online reviews left by customers using Airbnb themselves have various meanings. Research is needed to analyze online reviews and ratings by showing their needs frankly. Existing prior studies related to the Airbnb have also been conducted from various perspectives. However, it is time to actively analyze real-time data in this rapidly changing era to discover the inherent meaning and study the perception of Koreans about the Airbnb. Therefore, this study analyzed the meaning of the reviews recorded online by the Airbnb customers to derive meaningful results and also analyzed the relationship between overall satisfaction level according to customer ratings. Text mining techniques for reviews were used to identify these relationships. In order to achieve the research purpose, it has been explored the Airbnb selective attributes, online reviews and overall customer satisfaction through domestic and international literature review, and the literature related to web crawling, text mining techniques, statistical techniques, etc. First of all, the Airbnb online review was selected and analyzed to collect data that would be the most important basic data of this study. Online reviews of host accommodations in Seoul were collected on Airbnb's official site. Various text mining techniques were used to collect reviews, ratings, and prices left by customers using selected hosts. Using online review data, frequency analysis of the mentioned words was conducted and the meaning of the words (keywords) was understood through word cloud visualization of the derived top words. The online review document was then converted into a single document to derive the topic between the document and the word to confirm the meaning of the entire document. Finally, after the researcher identified the entire online review, a user-based emotional dictionary was established, and the percentage of positives and negatives in the entire document was identified by the researcher. Through the meaning of the exploratory results identified here, the Airbnb selective attribute factors determined by the researchers were selected and the relationship with overall satisfaction was confirmed. To collect the data needed for this study, the data were collected using deep learning techniques, and R and SPSS25.0 were used for text mining and statistical analysis. The analysis data examined regional distribution characteristics with 281 local data, and conducted text mining analysis and statistical analysis with 1,545 review data. The results of the study on the research tasks presented in this study are summarized as follows: First, after checking the characteristics of the regional distribution, it was confirmed that many accommodations were distributed around the area where tourist attractions were located. Second, the exploratory study of online reviews first identified keywords associated with the top keyword in the reference frequency analysis, and the difference in elicitation keywords between 'general host' and 'super host' through the top frequency keyword was identified. The topic modeling analysis then extracted the six topics (topic) of the overall review of the Airbnb to find the underlying meaning and, as a result of the sentiment analysis, identified the probability of positivity and negation of the entire review document. For the sentiment analysis, the researcher established a user-based sentiment dictionary, which resulted in 62% of the accuracy of the sentiment analysis, which proved to be significant. As a result of eliciting the meaning of online review data through various text mining analysis methods, we can see that online reviews contain a lot of customer opinions. The review analysis confirmed that the limitations that ratings alone cannot explain all the implications should be assessed and re-elected. Third, 'clean', 'communication' and 'smooth check-in' in the relationship of the host class (general host vs. super host) between the select property of the Airbnb accommodation and the overall satisfaction. The following results of host-level control effectiveness verification show that the general host affects overall satisfaction depending on the presence or absence of 'cleanness', 'communication', and 'round check-in'. Through this, the host needs to maintain cleanliness, prompt feedback with the guest, and smooth check-in. Through text mining techniques and statistical analysis, this study has shown that the Airbnb online review is very important information that can grasp the customer's sensibility and opinions, which is deeply related to overall satisfaction. Therefore, online reviews of various opinions and emotions have been shown to be important information for the Airbnb, and the scientific methods provided through this study will be the basis for further extended research.