The application and comparison of web services for sentiment analysis in tourism

Abstract

The popularity of social computing and sentiment analysis has attracted an increasing attention of tourism industry and academia. The sentiment analysis of residents' and tourists' plays an important role to the development of tourism. It aims to identify and analyze opinions and emotions contained in reviews which are expressed by residents or tourists. Although it's a challenging task, many companies and research institutes are developing web services to provide public-access and cost-effective solutions to the problem. However, to the best of our knowledge, there are very few studies to apply and compare these web services in tourism. Therefore, in this work we make an initial attempt to explore a set of three web services and compare their sentiment analysis capabilities using the well-known TripAdvisor data collection. Experiment results lead to some interesting conclusions.

Sentiment Analysis in Tourism: Capitalizing on Big Data

“…On the contrary, the availability of online user-generated content (UGC) and new technologies provided researchers with a new approach that travelers’ perceptions and possibly their level of satisfaction can be approached through “sentiment analysis.” Sentiment analysis, in general, aims to determine the overall contextual polarity of a text document, a review, an opinion, or an emotion expressed in online UGC, whereby polarity can be positive, neutral, or negative. While highly relevant for tourism, sentiment analysis in tourism is only beginning to gain in popularity (Feldman 2013; Gao, Hao, and Fu 2015; Ribeiro et al 2016).…”

“…Sentiment analysis, in general, aims to determine the overall contextual polarity of a text document, a review, an opinion, or an emotion expressed in online UGC, whereby polarity can be positive, neutral, or negative. While highly relevant for tourism, sentiment analysis in tourism is only beginning to gain in popularity (Feldman 2013;Gao, Hao, and Fu 2015;Ribeiro et al 2016).…”

Sentiment analysis of popular-music references to automobiles, 1950s to 2010s

“…To perform the Sentiment Analysis, we employ two open-source algorithms: IBM "Alchemy Language" (which at the time of writing has been integrated into the Watson line of products) 3 and IBM Watson "Tone Analyzer. 4 Both algorithms use Machine Learning approaches to identify sentiments, and have been widely applied to text sources including customer reviews (Gao et al 2015;Shah et al 2020) and social media posts (Cao et al 2018;Jussila and Madhala 2019).…”

“…A possible reason for the humans-algorithms differences is that the two algorithms are not trained specifically by music lyrics. The algorithm designers do not disclose the types of datasets used to train the two algorithms, however it is known that applications of the two algorithms have included social media postings and hotel reviews (Cao et al 2018;Gao et al 2015;IBM 2019). Of the two studies of which we are aware that employ the Watson Tone Analyzer algorithm on music lyrics (Al Marouf et al 2019;Napier and Shamir 2018), both used only the algorithm's determinations, without the inclusion of a comparison against human analysts' judgments.…”

Predictive Accuracy of Sentiment Analytics for Tourism: A Metalearning Perspective on Chinese Travel News

“…We also compared our model with two popular sentiment analysis tools: Semantria and ROST Content Mining. Semantria is a popular commercial online sentiment analysis software that performs well with online hotel reviews (Gao, Hao, and Fu 2015; Serrano-Guerrero et al 2015). ROST Content Mining is a sentiment analysis tool based on a general sentiment lexicon, which has achieved good performance with Chinese online reviews (Luo and Zhai 2017).…”