Date of Award
Open Access Dissertation
School of Hotel, Restaurant and Tourism Management
The widespread practice of photo-taking and sharing behavior fosters research on photographs to probe into consumer behavior and business strategies. Consequently, the critical role of photographic research has been acknowledged in the fields of marketing as well as tourism and hospitality. To delve deeper into understanding photographs, this dissertation carried out three independent but interconnected studies to examine photographs from both macro and micro perspectives.
The first study systematically reviews photographic research across sociology, marketing, computer science, and tourism and hospitality. It explores key topics, methodologies, and theories, and compares the merits and drawbacks of various research methods employed in the literature. Moreover, this study outlines future research directions for photographic studies in the fields of tourism and hospitality.
The second study develops a conceptual and measurement framework that associates hotel-generated photos on social media with the hotel brand image. By using mixed methods including survey, word-embedding, k-means clustering, and card sorting, four levels of pictorial attributes are conceptualized to explicitly connect with hotel brands: manifest content level (i.e., architecture/building, scenery, lobby, room, activity, food/beverage, and amenity/facility), photography level (i.e., photo aesthetics), latent cognitive level (i.e., price, location/accessibility, cleanliness, comfort, ambience, room quality, decoration/design, service, and technology/innovation), and brand personality level (i.e., relaxation, hospitableness, liveliness, distinctiveness, sophistication, and wholesomeness). Moreover, a dataset with ground truth labels of latent cognitive attributes and brand personality attributes has been established, and a measurement framework based on various deep learning algorithms has been developed to identify the various attributes.
The third study delves into the impact of multi-level photo attributes on social media engagement. Through the analysis of 37,553 Instagram posts, multiple machinelearning models are employed to predict social media engagement. The Random Forest model shows superior predictive accuracy in forecasting the number of likes. Pictorial attributes such as “ambience”, “location”, and “service”, “aesthetics”, “wholesome”, “relaxing”, and “lively” are found to be important in determining the number of likes a post would receive. In predicting the number of comments, the XGBoost model displays the best performance on the test data, with “ambience”, “location”, “technology”, “aesthetics”, “relaxing”, “distinct”, “wholesome”, “scenery”, “amenity”, and “lobby” being identified as the most substantial pictorial features.
The dissertation contributes to visual marketing and hotel brand image management through photographs, as well as content effectiveness in the era of social media. It also provides practical insights into brand marketing on social media platforms for industry professionals. Additionally, limitations and future research are addressed.
Li, N.(2023). Capturing Visuals in Hospitality: A Multi-Dimensional Exploration of Photographs in Interdisciplinary Research. (Doctoral dissertation). Retrieved from https://scholarcommons.sc.edu/etd/7484
Available for download on Sunday, November 30, 2025