Research on Methods of Service Quality Attributes Classification Based on Tourism User-Generated Content


Student thesis: Doctoral Thesis

View graph of relations


Related Research Unit(s)


Awarding Institution
  • Junhua HU (External person) (External Supervisor)
  • Kwai Sang CHIN (Supervisor)
Award date8 Jun 2023


The convenience, interactivity and personalization of the internet in the era of big data have promoted innovation and change of how customers and products interact, especially with regard to experiential products. The emergence and development of tourism e-commerce and social media platforms have produced a large amount of easily accessible tourism user-generated content (UGC), with which knowledge can be shared between tourists and tourism enterprises. UGC contains various consumer preferences concerning service quality attributes, and can serve as a reference for potential tourists making purchase decisions, travel plans and shopping arrangements. It also provides new ideas for how enterprises listen to the voice of customers in a timely manner and maintain customer satisfaction. Effective classification of service quality attributes contributes to a better understanding and analysis of customer satisfaction, and provides support for service quality management and competitive decisions. Existing research on analysis methods for classifying service quality attributes is mainly based on questionnaires that obtain information related to customer preferences for the study of the classification problems of service quality attributes. However, the quality of data obtained from such questionnaires and their limitations in time, space and cost, restrict the ability of enterprises to monitor and manage data to some extent.

As a reliable data source, UGC is of great significance to the study of classifying service quality attributes. The development of classification methods for service quality attributes based on UGC has become the key to ensure that tourism enterprises make effective quality management decisions and meet the needs of customers in time. In addition, characteristics of reviewers and review information determine the degree of adoption of review contents, and affect the value of UGC for the decisions of customers and enterprises. Therefore, it is necessary to systematically analyze existing methods, combine the characteristics of tourism UGC, consider the characteristics of reviewers and review information, and comprehensively use linguistic decision-making methods and service quality theories, as well as the techniques of sentiment analysis and deep learning, to conduct in-depth research on the classification methods of service quality attributes suitable for problems in various situations based on the improved existing methods. Specific research contents of this paper are outlined as follows:

(1) A sentiment analysis method based on a domain-dependent sentiment dictionary is proposed. In order to fully identify and explore sentiment preferences of customers from massive UGC, the defects of existing sentiment analysis methods are firstly analyzed. Then, tourism UGC from the hospitality industry is used to construct a domain-dependent sentiment dictionary. Finally, a sentiment analysis algorithm based on semantic sentiment words is developed.

(2) A service quality attribute classification method considering reviewer characteristics is proposed. This thesis mainly focuses on improving the importance performance analysis model, and proposes a classification method of service quality attributes based on the importance performance analysis model, to deal with the challenges of such classification in terms of conflict preferences that reflect reviewer characteristics in UGC. Firstly, this thesis utilizes linguistic distribution assessments, and quantifies multi-angle and specific descriptions in UGC to the greatest extent in a concise way. Secondly, by considering conflict preferences caused by differences in experience and expertise of reviewers, a performance determination method considering conflict preferences is designed. In addition, combined with the semantic deviation between ratings and sentiment orientations reflected in textual reviews, an importance measurement method is proposed based on semantic consistency.

(3) A classification method of service quality attributes considering review information characteristics is proposed. This thesis considers temporal information contained in UGC that reflect the characteristics of review information, and proposes a classification method of service quality attributes based on the Kano model, aiming to study the problem of classifying service quality attributes under the impacts of temporal dynamics brought by the COVID-19 pandemic and market segment dynamics. This thesis firstly combines the dynamic topic model with the theories of service quality, and proposes an identification method of service quality attributes, which successfully reveals service quality attributes and their dynamic evolution. Then, this thesis constructs a dynamic effect quantification method of sentiment preferences towards service quality attributes on customer satisfaction. To do so, an attention mechanism and a deep learning method are incorporated. This method breaks the limitations of the static assumptions in existing methods. Lastly, this thesis proposes a Kano classification method of service quality attributes considering dynamic effects, which generates classification and improvement decisions with respect to different dynamic situations.

    Research areas

  • tourism UGC, linguistic decision-making methods, service quality, classification methods, reviewer characteristics, review characteristics