Analyzing tourism reviews using an LDA topic-based sentiment analysis approach.

Area of Science:

• Natural Language Processing
• Computational Linguistics
• Tourism Management

Background:

• Growing volume of online user reviews on specialized platforms.
• Need for automated systems to organize and classify reviews by aspect and sentiment.
• Importance of understanding customer opinions for business and destination management.

Purpose of the Study:

• To automate the organization and classification of user reviews for domain-specific aspects and sentiment polarities.
• To extract actionable insights from TripAdvisor reviews about Marrakech city.
• To enable tourism practitioners to leverage aspect-based sentiment analysis for destination improvement.

Main Methods:

• Data collection and pre-processing of user reviews.
• Latent topic extraction using Latent Dirichlet Allocation (LDA) algorithm.
• Sentiment analysis applied to each identified topic, with human validation of topic labels.

Main Results:

• Successful extraction of latent topics from Marrakech TripAdvisor reviews.
• Application of sentiment analysis to categorize topics by polarity (positive/negative).
• Identification of specific aspects contributing to destination strengths and weaknesses.

Conclusions:

• Automated topic modeling and sentiment analysis are effective for analyzing large volumes of user-generated content.
• The methodology provides valuable insights for tourism practitioners to enhance destination management.
• Aspect-based sentiment analysis offers a granular understanding of customer feedback for strategic decision-making.