Detecting novelty seeking from online travel reviews: A deep learning approach

Travel online reviews is important experience related information for understanding an inherent personality trait, novelty seeking (NS), which influences tourism motivation and the choice of destinations. Manual classification these challenging due to their high volume unstructured nature. This paper aims develop a framework deep learning model overcome limitations. A multi-dimensional was created NS trait that includes four dimensions synthesized from prior literature: relaxation seeking, arousal boredom alleviation. Based on 30 000 TripAdvisor we propose using Bidirectional Encoder Representations Transformers (BERT)- Gated Recurrent Unit (BiGRU) recognize automatically reviews. The classifier based BERT-BiGRU scales achieved precision F1 scores 93.4% 93.3% respectively, showing can be relatively accurately recognized. study also demonstrates produce satisfactory results model. findings indicate BERT- BiGRU achieves best effect compared same kind models. Moreover, it proves traits identified travel computational techniques. For practical purposes, this provides comprehensive NS, used in marketing recommendation systems operating industry.