Page 169 - Kỷ yếu hội thảo khoa học lần thứ 12 - Công nghệ thông tin và Ứng dụng trong các lĩnh vực (CITA 2023)

Linh Bui Khanh, Anh Nguyen Quynh, Hai Tran Van, Ha Nguyen Thi Thu 153

Fig. 4. WordCloud of the most frequency aspect that are mentioned by customer

Figure 5 shows the percentage of customers who are interested in each hotel aspect.
When analyzing in detail, it is possible to see the service aspects that are of attracting
customers. From that, we understand which aspects are satisfied by customers and
which aspects are dissatisfied with. Hotels can improve and upgrade service quality.

Fig. 5. The list of aspects with the ratio that are satisfied by customers

5.2 Prediction result

We use Python language to set up LSTM architecture. With Epochs = 15, this model
has the results with an accuracy of up to 96%. Figure 6 below shows the value of the

ISBN: 978-604-80-8083-9 CITA 2023