RESEARCH ON THE CONCERNS OF TOURISTS DURING THE EPIDEMIC BASED ON TOPIC MODEL

The covid-19 has severely hindered the development of tourism, so it is necessary to study the tourism under the pandemic situation. Travel websites provide a lot of data and topic model is a commonly used model to mining online review. However, the traditional topic model will lead to serious sparsity problem when dealing with short text due to lack of word frequency and context information and will be interfered by bridge words. Therefore, we proposed BW-BTM which is effective and superior to the traditional model to learn higher quality topics from short text with the large number of bridge words. Finally, we found out five aspects that tourists paid more attention to during the epidemic period through detecting topic, they are entry and exit, ticket refund service, epidemic prevention in scenic spots, hotel epidemic prevention requirements and public transportation epidemic prevention.