Font Size: a A A

Spot Recognition And Sentiment Analysis Of Travel Travels Based On Deep Learning

Posted on:2021-10-22Degree:MasterType:Thesis
Country:ChinaCandidate:C FanFull Text:PDF
GTID:2518306032967809Subject:Computer technology
Abstract/Summary:PDF Full Text Request
With the rapid development of information technology and the massive growth of Internet users,online travel platforms are increasingly prosperous compared to traditional travel companies.These platforms provide users with the opportunity to communicate and share travel experiences.It allows travellers to understand information such as the route of the tourist destination and the evaluation of attractions through travel notes and comments shared by users.At present,travel notes have become one of the important references for tourists to choose attractions and travel routes.Due to the large number of travel notes and the uneven quality,the redundancy and fragmentation of travel notes make users spend a lot of time and energy.This article focuses on the following three issues:First,entity and emotion word recognition based on travel information in the travel text.For this study,the following four sets of comparative experiments were set up:CRF model,LSTM model,Bi-LSTM model considering context information,and Bi-LSTM-CRF model combined with CRF.The experimental results show that the Bi-LSTM model uses context information in named entity recognition,which is better than the LSTM model.The Bi-LSTM-CRF model that uses the CRF layer instead of the Softmax layer has the best effect,and its accuracy,recall and F1 values can reach 92.08%,89.27%,and 90.65%,respectively.Second,based on the aspect-level sentiment analysis of tourist evaluation in travel notes.For this study,the main set of comparative experiments the following four groups:the most basic method Majority,LSTM models,Deep Memory Network model and the improved model of Deep Memory Network.The experimental results show that the Deep Memory Network model and its improved model have the highest accuracy rate when the hop layer is 6,and the effect is better than the LSTM model.Among them,the Deep Memory Network improved model has the best effect,and the accuracy rate can reach 77.58%.Finally,after calculating the transfer rate of each scenic spot and its favorable rate,using ArcGIS software to render the path of popular attractions in the text,combined with its favorable rate to finally get the attractions path and evaluation map.In this way,it is possible to save the time for the tourists to plan the route before traveling and bring convenience to the tourists.
Keywords/Search Tags:Named entity recognition, Aspect-level sentiment analysis, Bi-LSTM-CRF model, MemNet model, GIS visualization
PDF Full Text Request
Related items