Scenic Spots Location Recognition And Information Search Based On Tourism Big Data

With the development of Internet technologies and social economics,tourism has become an indispensable part of people's lives.People's demand for tourism information services has become more diversified.More and more people not only choose to search for information on the Internet by texts,pictures and other information of attractions according to their needs to prepare for travel,but also tend to share their own travel itineraries and journey experience.Therefore,the research on location recognition and information search of scenic spots based on tourism big data has become a hot spot concerned by a lot of researchers,which is an important method to help people to improve their quality of life.The main work done in this thesis is shown as follows:(1)Combining the characteristics of scenic spot comment texts,the dynamic theme model is used to extract the theme distribution of the scenic spots in the time dimension,effectively obtained the theme distribution and evolution,and established an effective semantic theme model of scenic spots.The topic distribution of the scenic spots in the time dimension is obtained through training.The time information is introduced into the tourist information search,and the user's search intention is estimated through the change of the scenic spots in the time dimension to effectively improve the accuracy of the scenic spot information search.Based on the GPS information of the pictures taken by the tourists,the clustering algorithm is used to get the clustering centers of attraction images and obtained the corresponding popular scenic spots.The location information of the scenic spots is effectively extracted and the distribution of the popular scenic spots is obtained.(2)A method of location recognition of the spot image based on depth learning is proposed.The convolution neural network structure is used to extract the deep features of the images.And obtain the intersection of sets of images visually and geographically close to the query image as the candidate set.The candidate images are sorted according to the similarity between the query image features and the candidate image features.Then the location of the closest image on both visual semantics and location to the query image is past to the query image.Experiments were carried out on the Flickr images of Beijing's scenic spots and the European cities landmark dataset to verify the accuracy of the scenic spots recognition method based on deep learning.Compared with the methods based on the image basic features and methods without considering the location information,the method of based on deep learning obtains higher accuracy with an average increase of about 15%.(3)A tourism information search method based on the dynamic theme of attractions and search intention is proposed.Combining with the characteristics of the comment texts of the spots,the dynamic topic model is used to extract the effective semantic topic models of the spots and the KL distance is used to calculate the similarity between the query model and the document model to obtain effective query results.Combined with the comment data of scenic spots and the dynamic theme model,the users'search intentions are estimated through the changes of the attraction theme in the time dimension,and the accuracy of the scenic spot information search is improved.Combined with the search result diversification method,the search results have more representative contents.Furthermore,through the search intent classification algorithm,solved the problem that the traditional search method is hard to understand the user search intents,so as to improve the accuracy of the scenic spot information search.Experiments on the related data of the scenic spots in Beijing show that the search method based on the theme of the scenic spot and the search intention is effective.Compared with the search method based on the keywords,the search performance is improved.(4)Designed and developed a scenic spot location identification and information search system based on tourism big data.The system is divided into cross-media travel data analysis module,location-based image recognition and search module based on deep learning,and travel information search module based on scenic spot theme and tourist search intention.System is developed based on Java,JSP is used for front page design and display,the Struts 2 framework is used to complete front and back control,and Java is used to complete backend data processing and algorithm package.This thesis combines data mining algorithms and image and text search algorithms to provide tourists with rich and accurate travel information and to help tourists get travel information in line with their search intents to facilitate their travel planning.