| Summary: | 碩士 === 中華大學 === 資訊工程學系 === 104 === As technology advances, people progress from using desktop computers to portable smart mobile devices, such as: smart phones, tablets, and so on. And with the increasing popularity of 4G network and smart mobile devices, how to let users to obtain more information through smart mobile devices quickly has become an important issue for technology developers nowadays. At the same time, computing power of smart mobile devices is much more advanced than it in the past, so that many software applying augmented reality techniques into smart mobile devices are produced and augmented reality technology has been successfully applied in many fields. It can be found in many literatures that applications combining smart mobile devices and eco navigation system platform are still relatively small and most of existing applications use text or image mode. Therefore, to provide a rich and immediate ecological tour service, in this thesis, we present an ecological tour system based on augmented reality technology, which uses augmented reality technology to improve navigation immediacy in the ecological environment and to enhance the richness of ecological navigation platform. At the same time to improve the system's efficiency, the recommended system is applied to calculate the attribution degree whose objective value is the basis of items recommendation.
This thesis discusses how to use augmented reality and location-base service action navigation technology with images and sounds of special animal and plans in the ecological environment. Use of augmented reality technology, direction position of birds or plants which may occur nearby can be provided for users depending on their location of the user. Based on users’ interests the recommendation system can realize ecological navigation and education through calculating attribution degree by attribution degree algorithm and recommending items based on the calculated objective value. The recommendation system proposed in this thesis takes "Time - Count" as the basis. The "Time–Count” algorithm is compared with PIP similar algorithm using the MAE between the predicted specific groups and actual groups and F1 index value of the proportion of different cold-starting problems occurred. It can be seen from the practical data that the "Time-Count" algorithm has a certain accuracy, and can solve the cold-starting problem.
|
|---|
