澳大新語 • 2021 UMAGAZINE 24 27 封面專題 • COVER STORY can detect the number and types of cars from an image, and recurrent neural networks which can process temporal data. Also under development is a machine‑learning‑powered application for predicting bus‑waiting times. Prof Gong says that they are still tackling certain challenges to make better use of social data for smart tourism. These include the integration of data from different social media platforms, which provide varying degrees of data access and store their data differently. Moreover, he points to challenges in semantic analysis, such as enhancing the ability of computers to handle words and sentences that can have multiple meanings, a phenomenon known as polysemy. ‘Our team will continue to improve our machine learning models, so that we can offer better and smarter travel experiences to tourists in Macao.’ and museums to better understand what the visitors want and need, so that they can improve their services to attract more visitors,' says Prof Gong. Furthermore, the researchers have used social media data to calculate the density of tourists in different areas over the course of the day. Prof Gong says such statistics can inform government planning for tourist facilities and the transport system. Better Algorithms for Tourism To help residents and tourists get around the city while avoiding traffic jams, Prof Gong’s research team plans to launch another mobile application, which sources data from cameras at over 40 junctions in the city. The tool can show real‑time congestion levels on a map and make predictions accordingly. This mobile application is based on two machine learning methods: convolutional neural networks which
RkJQdWJsaXNoZXIy MTQ1NDU2Ng==