Finding representative images for POIs

摘要

In recent years, consumer-focused location-based services have been growing exponentially. People can get the nearby information or plan a trip easily by using these applications. However, there is tough competition between these similar applications. Hence how to attract users become a big challenge.People usually choose the service which provides the latest and the most correct information. In order to achieve this goal, we need to constantly update the database. In this paper, our goal is to build a system which can renew POI data automatically. We want to update the information of POIs such as telephones, business hours and images. Among all of these data, images are consider as the most important type since they can attract users’ eyes.In this study, we have designed an approach to find the correct images for POIs in several steps. First, we use a search engine to collect a POI’s information and extract its features. The Second part is to collect the images from the web pages. With the popularity of the internet, there are lots of resources like social media or blogs. For each image, we collect the related information such as title or Alt and the near-by texts. Finally, we match the POI’s feature with image’s information to find the best image.