A novel approach to mine knowledge from social images

摘要

With the popularity of various social media website,currently,lots of social images attached with different kinds of metadata have been uploaded to social media websites.Mining useful knowledge from social images has been an emerging important research topic in web search and data mining.In this paper,we propose a novel approach to find geographical difference of a given concept from social image community.We put a given concept to social image community,and then downloaded social images with metadata,particularly,the place where the photo was taken should be provided in advance.Firstly,concept is submitted to social image community,and then social images with different kinds of metadata are downloaded.Secondly,social images are clustered according to metadata of images.Finally,the information of concept's geographical difference is found.Experiments conducted on social image community proof the effectiveness of our approach.