| The education of the next generation is getting more and more concern from parents.Parents are willing to invest in their children's education and have their children take after-school courses outside,including English classes.However,it is less efficient for children to learn English words at home.This is because they could only learn English words by books,tapes or videos.Books like textbooks and dictionaries cannot provide interaction with users.Besides,tapes and videos couldn't guarantee acceptance of English words.A recommender system combine multimedia presentation and consideration about acceptance of English words could help.But currently the popular techniques of recommender systems in practical are collaborative filtering which has trouble in recommending new items without prior records and content-based recommendation which can not measure and match users' acceptance of English words.So,two recommendation methods mentioned above cannot help,and there is a need for a new recommendation technique.To address the problems above,this paper describes the design and implement detail of a crowd-sensing recommender system for English words,which includes three subsystems: a data collection system for mobile devices,a system of data analysis and recommendation and a system of content management.The whole recommender system adopts a crowd-sensing data collection method,which can collect and upload data through smart phones.The crowd-sensing method can avoid the lack of prior records by improving the coverage of English words and providing sufficient records for recommendation.Meanwhile,the whole recommender system recommends English words by introducing a context-aware and semantics-aware user-based recommendation algorithm.The algorithm first introduces user information other than records of English words to describe learning stage and then represents English words by a trained semantic model.Subsequently,it combines them to measure the acceptance of English words.Finally,the algorithm searches similar users and generates recommendation based on the acceptance. |
