Research And Implementation Of A Mobile News Reading System For Visually Impaired People

With increasing development of the Internet and popularization of mobile devices,such devices have acted as one of the major platforms of acquiring news for most people.When interacting with mobile devices,some disabled users are facing difficulties due to their visual impairment,while most of current news reader applications are redundant in functionality and user interface.These applications are lack in accessibility,making the information acquisition for them low in efficiency.Therefore,development and providing of a visually impaired targeting news reader mobile application is of import significance.We cooperate with China Braille Press in this content,with the purpose of implementing a news reader application specifically suitable for visually impaired users.To improve the efficiency of news acquiring process operated by visually impaired users,we propose and implement a user behavior based hybrid recommendation method.Applying such method,the problem encountered by visually impaired users,which it is not easily to find the news they are interested in while massive quantity of news is available,can be solved.First,the explicit interactions and implicit behaviors are collected while using the application.The browsed news is scored using collected information.A content-based recommendations method is then fed by scored news to produce the time-varying historical interest model of a user.Simultaneously,a potential interest model is generated by user-based collaborative filtering.Finally,the two models are combined to obtain a fusion interest model.Candidate news is matched with the fusion model,while recommendation result is returned.Test and analysis are performed to demonstrate that the new approach is more effective than the traditional algorithm.To solve the problem that visually impaired users are difficult to learn the whole event associated with single news,we proposed and implemented an automatic detect and organize method of hot topics.First,collected news are classified into a set of topics using Single-Pass incremental clustering method.Hierarchical clustering based on time similarity is then applied within each topic to generate a context model of the topic.Representative news is selected and presented to user sorted in chronological order,so that visually impaired users can easily learn the key progress of whole event over time.Based on the research above,we designed and implemented an accessibility mobile news reader system.Major components of system are described in this paper,including a server which provides news data and performs data processing,and a client on iOS mobile system with full accessibility compatible design.The system provides enhanced news reading experience specifically for visually impaired users.