| Currently, visually impaired users can not take full advantage of the Internet to obtain the required information, this paper studies for the visually impaired user’s information processing technology, dedicated to the research of personalized mobile Internet technology, enhance the possibility of the visually impaired’s reading news fast and accurately by recommending technology and automatic summarization. Nowadays, a variety of information navigations for the visually impaired brings them convenience, but it also increases the difficulty in reading experience, especially in the more complex information environment, there is no obvious related news, they have to use screen reader to listen to each page line by line in order to make the next step. Thus, to reduce the inconvenience in access to information of the visually impaired, this paper introduces an optimized navigation designed to improve the efficiency of the reading experience of the visually impaired. The experiment section is based on the feedback on Listening World from the visually impaired, which proves that this system can reduce the complexity of the access to news, help the visually impaired to achieve news more effciently.Traditional solutions are as follows:web portable screen reader. Spoken-Web is a broadcast software which can be used only under the environment of Internet connection. With the help of the server, users can update information such as: domestic news, foreign news, finance and technology news. Other similar software like BrowseAloud and ReadSpeaker are also able to broadcast some information, however, such software does not provide information to the visually impaired, but to the learning disabled or language barriers. This type of server-based screen reader is not rich in functions, and can only provide information from the server, so the limitation is relatively large. But it is convenient when using public computers.Such principle is based on the conversion of HTML code, by changing the page layout, making it a linear structure (general web contents are too scattered for the visually impaired, because they need to rely on the reader to read line by line), thus it will be easier for the screen reader to convert it to audio.With the help of screen reader and W3-browser, the visually impaired can access to the W3page, however, the web structure and metadata information is difficult to identify, because the visually impaired can hardly identify or locate the title, link and non-text elements (such as inline images), also, it is difficult for them to achieve navigation between the title and link. For example, when the visually impaired are browsing pages, they need to scan the entire page by screen reader, when they find useful links they will jump to the next page, but if they want to return to the previous page, they need to scan the entire page again by news reader, line by line. Such reading is a waste of time, in the numerous miscellaneous information chaos, user is likely to get lost in meaningless information.The system uses a proxy server to complete the work, users do not need to install the software. This is the advantage of proxy server, it also facilitates the visually impaired access to information. But there are plenty of limitation in this kind of system, the algorithms is relatively simple, the most serious thing is that it is likely to lead to distortion in process, it is also helpless when it comes to pictures.Because of the way of HTML linear transformation is inefficient, and the web page design increasingly tends to interact with interactive graphics instead of text, early linear transformation will inevitably result in the loss of information. As a result, some people put their attention on the summarization (text summary, gist summary):they extracted summary from the web content, help the visually impaired to have a rough understanding of the page, thereby improving reading efficiency.In2000, from IBM in Japan, Chieko Asakawa and Hironobu Takagi decided to develop an annotation-based transcoding system which works between the user and the Web server to convert the existing Web pages. The system consists of two components:structural annotation and commentary notes.Structural annotation is used for recognizing fragmentary visual groups, displays the basic role and importance of each group, such as a description of a paragraph if it is title, tag or text. The commentary annotation is used for explaining each group for users. Through the comments, tags, and description of the image content, making it easier for users with visual impairments to access information online. Annotation function of the system improves greatly the reading efficiency of the visually impaired, but the system depends heavily on labor, especially in the description of images, people need to manually upload them to the server after describing it, which requires a lot of manpower support, in today’s vast network information, it is almost impossible.After that, Harper S, Patel N designed summarization software for the BBC and other news webs, which forms a small plug-in implementation of the Firefox browser, when browsing news, the plug-in text crawl across the news, and returned the first sentence of each paragraph to generate a summary of the text. To ensure simplicity, the plug-in returned up to four sentences. The first sentence of the first paragraph and last paragraph must be returned, and the remaining two sentences are the first sentence from top75%and rest25%. In order to ensure the integrity and readability of the summary, if the correlation degree of two sentences is high, it will be processed as one sentence. Finally, the plug-in will give a performance score of this summary:HIGH, MEDIUM, LOW for reference.Method proposed in the thesis is rather simple, plug-in in Firefox is also easy for users to use, no installation is needed, interception forms of summarization is good, the results achieved are satisfactory, but the plug-in is only for text-based content, such as news web, if there are more pictures in content, or when it comes to navigation, the plug-in has little help. Information accessibility study of this period is mainly designed for screen readers, it has many limitations.In the construction of Listening World, the paper devoted to making news reading more convenient, fast and efficient for the visually impaired. Recommendation algorithm is based on SVM applications, coupled with the impact of user characteristics, making recommendations more accurate, in order to achieve fast and efficiency in news reading, In keyword-based vector space model recommendation, we propose a multi-user word parameter calculation of interest rates. Finally, the system generates multi-document summarization. Integrity and efficiency of the entire system reaches the original design purpose, that is to meet the needs of visually impaired people in reading. Research work is mainly developed from the following four aspects:First of all, the first chapter introduces the background and significance of the study, elaborating the work in two main lines. It will introduces the current situation of the world and domestic visually impaired, normal reading habits’ changing of current. This naturally occurring contrast between visually impaired and normal, and then produce dissatisfaction in the visually impaired. We look back at history, introduces previously systems and software developed for the visually impaired. From the command line to the graphical, step-by-step to elaborate the development of history. At the same time, it describes the background of the application system, giving significance of the research system, which is among the network resources, how to make visually impaired users to obtain information quickly and accurately. In reviewing, it also details for a variety of software oriented to visually impaired users nowadays, as well as the advantages and disadvantages of these software and systems, advantages of the system compared to other software. All the things above lead to the thinking and contributions of this paper. Then the paper describes detail the significance of the system for the visually impaired and present the research orientation of recent scholars.The beginning of the second chapter introduces the main features of the system. And observed the system from the perspective of different users, analyze the overall framework of the Listening World, And highlighted the important role for the visually impaired in the application areas. Then follow the direction of data flow, the paper gives detailed explanation of the framework of Listening World, describes working principle and working flow module by module. From news acquisition, storage to classification, recommendation, user feedback gathering, weighs adugsting, all the modules are elaborately illustrated. The format of the data and the transmission format between the modules are also described. Finally, the paper give as simple explanation for interface optimization and navigation optimization.The third chapter discussed improved algorithm in the system in detail. The first is the introduction of SVM, gives the usual tutorial reluctant to write out a detailed description of the derivation process. In particular, the idea of SVM is explained thoroughly and understandingly. Then I compares my SVM model with the traditional SVM model. In the introduction of the keyword-based recommendation system, the paper explains the vector space model. Because the vector space model is actually not designed for recommendation system, how to apply this theory to the recommendation system, the middle of the parameter is selected based on what grounds, are all explanation. Followed by a description of the experimental data. In the multi-document summary, the paper introduces themes identified with news of the split tree alignment, rotation of the split tree. Then, based on word frequency of automatic summarization.In the last two chapters, we described the online operation of Listening World, the approval of the app store, data traffic of IFLYTEK. Reactions with the facts of our software. The research work carried out conclusion and futurework. Summarizing the summarization the recommended techniques and the overall design of our study and the concrete realization in the Listening World system, and pointed out the shortcomings and deficiencies which the system has, and propose future research directions and system improvements based on it. |
PDF Full Text Request