Research On Web Pages Content Adaptation Based On Eye Tracking Characteristics For Small Screen Devices

Nowadays, with the rapid development of Internet and mobile computing technology, mobile devices are widely used in our daily lives, and mobile learning has become a general trend. When using mobile terminals with small screens to visit massive web resources designed for large-screen devices, problems such as slow Internet speed, small memory, weak computing power, and short standby time, will be better solved in the near future. On the other hand, there is an irreconcilable contradiction between screen size and portability of mobile devices, and it is difficult to solve the contradiction with current technologies in the short term completely. Therefore, while using small screen mobile terminals to visit massive Internet resources; there exists an inevitable problem of web page adaptation. To solve the problem, current solutions include thumbnail-based navigation, automatic transformation based on markup language, intelligent layout of webpages, etc. Nevertheless, these methods only used web page itself as the research object. Moreover, the adaptive transformation will disrupt the semantic consistency and break the layout of original web pages, and further creates unnecessary obstacles for users to locate information and results in unsatisfactory user experience. From the users’ perspective, the study presented a novel method by introducing users’ eye-movement characteristics while browsing web pages on personal computers, and aimed at resolving the above problems of web pages content adaptation browsing through the small screen mobile terminals. This paper is supported by "Twelfth Five-Year" National Science and Technology Support Project (NO.2012BAD35B02) and National Key Project (NO.2010ZX01045-001-005-3).The main contribution of this paper includes the following aspects:(1) Constructed an eye-movement characteristics model after browsing webpages. Unlike previous researches which mainly considered web page itself in page transformation, from end users’ perspective, we introduced eye-movement characteristics into webpage adaptation research while using small screen mobile terminals. Firstly, we used eye tracker to collect data of users’ eye-movement characteristics when browsing web pages on ordinary PC, generated heat maps of users’ browsing behavior, and then classified and extended the HTML Document Object Model (HTML DOM). On this basis, we designed the attention model based on eye-tracking heat map. Finally, we presented a mapping mechanism based on location coordinates, and then proposed two eye-movement browsing patterns, free web browsing pattern and full web browsing pattern.(2) Developed a web page classification algorithm based on layout characteristics. According to layout characteristics of web pages, we divided web pages into two categories, directory-based web page and content-based web page, and used link density and text density of web pages to represent layout characteristics. The algorithm has been implemented and experimental result showed the web pages classification algorithm based on layout characteristics is valid.(3) Developed a web page segmentation algorithm based on heat map. In this paper, we proposed the concept of atom information unit with clear definition and attribute descriptions, and then provided rule-based automatic recognition algorithm of atom information unit. Utilizing heat map generated by full web browsing, comprehensive considering the similarity of heat maps and label attributes among adjacent atom information units, using similarity calculation and fusion method of adjacent atom information units, we implemented segmentation web pages. Experimental results showed that the algorithm performs higher accuracy than Microsoft’s VIPS algorithm, partially solved the problem of semantic inconsistency after web page segmentation.(4) Developed a web page adaptive transformation algorithm based on eye-movement characteristics. For directory-based web pages, we collected information of heat maps by using free web browsing, calculated the importance value index of web page blocks, and then sorted webpage blocks by importance value. For content-based web pages, we extracted main contents using information of heat maps, along with text density and link density of web pages, then adaptively transformed and represented index webpages and content webpages respectively. Experimental results showed that the accuracy of our importance calculation method of web page blocks based on eye-movement characteristics is much better than the importance calculation method based on location and content features proposed by Microsoft; our method is equivalent to Block-based method on main content extraction, but significantly superior to the DOM-Based method.From users’ perspective, we introduced eye-movement characteristics while browsing webpages on personal computers into small screen adaptive webpage content research, constructed eye-movement characteristics model, and then segmented webpages according to eye-movement data collected from full web browsing. Meanwhile, we used these data to calculate the importance of webpage blocks and extracted main contents of web pages, and finally adaptively transformed and represented directory-based web pages and content-based webpages. To a certain degree, our methods reduced semantic inconsistency of web pages on small screens and thus helped users to navigate web pages on mobile devices, Experiments showed that compared to previous methods, methods proposed in this paper provided better accuracy and satisfaction, and have great useful value.