Web searching on a mobile device, especially based on a cell phone is very common and convenient. Since most Web pages are with rich content and mainly are specifically developed for desktop computers, it is hard to navigate them with (relatively) small screen of handled devices. Hence, there is a great interest in computer science to adopt such pages with rich content into small screens of mobile devices. However, a Web page has different parts that do not have equal importance to the end user. This fact made us to consider a novel approach to identify the most useful portion of a Web page to the end user regarding its search query. Typically, mobile phones are common in the following characteristics:
Small Screen: limited screen area compared to a conventional PC.
Limited Keyboard: Limited area for interaction with the device. Therefore it is necessary to put the relevant part at first.
Distracted User: A desktop user is mainly in an (relatively) isolated environment with focused attention to its working. Whereas, a smartphone user may be in an open area with distractive surrounding which makes him or her to devote only a portion of his or her attention to the device. Consequently, less scrolling is desired.
Authors in (Sugaya, 2007) have proposed the architecture of a page segmentation based on content-distance. The approach is basically based on checking a page score and the distance of the content according to the distance in the source code of the corresponding Web page. Issues with this approach can be categorized as followings:
The distance between elements in the code and between corresponding elements in the actual page might be different, i.e., the distance between 2 cells in the source code vs. the distance of corresponding cells in the actual table in the actual Web page.
Most Web pages contain numerous parts which makes the introduced method slow and time-intensive.
The low accuracy of the proposed method (55-70% based on the authors report) makes it not a desirable technique.
The proposed solution (a hybrid version of the framework) still can be time-intensive for the fact it needs to check a considerable sets of conditions.
Efforts on displaying web pages on small screens fall into following categories: