【作者】 陈颖；

【导师】 徐海银；

【作者基本信息】 华中科技大学 ， 计算机技术， 2012， 硕士

【摘要】 因特网技术的发展使得Web成为当前最大的信息来源。但是Web在给人们提供丰富信息的同时，又使人们在有效找到自己需要的信息方面面临巨大的挑战。大多数信息检索系统将页面视为最小的不可再分的单元，然而一个页面中往往包含多个并不紧密相关的主题，还含有导航、装饰、联系方式等零碎信息，因此用整个页面来表示一个单一的语义有时候并不合适。另一方面，随着通信技术的发展，手持移动设备(PDA、手机、智能手机等)正在快速发展。移动设备和因特网的结合是最具潜力的发展方向，如何在手持移动设备的小屏幕上显示合适的网页，也是很有价值的一个议题。Web页面分块技术，为上述问题提供了一种解决方案。论文首先对现有的Web页面分块技术进行了详细分析，总结了各类技术在实际应用中的优势和不足之处。在前人工作的基础上，提出并实现了一种基于多线索的Web页面分块方法。该方法考虑到了实际网页设计常用的布局模式，结合了页面视觉信息和页面对应标签树中的结构信息，最后生成的分块既符合人分析页面的逻辑，也保留并体现了网页的整体结构，还为后续子块信息的提取、生成提供了方便。实验结果表明新的方法能对常规Web页面进行有效的分块，并弥补了现有的一些页面分块技术在实际应用中的不足，结合了结构信息和视觉信息线索，在算法的效果和效率间找到了平衡。进一步，在面向移动设备的Web页面转换应用中，新方法中产生的符合常规页面布局的分块提供了额外的重要信息，有较好的应用价值。更多还原

【Abstract】 With the development of the Internet, Web has become the largest information source.Yet, Web makes people face a huge challenge in looking for useful information efficiently,while offering a wealth of information. Most information retrieval systems on the Webconsider web pages as the smallest and undividable units. However, a page usuallycontains multiple topics that aren’t much relevant to each other, and various contents suchas navigation, decoration, contact information, etc. So a web page as a whole may not beappropriate to represent a single semantic. On the other hand, with the development ofcommunication technology, handhold mobile devices, such as PDAs, smartphones,improve rapidly. How to display web pages on small screens of handhold mobile devicesproperly is another valuable topic. Web page segmentation offers a solution to thoseproblems.In this paper, through the detailed analysis of popular methods of Web pagesegmentation, advantages and shortcomings of every kind of methods applied in the realsituation are concluded firstly. Based on previous researches, a new method of Web pagesegmentation is proposed and implemented making use of various cues. It takes severalaspects into consideration, including general layout patterns of web pages, visualinformation of web pages, and structural information of pages’ tag trees. Suchsegmentation matches how and what people segment web pages into, preserves the overallstructure of web pages, and makes it easier for extraction and production of subblocks’information later on.The experimental results show that the new method can segment general web pagesefficiently, and it overcomes the insufficiency of some existing approaches. We furtherapply it to offer extra important information with pieces that follow the general layoutpatterns in the application of web page transformation especially for handhold mobiledevices, which gives the new approach of web page segmentation good application value.更多还原