面向网页信息无障碍合规性检测的样例推荐系统

发布时间：2018-02-12 21:23

本文关键词： 信息无障碍 DOM树网页相似度网站无障碍检测　出处：《浙江大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着信息技术的飞速发展,人们日常的生活已离不开互联网。互联网给人们带来便捷的同时,由于多数网站不符合无障碍标准,使得残疾人无法像健全人那样方便的获取互联网信息。为了使残疾人能够与健全人一样没有障碍地浏览网站,对网站进行无障碍建设意义重大。为了对网站的无障碍能力进行评估并找出其中不符合无障碍标准的网页,需要对其进行网页信息无障碍合规性检测。然而,当前的网站无障碍检测系统无法完全自动化,仍有部分检测规则需要专门的检测人员进行人工检测。由于待检测网页类型各异、检测过程依靠大量主观判断,因此人工检测困难重重,检测的准确率与效率无法得到保证。本文针对网页信息无障碍合规性检测提出样例推荐算法,在人工检测过程中为检测人员实时推荐与当前检测页面具有最高相似度的样例页面,以帮助检测人员更加快速的理解检测规则,提高检测速度和准确率。为了有效衡量页面间的相似度,本文利用DOM树来保存页面结构并采用改进的简单树匹配算法来计算页面结构相似度。同时采用URL比较用来快速识别明显的模板页面,以达到加快页面相似度计算速度的目的。此外还增加检测内容相似度计算,并对结构相似度和检测内容的相似度赋予相应的权重作为总的页面相似度,以达到提高推荐准确率的目的。基于已有的页面相似度计算算法,本文提出面向样例推荐的样例库创建方法,随着样例的不断增加提出改进后的更新算法以保证样推荐的高效和准确。最后,本文通过在真实数据上的实验,验证页面相似度计算算法的可靠性,并通过推荐系统功能实现与用户对推荐样例的反馈,验证了样例推荐系统能够在无障碍检测过程中为检测人员推荐有参考价值的样例页面。
[Abstract]:With the rapid development of information technology, people's daily life has been inseparable from the Internet. Make it impossible for people with disabilities to have access to Internet information as easily as the able-bodied. Accessibility of websites is significant. In order to assess the accessibility of websites and identify those pages that do not meet accessibility standards, they need to be tested for information accessibility. The current website barrier-free detection system can not be fully automated, and there are still some detection rules that require manual inspection by specialized examiners. Due to the different types of pages to be tested, the detection process relies on a large number of subjective judgments. Therefore, manual detection is difficult, and the accuracy and efficiency of detection can not be guaranteed. In the process of manual detection, we recommend the sample pages with the highest similarity with the current detection pages in real time, so as to help the examiner to understand the detection rules more quickly. Improve detection speed and accuracy. In order to effectively measure the similarity between pages, In this paper, the DOM tree is used to save the page structure and the improved simple tree matching algorithm is used to calculate the similarity of the page structure. At the same time, the URL comparison is used to quickly identify the obvious template pages. In order to speed up the computing speed of page similarity, the computation of detection content similarity is added, and the weight of structure similarity and detection content similarity is given as the total page similarity. In order to improve the accuracy of recommendation, based on the existing page similarity calculation algorithm, this paper proposes a method to create sample library for sample recommendation. With the continuous increase of samples, an improved updating algorithm is proposed to ensure the efficiency and accuracy of the sample recommendation. Finally, this paper verifies the reliability of the page similarity calculation algorithm through experiments on real data. Through the function of recommendation system and user feedback to the recommended sample, it is verified that the sample recommendation system can recommend the reference sample page for the examiner in the process of barrier-free detection.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP393.092;TP391.3

【相似文献】