不原创精品良网页判定算法的研究与实现.docxVIP

不良网页判定算法的研究与实现 赖勇浩（计算机科学与技术师范专业）指导老师：谢赞福副教授 摘要:不良网络信息过滤是一个切合社会迫切需求，应用面极广，体现大学生社会 责任感的极具商业价值的课题，是人工智能与数据挖掘相关技术的综合性应用，其最 垂要组成部分是对网页性质的判定?本文的不良网页判定算法对文本和图像这两个 最重耍的网页组成元素进行检测分析，由软件系统自行判定网页是否包含色情，暴力, 反动等不良信息.提出了基于学习型知识库的文本内容检测分析算法和反误判算法, 综合了基于肤色模型和人脸特征的图像检测算法，建立文木与图形结合的数学模型, 创造性地将文木与图像两面结合起来判定网页的性质，极大地提高了识别率,降低了 误识率.在已经实现的过滤系统一一《上帝Z手》上，对不良网页的识别过滤率高达 99.2% ? 关键字:不良信息过滤判定算法识别文本检测 Research and Implementation of Judging Bad Web Pages Algorithm Yonghao Lai Directed by: Professor Zanfu Xie Abstract： Bad network information filter is a commercially valuable task that suits to the exigent requirement of the society and has a huge application.lt adequately incarnates the social responsibility of the undergraduates.lt is a correlation technique all-around application of Arificial Intelligence and Data Mining.The most important ingredient is the judgement to the web page proterty.Judging bad web pages algorithm of this passage is a detecting analusis towards to text and image,which are the most important component element of web pages.This software system itself judges the web pages whether contains the web information such as eroticism,violence and counteraction.Suggest the algorithm of detecting text content based on studying repository and anti-misdetecting algorithm. Bi nd the image detectomg algorithm which based on complexion model and face feature,eatablish the mathematical model which bind text and image,creatively judge property of the web pages by detecting text and image,infinitely improve the recognition rate,reduce the anti-missing rate.In our filter system-GODfS HANDjhe reconition rate to bad web page have been up to 99.2%. Keywords: information filter Judging bad web pages algorithm5Recognition,Text detecting. 引言 中国互联网业形成声势也就5年时间，但形象和境况却已变幻多次，互联网会 在短短2?3年Z间迅速“变色”。以短信为契机，互联网已经成为色情/暴力/反动的 主战场?一切正在往相反的方向加速滑行，越来越“黄”，越来越“黑”，形成恶性循 环，失去了界限，使得互联网的主力——青少年成为主要的受害者。由于高匿名性、 高隐私性、高互动性和无地域性等特点，使得成人内容根本无法有效管理和制约。 对青少年的侵害，远比不良的网络游戏危害更人、更严重。 令人担忧的是，我们传统的监管法规、技术手段乃至对网络色情业的认知都还 停留在初级阶