基于网络的多媒体节目编目信息抽取技术的研究-信号与信息处理专业论文.docxVIP

的抽取做好准备。因此，论文在研究了 HTML 网页的特点后，提出 了一套适用于多媒体节目编目信息抽取的 WEB 预处理技术，其中包 括 HTML-Tree 中心内容确定方法、基于 HTML-Tree 方法的网页特征 提取等关键技术，通过这些关键技术的实现，很好地达到了对 WEB 信息进行预处理的目的，提高了网页自动分类的查准率和查全率。 然后论文研究了多媒体节目编目信息抽取的主要实现技术，使用 了基于主题的信息抽取方法，该方法通过多媒体节目编目信息模板的 构建、主题相似度的判断和模式匹配最终得到较为完整的多媒体节目 编目信息，该方法可以较好地完成预期目标，获取一些简单的常用的 编目信息。 最后，论文在 Java 平台上实现了论文所提出的多媒体节目编目 信息自动抽取系统 NMPIES，并进行了大量的实验，取得了较好的结 果。 关键词： 多媒体节目编目信息，HTML 自动分类，信息抽取， 特征提取 STUDY ON INFORMATION EXTRACTION FOR MULTIMEDIA PROGRAM CATALOGUE BASED ON INTERNET ABSTRACT Recently with continuous expansion of digital multimedia services demands to catalogue information of digital multimedia programs from the audience is becoming much more. The demand is a great power to this dissertation. With rapid development of Internet a great increase of web data results in generating a great amount of semi-structured data which include much information about multimedia program thus make it possible to acquire multimedia program catalogue information. So the dissertation takes automatic extraction technique for multimedia program catalogue information as our task and goal with a view to Internet. For solving this problem the dissertation proposes a network automatic extraction system for multimedia program catalogue information (NMPIES) based on the studies on the general classes for IE (Information Extraction) and the general implementation of IE system. Web page pretreatment and HTML automatic classify is the precondition of IE for multimedia program catalogue, which is also the important point of this dissertation. It proposes a set of web page pretreatment technique for multimedia program catalogue information which includes HTML-Tree Center Content Decision Method and Feature Selection based on HTML-Tree. With these key methods the expectation goal can be completed very well. For information extraction for multimedia program catalogue, a method of theme-based information extraction is used. The catalogue information for multimedia program can be acquired