基于属性agent模型的新闻信息网页筛选技术  

Technology for selection of news information pages based on attribution agent

作　　者：胡军[1] 杨先芬[1] 刘喜亮[2] 黄慎[1] 

机构地区：[1]湖南大学计算机科学与通信学院,长沙410082 [2]湖南大学软件学院,长沙410082

出　　处：《计算机应用研究》2009年第5期1760-1763,共4页Application Research of Computers

基　　金：国家自然科学基金资助项目(60773208);高等学校博士学科点专项科研基金资助项目(20070532075)

摘　　要：随着人工智能技术的不断发展和复杂动态系统建模手段的不断完善,agent技术因其自主性、反应性、预动性等特点和在协作、推理及规划方面的优势,使其对非结构性的决策与不确定性的推理有很强的刻画能力,能很好地解决一些非数学模型的动态推理与筛选特征事件为基础的问题,从而为人们解决类似的问题提供非常好的新途径。从构建的属性agent模型出发,利用网页筛选技术的风格特征:HTML标签、URL字符、文本内容和视觉效果等获取风格特征的属性值,从而构建新的风格决策树模式,提高网页识别、筛选的精度。实验表明,该技术能提高网页筛选的精度。With the development of artificial intelligence and complex system modeling, The agent technology gives us a new approach to solve the real complex problems for the agent has the attribution of autonomy, reactivity, pro-activeness and rationality. It can build the model for non-structure decision and uncertain reason, and also can solve the dynamics reason and selection the attribution from huge data that can＇ t easy to be described by the mathematic model. This paper built the attribution agent model to select the Web page about news information through the genres attribution： HTML tag, URL char, text content and visual style. And could get the new gene decision tree to improve the Web page about news information selection and identification correctness. The experimental result shows that the correctness of proposed technology get great improved.

关 键 词：属性agent 网页特征风格 决策树 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]