公开(公告)号：US20030236789A1

公开(公告)日：2003-12-25

申请号：US10180711

申请日：2002-06-25

申请人： International Business Machines Corporation

发明人： Dashi Agrawal ,  David P. Olshefski ,  Dinesh C. Verma

IPC分类号： G06F007/60 ,  G06F017/10 ,  G06F017/00 ,  G06F007/00

CPC分类号： G06F17/30333 ,  Y10S707/955 ,  Y10S707/957 ,  Y10S707/958 ,  Y10S707/99942

摘要： The present invention provides methods, apparatus and systems for cost conversant classification of objects. In order to achieve this, we create an efficient tree structure to store different classification rules. The methods particularly relate to objects with multiple attributes and classification rules that are defined in terms of these attributes. The techniques can handle rules that can be represented as a single or multiple hypercubes in attribute space. Furthermore, the present techniques are effective when the number of relevant object attributes are large and no specialized classification hardware is accessible. Such methods can be used in various applications such as classification of web sessions in an e-commerce application, classification of connection request in TCP connections etc.

摘要翻译： 本发明提供了用于成本分类对象的方法，装置和系统。 为了实现这一点，我们创建一个有效的树结构来存储不同的分类规则。 该方法特别涉及具有多个属性的对象和根据这些属性定义的分类规则。 这些技术可以处理可以在属性空间中表示为单个或多个超立方体的规则。 此外，当相关对象属性的数量大并且不可访问专用分类硬件时，本技术是有效的。 这样的方法可以用于各种应用中，例如在电子商务应用中对网络会话的分类，TCP连接中的连接请求的分类等。