基于语义拓扑网的反馈学习技术  

A Feedback-learning Mechanism Based on the Semantic Topological Network

作　　者：杨建武[1] 万小军[1] 

机构地区：[1]北京大学计算机研究所,文字信息处理技术国家重点实验室,北京100871

出　　处：《计算机工程》2005年第1期6-8,共3页Computer Engineering

基　　金：国家技术创新计划资助项目

摘　　要：为了从海量数据中快捷有效地获取所需的信息,提出了语义拓扑网的概念以及基于语义拓扑网的反馈学习方法。通过将数据对象的内容特征与语义特征进行有机地结合并构成语义拓扑网。在反馈过程中利用语义拓扑网,不断学习记忆并指导搜索。实验表明,基于语义拓扑网的反馈系统具有良好的学习能力与记忆能力,能有效地提高检索系统的性能。In order to accurately and quickly gain useful information and knowledge from huge amount of digital objects, relevance feedback has been put on many efforts. In this paper, a new concept is proposed, which is named semantic topological network where digital objects' contents features and semantic features are combined via classification learning. And a feedback-learning mechanism based on semantic topological network is discussed. The experimental results show that the information retrieval system with the feedback-learning mechanism achieves high accuracy and effectiveness on real-word text collections.

关 键 词：相关反馈 语义拓扑网 检索系统 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]