一种面向非结构化信息的知识获取框架  

Knowledge Acquisition Framework for Unstructured Information

作　　者：邹志鹏[1] 饶若楠[1] 

机构地区：[1]上海交通大学软件学院

出　　处：《微型电脑应用》2010年第8期18-21,5,共4页Microcomputer Applications

基　　金：上海市科委科技发展基金项目(项目编号:08DZ2292902和08dz1500507)的资助

摘　　要：在文本、图片、音频等非结构化信息中蕴含着大量的知识,如何从海量非结构化信息中,自动获取这些知识是一个亟待解决的问题。该文提出一种非结构化信息知识获取框架KAFUI,该框架以非结构化信息管理架构UIMA为基础,支持异构的非结构化信息资源的接入,可整合和复用多种分析引擎,分析结果中标注了非结构化信息中的实体及其关系,这些实体和关系,进一步转换为以OWL描述的知识,并可通过推理发现新的事实。该框架的原型已在实际场景中应用,其实验情况表明是可行而有效的。Unstructured information such as text,images,audio,contains a great deal of knowledge,how to acquire these knowledge from the massive unstructured information automatically is a serious problem.This paper presents a knowledge acquisition framework for unstructured information called KAFUI which is based on UIMA.KAFUI support heterogeneous accessing of unstructured information resources,integrating and re-using of analysis engines.The analysis results got from unstructured information contains lots of annotations of knowledge objects such as entities and relations between them.These knowledge objects will be transformed to the knowledge form presented by OWL.In addition,KAFUI supports the discovering new facts through reasoning.The prototype implementation of the framework has been applied in real scenarios,and experimental conditions shows it is feasible and effective.

关 键 词：UIMA 知识获取 非结构化信息分析 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]