摘要
为了解决自动文摘系统目前所面临的领域通用性和文摘质量的矛盾 ,提出了建造基于多Agent技术的自动文摘系统的方案 .建造这样的系统时 ,有两个关键问题 :首先是确定在一定负载下 ,面向各个领域的合适的文摘Agent数目 ,其次就是选择什么样的协调算法 .本文给出了在Internet环境下基于多Agent技术的自动文摘系统模型 ,提出了三种协调算法 ,在仿真的基础上分析了系统的性能 ,得到了在一定负载下面向各个领域的合适的文摘Agent数目 ,并对三种协调算法进行了比较研究 .
To resolve the contradiction of domain currency and abstract quality of automatic abstracting systems, we put forward the solution to build automatic abstracting systems based on multiagent technologies. To build such systems, there are two key problems: firstly, how many abstracting agents for each domain are suitable in some load; secondly, what coordination strategy should be used. The model of an automatic abstracting system based on multiagent technologies is given, and three kinds of coordination algorithms are set forth. We analyze the performance of the system based on the simulations, and get suitable number of abstracting agents for each domain in given load. Furthermore, we compare three given coordination algorithms.
出处
《电子学报》
EI
CAS
CSCD
北大核心
2001年第2期247-249,共3页
Acta Electronica Sinica
基金
国家自然科学基金! (No .69982 0 0 1 )
