CAJ | 학술논문

随着互联网的发展和Web2．0的应用，网络信息呈现出越来越明显的动态演化性，传统的静态文摘方法不能很好地反应文摘的动态变化和新颖性，难以满足人们对于摘要获取效率的要求，动态文摘技术成为新的研究热点。提出一种基于改进K‐means算法的动态文摘提取和更新方法。为了减小聚类结果对初值的依赖性，提高聚类的稳定性，采用聚类中心的搜索算法获得较优的初始聚类中心。它能解决使用传统聚类方法进行动态文摘提取中存在的重复计算问题，并能保持较高效率。在T AC2008上的实验证明，该方法生成的动态文摘效果较好、算法效率高。
수착호련망적발전화Web2．0적응용，망락신식정현출월래월명현적동태연화성，전통적정태문적방법불능흔호지반응문적적동태변화화신영성，난이만족인문대우적요획취효솔적요구，동태문적기술성위신적연구열점。제출일충기우개진K‐means산법적동태문적제취화경신방법。위료감소취류결과대초치적의뢰성，제고취류적은정성，채용취류중심적수색산법획득교우적초시취류중심。타능해결사용전통취류방법진행동태문적제취중존재적중복계산문제，병능보지교고효솔。재T AC2008상적실험증명，해방법생성적동태문적효과교호、산법효솔고。
With the development of the Internet and Web2 .0 applications ,network information showing the dynamic evo‐lution of more and more obvious ,traditional static methods can not be a very good response Digest Digest of dynamic change and novelty ,can not satisfy people's summary Get efficiency requirements ,dynamic summarization technology be‐comes a new hotspot .This paper presents a document update summarization Algorithm based on improved K‐means algo‐rithm .In order to reduce dependence on initial clustering results ,to improve the stability of clustering using cluster cen‐ters on better search algorithm to obtain the initial cluster centers .It can solve the problem using traditional clustering method to extract dynamic abstracts in double counting ,and can maintain a high efficiency of the algorithm .Experimental on the TAC 2008 shows the effectiveness of the method to generate better dynamic Summarization ,the high efficiency of the Algorithm .