CAJ | 학술논문

微博舆情事件会带来严重后果，而碎片化和无序化使得难以从海量的微博内容中全面地认知微博舆情事件的全貌。从微博舆情事件内容聚合出其主要内容具有重要实践价值和研究价值。基于关键词抽取技术KEA提出了一种微博舆情事件内容聚合方法。根据对微博内容篇幅、规范性和网络性的分析，增加中心度特征、词性特征和改进词位置特征，并以朴素贝叶斯分类器抽取关键词集作为聚合输出。最后通过现实微博平台上的内容聚合实验证明本研究所提方法相对基线方法在准确率和召回率上的优势。
미박여정사건회대래엄중후과，이쇄편화화무서화사득난이종해량적미박내용중전면지인지미박여정사건적전모。종미박여정사건내용취합출기주요내용구유중요실천개치화연구개치。기우관건사추취기술KEA제출료일충미박여정사건내용취합방법。근거대미박내용편폭、규범성화망락성적분석，증가중심도특정、사성특정화개진사위치특정，병이박소패협사분류기추취관건사집작위취합수출。최후통과현실미박평태상적내용취합실험증명본연구소제방법상대기선방법재준학솔화소회솔상적우세。
Microblogging public opinion events(MPOE) bring serious consequences, it's difficult to recognize the full view of MPOE from huge number of microblogging content for its fragmentation and disordering. Aggregating main point from MPOE is of important practical value and research value. Based on KEA, the keyword automatic extraction, the paper proposes a method to aggregate main point from MPOE. According to the length distribution, normative and network characteristic, the paper adds centrality feature, part-of-speech feature and modifies the position feature, then extracts the keywords as output of MPOE aggregation by Na?ve-Bayes classifier. Finally, advantage on precision and recall of method we proposed comparing with baseline method is proved by experiment of MPOE aggregation performed on real microblogging platform.