摘要
跨语言事件映射主要研究的是不同语言的事件之间的相似性。针对传统方法仅从文本内容来获取特征词导致准确率不高的问题,提出从文本标题、文本内容以及新词发现三方面综合分析,通过计算候选词的综合权重来得到最终的特征词。实验证明了与传统方法相比,该方法准确性大大提高。
Cross language event mapping researches on the event similarity between two different languages. Existing researches only extract feature terms from the content, which cause a low accuracy. Aiming at the problem, a new method was proposed, which considered text title, content and new words simultaneously, and the final feature words were got by the calculation of comprehensive weight of candidate words. Compared with the traditional method, the method greatly improves the accuracy. In the end, the experiment proves its validity.
出处
《计算机应用》
CSCD
北大核心
2016年第A02期247-250,共4页
journal of Computer Applications
基金
国家973计划项目(2014CB340400
2012CB316303)
国家自然科学基金重点项目(61232010)
国家自然科学基金面上项目(61173064)
国家科技支撑计划项目(2012BAH39B04)
关键词
事件相似度
跨语言对齐
特征向量提取
文本聚类
概念扩展
event similarity
cross language alignment
feature vector extraction
text clustering
conceptual expansion