【作者】 倪耀群； 许洪波； 谭婧霞； 唐慧丰； 程学旗； 孟丹；

【Author】 Yaoqun Ni~(1,2,3),Hongbo Xu~1,Jingxia Tan~3,Huifeng Tang~3,Xueqi Cheng~1,Dan Meng~1 Institute of Computing Technology,Chinese Academy of Sciences,Beijing,China,100190

【机构】 中国科学院计算技术研究所； 中科院计算所； 中国科学院研究生院； 解放军外国语学院；

【摘要】 为方便维吾尔文与汉文使用者的信息交流和双语教学,作者使用经过优化的双语词典构建了维、汉多语言信息检索系统,该系统分为索引生成、多语言检索和摘要三个组成部分。首先采集网络上包含这两种文字的网页,将网页正文统一转换为UTF-8编码存储,然后按照汉维文词典合并后的UTF-8词表分词,使用中科院计算计算研究所I3Search(UTF-8版)建立索引。多语言检索通过翻译查询语句实现,支持多语言混合检索(如(?)网站)和单一语言检索(如维吾尔文网站)两种方式。多语言混合检索结果比较精确但是对使用者的语言要求很高;而根据双语词典,允许用户输入单一语言检索相对容易,但是词典中多义词和同义词现象导致检索结果精确率不高。多语言的摘要通过在正文中以字符为单位滑动寻找查询关键词出现最密集的窗口,加标签并截断尾部残缺字符后显示,取得了较为满意的效果。更多还原

【Abstract】 In order to make the information exchanging and bilingual education convinent for Uyghur and Chinese users,the author used optimized bilingual dictionary to construct an Uyghur-Chinese Cross Language Information Retrieval System which consists of index building,corss language information retrieval and multilingual absctract parts.In the index building part,the author gathered webpages contained the two languages at first,then transformed the main contents of every vvebpage into UTF-8 format and segmented the contents using Chinese-Uyghur combined word list,finally built index using ICT software 13Search(UTF-8 edition).The second part which provided multilingual and monolingual retrieval was implemented by translating query keywords.Multilingual retrieval is more accurate but requests foreign language knowledge of the users.On the other hand,monolingual retrieval is simpler by using the bilingual dictionary,but there are many conditions such as polysemies and synonyms,both of which made accuracy not very high.Mutillingual absctract found the window with the maximum keywords density via sliding along in the main contents character by character, tagged them,and truncated.So it gives a better result.更多还原