基于术语自动抽取的科技文献翻译辅助系统的设计.docxVIP

基于术语自动抽取的科技文献翻译辅助 系统的设计 黄政豪崔荣一 延边大学工学院计算机科学与技术学科智能信息处 理研究室 摘要： 设计了一种中韩科技信息综合平台中的翻译辅助系统.首先，依据关键词确定的 组词特证获取候选术语，并使用互信息评估候选术语以实现术语自动提取.其次, 将已有术语、抽取到的新术语、术语译文和历史翻译记录等信息存储到系统数据 库屮建立术语库?最后，设计翻译工作者的用户接口，使其通过该接口获取已有 术语的译文信息、新术语的相似译文信息和译文记忆库为基础的历史翻译数据. 测试结果表明，本文设计的术语自动抽取功能和辅助译文生成功能达到了预定 的设计目标，术语自动抽取算法召回率达到61.8%,结合优化方法进行优化后 达到66. 9%;辅助译文生成平均延时为0.03Is, MRR为0.951,测试结果满足用户 需求. 关键词： 术语自动识别;术语抽取;翻译辅助系统; 作者简介：崔荣一(1962—),男，博士，教授，研究方向为模式识别、智能计 算. 收稿日期：2017-04-19 基金：吉林省自然科学基金资助项目(20140101186JC) Design of translation assistant system based on automatic extraction of terms HUANG Zhenghao CUI Rongyi [nte11igent Information Processing Lab?， Dept? of Computer Science Technology, College of Engineering, Yanbian University; Abstract： This paper describes the design method of Chinese and Korean science and technology information aided translation system.Firstly, extracting can didatc tc rms based on word forma tion charac ter is tics of keywords, and using mutual information to evaluate Candidate terms for automatic term extraction. Secondly, existing terminology, extraction of the new terminology, terminology transla.tion and history of translation records and so on are stored in the system database and the establ ished terminology database. Finall)s dcsign the user intcrfacc for tremslators, so the translators can obtain the translation of exiting terminology, the similar translation of additional terms, and the history translation data based on translation memory through this interface. The results of the system test show that automatic extraction of term and the auxi1iary tremslation function reach the desired goals. The recall rate or term automatic extraction algorithmis61. 8%, and after optimization the rate is improved by optimizationmethodtoreach66. 9%. Thegeneration of auxiliary translation averagely delaysO. 031 seconds, andtheMRRisO.951, so the test results fulfil the users needs. Keyword： automatic term recognition; term extraction; computer aided translation; Received： 2017-04-1