Design And Implementation Of Translation Assistance System For Scientific And Technical Literature Based On Automatic Term Extraction

Scientific and technical literature serves as significant source of information for researchers to understand and research the latest development of trends in science and new technology.Hence,the multilingual scientific and technical literature sharing platform makes researchers understand the progress of the scientific research among different countries,which is of much meaningful to promote the innovation of modern science and technology.The main difficulty in translation of scientific and technical literature lies in translating technical terms in various fields.Locating accurately and complying with the uniform rules in translating the technical terms is a critical step and content for the translation of scientific and technical documents.For this reason,this thesis designed the translation assistance system for Chinese-Korean scientific-technical information integrated platform.The system captures new terminology and signs it automatically when the new science and technology literature entering,more than that,it quickly generates the translation,similar translation and the related historical translation record of terms selected by translator through auxiliary production module,thus the non-professional translators could translate the scientific literature efficiently.First of all,we designed and implemented an automatic term extraction algorithm based on structural features and mutual information.The algorithm segments the domain terms based on the keywords in the scientific and technical literature and analyzes the combinational characteristics of terms:the number of words and parts of speech for the words.We used these characteristics to construct filter template and stop words template,and scan the new entering literature to obtain candidate terminologies.And the mutual information was utilized to judge the degree of stability of the candidate terminologies adjacent lexical entry and determine whether the candidate terms could be accepted by setting the threshold value.Secondly,the existing terminologies,the extracted new terminologies,the terminology translations and historical records of translation etc.were stored in the system database forming terminology database.The terminology database is available to the translator in acquiring all the relevant information of the selected term through translation assistance system and the term search function.Finally,we designed and implemented the translation assistance system.The system includes the user interface for translators,by which translator can obtain the translation of exiting terminology,the similar translation of new terms and the historical translation data based on translation memory.The system sorted the accessory translation information according to text similarity algorithm,so translators can quickly find the matching translation results,which way improve the translation efficiency.The results of the system test show that the automatic extraction of the term and the accessory translation function reaches the desired goals.The recall rate or term automatic extraction algorithm is 61.8%,and after optimization the rate was raised to 66.9%,which was increased by 5.1%.The generation of accessory translation averagely delays 0.031 seconds,and the MRR is 0.951.In conclusion,the test results turn out that the system fulfils the users’ needs.