Example-Based Grammar Abstraction and Machine Translation via Spread of Activation

摘要：

This article presents an experimental machine translator capable of learning the grammatical and translational properties of languages based on bilingually aligned example sentences. Newly-learned linguistic information is inte-grated into the knowledge network, which forms grammatical abstraction hierarchies for each language learned. Translation is achieved via spreading activation over the knowledge net-work. The systems performance was assessed in terms of general learning capacity, one-to-many and flexible word order translation tasks. Its learning algorithm and potential are discussed, and finally the direction of future research is suggested.

作者: Toshiyuki Uchino

作者单位: School of Informatics and Computing Indiana University Bloomington IN, USA

会议类型: 国际会议

会议名称: 2010 China-Ireland International Conference on Information and Communications Technologies(2010年中国-爱尔兰信息与通信技术国际会议 CIICT 2010)

会议地点: 武汉

会议语种:英文

页码: 123-128

在线出版日期: 2010-10-10（万方平台首次上网日期，不代表论文的发表时间）