摘要
结合商品命名实体自身结构的特点，提出一种基于模板的产品名命名实体的翻译方法。该方法在传统模板中包含的固定项和可变项的基础上，添加候选项。首次把TTL模板抽取算法应用于“汉语-维吾尔语”，并加以改进。实验结果表明模板中候选部分的添加可大大减少模板的数量，提高模板的泛化能力和翻译质量。
Abstract
Combined with the structure characteristics of the commodity named entities, puts forward a kind of commodity named entity translation method based on template. The method in traditional template contains fixed and variable, on the basis of adding the candidate items. For the first time, puts the TTL template extraction algorithm applied in "Chinese - Uighur", and improves it. The experimental results ex-press that adding a candidate in the template can greatly reduce the number of templates and improve the generalization ability and the quality of translation templates.
NETL
NSTL
维普
万方数据