摘要
一位新闻记者-机器人与机器学习的工作人员新闻编辑每日新闻-机器翻译的研究结果在一份新的报告中讨论。根据NewsRx记者在南京的新闻报道,研究表明:“机器翻译系统已经广泛地应用于我们的日常生活中,使生活变得更轻松、更方便。不幸的是,错误的翻译可能导致严重的后果,如经济损失。”我们的新闻记者从南京大学的研究中得到一句话：“这就需要提高机器翻译系统的准确性和可靠性。”然而，由于底层神经模型的复杂性和难解性，机器翻译系统的测试具有挑战性。针对这些挑战，我们提出了一种新的基于语法树剪枝(STP)的变形测试方法来验证机器翻译系统。我们的主要观点是，修剪后的句子应该与原句具有相似的关键语义。具体地说，STP(1)在句法树表示层次上提出了一种基于基本句子结构和依存关系的核心语义保持prun策略；(2)基于变形关系生成源句对；(3)通过词袋模型报告翻译具有一致性的可疑问题。我们在两个最先进的机器翻译系统（即Google Translate和Bi NG Microsoft Translator）上进一步评估了STP，其中1200个源句子作为输入。结果表明，STP在Google Transla TE和Bing Microsoft Translator中分别准确找到5073个和5100个唯一错误翻译（比最先进的技术高出400%），准确率分别为64.5%和65.4%。
Abstract
By a News Reporter-Staff News Editor at Robotics & Machine Learning Daily News Daily News-Research findings on machine translati on are discussed in a new report. According to news reporting from Nanjing, Peop le's Republic of China, by NewsRx journalists, research stated, "Machine transla tion systems have been widely adopted in our daily life, making life easier and more convenient. Unfortunately, erroneous translations may result in severe cons equences, such as financial losses." Our news reporters obtained a quote from the research from Nanjing University: " This requires to improve the accuracy and the reliability of machine translation systems. However, it is challenging to test machine translation systems because of the complexity and intractability of the underlying neural models. To tackle these challenges, we propose a novel metamorphic testing approach by syntactic tree pruning (STP) to validate machine translation systems. Our key insight is t hat a pruned sentence should have similar crucial semantics compared with the or iginal sentence. Specifically, STP (1) proposes a core semantics-preserving prun ing strategy by basic sentence structures and dependency relations on the level of syntactic tree representation, (2) generates source sentence pairs based on t he metamorphic relation, and (3) reports suspicious issues whose translations br eak the consistency property by a bag-of-words model. We further evaluate STP on two state-of-the-art machine translation systems (i.e., Google Translate and Bi ng Microsoft Translator) with 1,200 source sentences as inputs. The results show that STP accurately finds 5,073 unique erroneous translations in Google Transla te and 5,100 unique erroneous translations in Bing Microsoft Translator (400% more than state-of-the-art techniques), with 64.5% and 65.4% precision, respectively."
NETL
NSTL