China
Africa
Explore chapters and articles related to this topic
Results and Conclusions
Published in Krzysztof Wołk, Machine Learning in Translation Corpora Processing, 2019
The quality of machine translation has greatly improved over time. Several general-purpose translation services, such as Google Translate [227], are currently available on the web. However, they are intended to translate text from a wide variety of domains, and are therefore neither perfect nor well-tuned to produce high quality translations in specific domains.
Speech and Language Interfaces, Applications, and Technologies
Published in Julie A. Jacko, The Human–Computer Interaction Handbook, 2012
Clare-Marie Karat, Jennifer Lai, Osamuyimen Stewart, Nicole Yankelovich
Like speech applications, machine translation systems are imperfect (and probably will always be) as the interaction is complicated by two potential sources of errors: recognition and translation. The same three categories of recognition errors (rejection, substitution, and insertion) that were earlier described for speech applications can also be found in the speech component of speech-to-speech machine translation applications. However, unlike speech applications, speech-to-speech machine translation applications do not currently have any systematic way of handling these errors in the interface. As we stated in Section 16.3.7, this is a consequence of how machine translation works. Translation depends on the accuracy of the input (in this case speech), but there is no provision in the interface to handle any of the recognition errors whatsoever, rather, the output from the recognition component is passed on directly to the translation component. For future speech-to-speech applications, this is one area in which the strategies already in place for speech applications can be applied or customized. The errors contained in the output from the translation component can also be modeled after the three types of errors in speech application: rejection, substitution, and insertion. Accordingly, a rejection error occurs when the machine translation algorithm has no valid hypothesis about what the user said. For example, a user says “May I run inside and give the keys to my visitor” and the translation outputs only function words like “in the to my” which is complete gibberish. A substitution error involves the machine translation algorithm changing the entire meaning of the original utterance by mistaking the user’s utterance for a different valid utterance. For example, a user’s initial message “How are you?” is translated as “How old are you?” which, though valid, sets the conversation on an entirely different path (and because two separate cultures are involved, this may even come across as too direct or offensive). In the case of an insertion error, the machine translation algorithm may decode multiple words when only one was spoken. For example, a user says “hello” but the system says “help me get all.” The same vacuum in the handling of recognition errors also exist in translation errors; there are currently no formal or systematic ways in the interface for dealing with these categories of translation errors. As a result, the users are left to interpret errors and then try to repair them by picking from any number of available strategies that they think might work. This creates inconsistency in the interface and impedes the ability to learn how to use the system (learnability) as well as the usability of collaborative machine translation systems in general.
Improving Efficiency and Accuracy in English Translation Learning: Investigating a Semantic Analysis Correction Algorithm
Published in Applied Artificial Intelligence, 2023
Machine translation reduces labor costs, improves translation efficiency, and has high application value. Traditional machine translation uses the pipeline traversal method to identify and analyze the input language one by one to obtain syntactic structure, and this method will pass on the errors that occur during the translation process, resulting in lower translation accuracy. To address this problem, this paper investigates the correction algorithm in English translation learning. First, we calculate the semantic similarity between the input sentences to be translated and the vocabulary in the source language in the instance library and construct a log-linear model, then use the dependency tree to string approach to select a suitable translation, implement the dependency structuring process on the source language side to assure the correspondence between Chinese and English, and further proofread the precise translation of English via the data-oriented translation approach. The test results show that the system improves the efficacy and precision of translation in English and can meet the needs of the users concerning English translation correction.
Machine translation model for effective translation of Hindi poetries into English
Published in Journal of Experimental & Theoretical Artificial Intelligence, 2022
Rajesh Kumar Chakrawarti, Jayshri Bansal, Pratosh Bansal
Today, most nations acknowledge Indian societies and attempt to get familiar with the Hindi language to achieve Indian culture. Hence, interpretations of Hindi writing and sonnets are extremely basic and significant to understand. The machine translation systems are automated computer programmes (software) capable of translating information available in one language (called the source language) into different dialects (called the objective language) (Alqudsi et al., 2014). With the support of machine translation frameworks, interpretations of Hindi’s writing into English are genuinely simple. Existing frameworks on interpretations of Hindi sonnets into English is vital and an inconceivable exercise in machine translation model. In this, lyrics assume a significant role in contrasted with written work of interpretations. Since ballads give sentiments, feelings, expression and more, the genuine interpretations of the sonnets are extremely significant.
Ambient intelligence framework for real-time speech-to-sign translation
Published in Assistive Technology, 2018
Mwaffaq Otoom, Mohammad A. Alzubaidi
Machine translation is the process of translating one language to another (Koehn, Och, & Marcu, 2003). To do so, parallel corpora need to be first prepared—where corresponding sentences, phrases, or words from the two languages are identified. For instance, in the context of English-text-to-ASL-sign translation, signs need to be first represented in a written form, called ASL text. English text and the counterpart ASL text represent the parallel corpora that are used by the different learning mechanisms of translation. An example of English text and its counterpart ASL text is: English Text: “Do you understand him?”ASL Text: “YOU UNDERSTAND HE?”