translation accuracy
Recently Published Documents

One of efforts by the Indonesian people to defend the country is to preserve and to maintain the regional languages. The current era of modernity makes the regional language image become old-fashioned, so that most them are no longer spoken.  If it is ignored, then there will be a cultural identity crisis that causes regional languages to be vulnerable to extinction. Technological developments can be used as a way to preserve regional languages. Digital image-based artificial intelligence technology using machine learning methods such as machine translation can be used to answer the problems. This research will use Deep Learning method, namely Convolutional Neural Networks (CNN). Data of this research were 1300 alphabetic images, 5000 text images and 200 vocabularies of Minangkabau regional language. Alphabetic image data is used for the formation of the CNN classification model. This model is used for text image recognition, the results of which will be translated into regional languages. The accuracy of the CNN model is 98.97%, while the accuracy for text image recognition (OCR) is 50.72%. This low accuracy is due to the failure of segmentation on the letters i and j. However, the translation accuracy increases after the implementation of the Leveinstan Distance algorithm which can correct text classification errors, with an accuracy value of 75.78%. Therefore, this research has succeeded in implementing the Convolutional Neural Networks (CNN) method in identifying text in text images and the Leveinstan Distance method in translating Indonesian text into regional language texts.  

Language translation is essential to bring the world closer and plays a significant part in building a community among people of different linguistic backgrounds. Machine translation dramatically helps in removing the language barrier and allows easier communication among linguistically diverse communities. Due to the unavailability of resources, major languages of the world are accounted as low-resource languages. This leads to a challenging task of automating translation among various such languages to benefit indigenous speakers. This article investigates neural machine translation for the English–Assamese resource-poor language pair by tackling insufficient data and out-of-vocabulary problems. We have also proposed an approach of data augmentation-based NMT, which exploits synthetic parallel data and shows significantly improved translation accuracy for English-to-Assamese and Assamese-to-English translation and obtained state-of-the-art results.

AbstractTranslation of the genetic code into proteins is realized through repetitions of synchronous translocation of messenger RNA (mRNA) and transfer RNAs (tRNA) through the ribosome. In eukaryotes translocation is ensured by elongation factor 2 (eEF2), which catalyses the process and actively contributes to its accuracy1. Although numerous studies point to critical roles for both the conserved eukaryotic posttranslational modification diphthamide in eEF2 and tRNA modifications in supporting the accuracy of translocation, detailed molecular mechanisms describing their specific functions are poorly understood. Here we report a high-resolution X-ray structure of the eukaryotic 80S ribosome in a translocation-intermediate state containing mRNA, naturally modified eEF2 and tRNAs. The crystal structure reveals a network of stabilization of codon–anticodon interactions involving diphthamide1 and the hypermodified nucleoside wybutosine at position 37 of phenylalanine tRNA, which is also known to enhance translation accuracy2. The model demonstrates how the decoding centre releases a codon–anticodon duplex, allowing its movement on the ribosome, and emphasizes the function of eEF2 as a ‘pawl’ defining the directionality of translocation3. This model suggests how eukaryote-specific elements of the 80S ribosome, eEF2 and tRNAs undergo large-scale molecular reorganizations to ensure maintenance of the mRNA reading frame during the complex process of translocation.

The most popular video website YouTube has about 2 billion users worldwide who speak and understand different languages. Subtitles are essential for the users to get the message from the video. However, not all video owners provide subtitles for their videos. It causes the potential audiences to have difficulties in understanding the video content. Thus, this study proposed a speech recorder and translator to solve this problem. The general concept of this study was to combine Automatic Speech Recognition (ASR) and translation technologies to recognize the video content and translate it into other languages. This paper compared and discussed three different ASR technologies. They are Google Cloud Speech-to-Text, Limecraft Transcriber, and VoxSigma. Finally, the proposed system used Google Cloud Speech-to-Text because it supports more languages than Limecraft Transcriber and VoxSigma. Besides, it was more flexible to use with Google Cloud Translation. This paper also consisted of a questionnaire about the crucial features of the speech recorder and translator. There was a total of 19 university students participated in the questionnaire. Most of the respondents stated that high translation accuracy is vital for the proposed system. This paper also discussed a related work of speech recorder and translator. It was a study that compared speech recognition between ordinary voice and speech impaired voice. It used a mobile application to record acoustic voice input. Compared to the existing mobile App, this project proposed a web application. It was a different and new study, especially in terms of development and user experience. Finally, this project developed the proposed system successfully. The results showed that Google Cloud Speech-to-Text and Translation were reliable to use in video translation. However, it could not recognize the speech when the background music was too loud. Besides, it had a problem of direct translation, which was challenging. Thus, future research may need a custom trained model. In conclusion, the proposed system in this project was to contribute a new idea of a web application to solve the language barrier on the video watching platform.

In the traditional sense, the translation evaluation of English complex long sentences is often limited to the idea of whether or how to realize the semantic transformation of the original text, so many phenomena that have nothing to do with language but directly affect the translation evaluation are not included in the field of vision and can be interpreted. In order to solve the above problems, a multi-label clustering algorithm is proposed to evaluate the translation accuracy of English complex long sentences. The multi-label clustering algorithm is introduced into the translation evaluation activities to carry out the translation and detection parameters of complex long sentences. The comprehensive description, the accuracy of generalization and the rationality of interpretation lay a solid foundation for English translation activities.

Banjar language used in conversation and daily life around the area. So foreigners who come to the regions of South Kalimantan will have difficulty in communicating. Besides, most local residents in the backwoods of South Kalimantan can not use Indonesian language properly, they would be more convenient to use regional language to interact. For that reason we need an Android application can help users to find the translation of a word or phrase whenever and wherever. With the help of Google Voice Search, this application can also listen to the voice of the user to be converted into text and insert into the input translation. Speech recognition of Banjar language required a literacy training data by using the method of statistical inference to make results appropriated. Testing using method of Black Box Testing to measure the percentage of suitability of the results of translation, speech recognition for Indonesian language and speech recognition Banjar language using method of Statistical inference. So the results of translation accuracy 100% and accuracy of speech recognition Indonesian language and Banjar language by 97.85% and 82.74%.

The use of neural machine algorithms for English translation is a hot topic in the current research. English translation using the traditional sequential neural framework, which is too poor at capturing long-distance information, has its own major limitations. However, the current improved frameworks, such as recurrent neural network translation, are not satisfactory either. In this paper, we establish an attention coding and decoding model to address the shortcomings of traditional machine translation algorithms, combine the attention mechanism with a neural network framework, and implement the whole English translation system based on TensorFlow, thus improving the translation accuracy. The experimental test results show that the BLUE values of the algorithm model built in this paper are improved to different degrees compared with the traditional machine learning algorithms, which proves that the performance of the proposed algorithm model is significantly improved compared with the traditional model.

The research addresses how Sundanese terms of address are translated into English. It discusses specifically the translation techniques practiced by the translator which affect the quality of the translation. The data are taken from Sundanese short stories and their translation in English. The theories exercised to find out the applied techniques are based on Molina and Albir (2002). The translation quality is examined using translation quality assessment of Nababan et al (2012). The analyses are carried out using Santosa’s methods (2017), a modification of Spreadly’s, following the analysis steps of domain, taxonomy, componential and culture findings. The results show that there are five translation techniques practiced by the translator, namely established equivalent, pure borrowings, deletion, variation, and implicit. The translation quality appears to gain 2.7. This score means that the translation is quite good. While the translation accuracy takes the highest score of 2.9, followed by acceptability 2.8, and readability having 2.5.