Doc-Handler: Document Scanner, Manipulator, and Translator based on Image and Natural language processing

Physical documents may easily be converted into digital versions in the modern digital era by employing scanning software and the internet. The day when this activity needed printers and scanners is long gone. Nowadays, even our smartphones and cameras may be used to quickly convert paper documents into digital ones. This is especially useful in the wake of the COVID-19 pandemic, where the ability to share and access documents online is more important than ever. This study proposes an application for illiterate people to quickly translate scanned papers or photos into their native language and save them in a digital format. The Application makes use of image processing methods and has capabilities including PDF conversion, image colour adjustment, cropping, and Optical Character Recognition (OCR). A user-friendly application, developed using the Flutter Framework and programmed in Python and Dart, serves as the interface for the system. The proposed application is cross-platform and works with a variety of gadgets. This method intends to increase accessibility and productivity for illiterate people in the digital age by integrating image processing with language translation. © 2023 IEEE.

Continuous Sign Language Interpretation to Text Using Deep Learning Models

The COVID-19 pandemic has obligated people to adopt the virtual lifestyle. Currently, the use of videoconferencing to conduct business meetings is prevalent owing to the numerous benefits it presents. However, a large number of people with speech impediment find themselves handicapped to the new normal as they cannot communicate their ideas effectively, especially in fast paced meetings. Therefore, this paper aims to introduce an enriched dataset using an action recognition method with the most common phrases translated into American Sign Language (ASL) that are routinely used in professional meetings. It further proposes a sign language detecting and classifying model employing deep learning architectures, namely, CNN and LSTM. The performances of these models are analysed by employing different performance metrics like accuracy, recall, F1- Score and Precision. CNN and LSTM models yield an accuracy of 93.75% and 96.54% respectively, after being trained with the dataset introduced in this study. Therefore, the incorporation of the LSTM model into different cloud services, virtual private networks and softwares will allow people with speech impairment to use sign language, which will automatically be translated into captions using moving camera circumstances in real time. This will in turn equip other people with the tool to understand and grasp the message that is being conveyed and easily discuss and effectuate the ideas. © 2022 IEEE.

Towards Large Vocabulary Kazakh-Russian Sign Language Dataset: KRSL-OnlineSchool

This paper presents a new dataset for Kazakh-Russian Sign Language (KRSL) created for the purposes of Sign Language Processing. In 2020, Kazakhstan's schools were quickly switched to online mode due to COVID-19 pandemic. Every working day, the El-arna TV channel was broadcasting video lessons for grades from 1 to 11 with sign language translation. This opportunity allowed us to record a corpus with a large vocabulary and spontaneous SL interpretation. To this end, this corpus contains video recordings of Kazakhstan's online school translated to Kazakh-Russian sign language by 7 interpreters. At the moment we collected and cleaned 890 hours of video material. A custom annotation tool was created to make the process of data annotation simple and easy-to-use by Deaf community. To date, around 325 hours of videos have been annotated with glosses and 4,009 lessons out of 4,547 were transcribed with automatic speech-to-text software. KRSL-OnlineSchool dataset will be made publicly available at https://krslproject.github.io/online-school/. © European Language Resources Association (ELRA), licensed under CC-BY-NC 4.0.

COVID-19 Mythbusters in World Languages

This paper introduces a multi-lingual database containing translated texts of COVID-19 mythbusters. The database has translations into 115 languages as well as the original English texts, of which the original texts are published by World Health Organization (WHO). This paper then presents preliminary analyses on latin-alphabet-based texts to see the potential of the database as a resource for multilingual linguistic analyses. The analyses on latin-alphabet-based texts gave interesting insights into the resource. While the amount of translated texts in each language was small, character bi-grams with normalization (lowercasing and removal of diacritics) turned out to be an effective proxy for measuring the similarity of the languages, and the affinity ranking of language pairs could be obtained. Additionally, the hierarchical clustering analysis is performed using the character bigram overlap ratio of every possible pair of languages. The result shows the cluster of Germanic languages, Romance languages, and Southern Bantu languages. In sum, the multilingual database not only offers fixed set of materials in numerous languages, but also serves as a preliminary tool to identify the language family using text-based similarity measure of bigram overlap ratio. © European Language Resources Association (ELRA), licensed under CC-BY-NC-4.0.

Harnessing Unsupervised Word Translation to Address Resource Inequality for Peace and Health

Research geared toward human well-being in developing nations often concentrates on web content written in a world language (e.g., English) and ignores a significant chunk of content written in a poorly resourced yet highly prevalent first language of the region in concern (e.g., Hindi). Such omissions are common due to the sheer mismatch between linguistic resources offered in a world language and its low-resource counterpart. However, during a global pandemic or an imminent war, demand for linguistic resources might get recalibrated. In this work, we focus on the high-resource and low-resource language pair ⟨ en, hie⟩ (English, and Romanized Hindi) and present a cross-lingual sampling method that takes example documents in English, and retrieves similar content written in Romanized Hindi, the most popular form of Hindi observed in social media. At the core of our technique is a novel finding that a surprisingly simple constrained nearest-neighbor sampling in polyglot Skip-gram word embedding space can retrieve substantial bilingual lexicons, even from harsh social media data sets. Our cross-lingual sampling method obtains substantial performance improvement in the important domains of detecting peace-seeking, hostility-diffusing hope speech in the context of the 2019 India-Pakistan conflict, and in detecting comments encouraging compliance with COVID-19 guidelines. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.

Approaching SMM4H with auto-regressive language models and back-translation

We describe our submissions to the 6th edition of the Social Media Mining for Health Applications (SMM4H) shared task. Our team (OGNLP) participated in the sub-task: Classification of tweets self-reporting potential cases of COVID-19 (Task 5). For our submissions, we employed systems based on autoregressive transformer models (XLNet) and back-translation for balancing the dataset. © 2021 Association for Computational Linguistics.

A Real-Time Translation of Arabic Video Contents to Arabic Sign Language

The shift to online education due to the COVID-19 pandemic has brought various challenges to different individuals. Among these affected individuals are deaf and hearing-impaired. They face difficulties in understanding online videos because they do not provide real-time translation into Sign languages. The main idea of this project is derived from this point. It proposes a PC-based application called "Al-Banan"that supports deaf and hearing-impaired in the Arabic region to understand media in general including learning video and live virtual classes by translating Basic Arabic Speech into Arabic Sign language using a three-dimensional avatar. The application was built done using Python and C# languages. The application was tested on many deaf through social media and proved to be effective if the word exists in the dictionary a three-dimensional cartoon avatar showing the corresponding Saudi Arabia Sign Language. Otherwise, the word is represented by finger spelling. The results proved the understanding and acceptance of the idea by the deaf. © 2021 IEEE.

Legal Informatics of HS Code Automatic Compliance Translation Based on Cross-border Trade Digitization

Trade facilitation has been an inseparable topic in international trade since the adoption of the agreement in cross-border trade. Trade facilitation implementation has also shifted from trade informatization to trade intelligence and then to trade digitization. Furthermore, in the post-COVID-19 era, people are more aware that the future world will be digital. The research focuses on the use of legal informatics in cross-border trade digitization. The study will attempt to construct a system model that includes the core of HS Code automatic compliance translation as well as the functions of automatic production, translation, sharing, and intelligent pre-review of legal texts at various ports. To reduce the implementation cost of cross-border trade digitization and improve customs clearance efficiency, attempt to understand, and process legal texts with informatization and intelligence, including the processing and reprocessing of legal data. © 2022 IEEE.

Findings of the WMT Shared Task on Machine Translation Using Terminologies

Language domains that require very careful use of terminology are abundant and reflect a significant part of the translation industry. In this work we introduce a benchmark for evaluating the quality and consistency of terminology translation, focusing on the medical (and COVID-19 specifically) domain for five language pairs: English to French, Chinese, Russian, and Korean, as well as Czech to German. We report the descriptions and results of the participating systems, commenting on the need for further research efforts towards both more adequate handling of terminologies as well as towards a proper formulation and evaluation of the task. © 2021 Association for Computational Linguistics

Improving the Virtual Educational Platforms for the Deaf and Dumb under the Covid-19 Pandemic Circumstances

Arabic Sign Language (ArSL) is a language of communication between deaf dumb people in Arab countries. This study focuses on establishing a real-time speech conversion service into ArSL by 3D animation videos, to facilitate the learning process in virtual educational platforms. This study to help the deaf dumb students in distant learning under the Covid-19 pandemic. This study applied on the TEAMS platform. A database was created containing more than 550 ArSL videos from the Al-tarjuman application and connected it with the TEAMS platform. A Python was used to link the System units. After studying some related works that were concerned with the language of the deaf dumb, we found that some studies do not support real-time translation, has lack Arabic sign language translation, has limited DB, resulting in a low level of system performance, and use of high-cost devices such as gesture motion sensors 'Kinect and Leap Motion'. As future work, we will increase the words in the DB to improve the results. We advise researchers to contribute by working on merging the videos retrieved from the DB to make an integrated serial video. Also, adding a syntactic analysis stage to extract the sentence structure. © 2022 IEEE.

Exams with Internet Access in a Fundamental Engineering Course

This study integrated online resources into some exams in the Statics course in a Thai University. As students were expected to use mathematical software to solve complex mathematical formula, internet access was also allowed to simulate a more realistic working context. The exam questions had to be carefully set to assess the in-depth learning and application, focusing on the physical meanings, problem setting, modeling and result interpretation as the previously time-consuming mathematical solving was taken care of by the computer. The internet access had affected various aspects of exams, particularly the data formula, language translation, unit conversion and the ability to describe questions with colored photos or videos. The availability of internet resources further penalized students who did not properly prepare for the exams;they faced the problems in the online time management, utilization of academic and professional database, and the use of online translation in the disciplinary context. The initial concern for cheating appeared to be minor. Generally, students were either eager for or wary on exams with internets but they all seemed to recognize the reality of the needs and challenges. The proposal could also address some of the online exam issues in the COVID-19 era and the new normal after. © 2021 IEEE.

A Cross-Cultural Adaptation of Chatbot Usability Questionnaire (CUQ): Indonesian Version

Chatbots are a highly developed technology, especially in the current COVID-19 pandemic to apply contactless activities. Unfortunately, most of companies that have been implemented chatbot has failed gained benefit from the chatbot. This shows companies need to evaluate usability of the chatbot. One of specific questionnaire to evaluate usability of chatbot is Chatbot Usability Questionnaire (CUQ). Currently CUQ is only available in English and there is no research that provides CUQ in the Indonesian version. This research focuses on translating CUQ and producing a reliable Indonesian version of CUQ. CUQ needs to be adapted naturally into Indonesian so it can be understood by all types of users. Cross-cultural adaptation approach is used to translate CUQ to Indonesia. The reliability test was run with 100 respondents and resulted in Cronbach's alpha value of 0.749. This value indicated that CUQ is reliable and can be used in chatbot research and evaluation. © 2021 IEEE

(Re)Designing Technical Documentation About COVID-19 with and for Indigenous Communities in Gainesville, Florida, Oaxaca de Juárez, Mexico, and Quetzaltenango, Guatemala

Background: In this article, we document how our team of translators, interpreters, technical communicators, and health justice workers is collaborating to (re)design COVID-19-related technical documentation for and with Indigenous language speakers in Gainesville, FL, USA;Oaxaca de Juarez, Mexico;and Quetzaltenango, Guatemala. Literature review: Although (mis)representations of Indigenous communities have been an ongoing issue in and beyond technical communication, the COVID-19 pandemic has brought added attention to how government institutions and other agencies fail to consider the cultural values, languages, and communication practices of Indigenous communities when writing, designing, and sharing technical information. Research questions: 1. How can technical communicators work toward social justice in health through collaborative design with Indigenous language speakers?2. How can technical documentation about COVID-19 be (re)designed alongside members of vulnerable communities to redress oppressive representations while increasing access and usability?Methodology: Through interviews and other participatory design activities conducted with Indigenous language speakers in the US, Guatemala, and Mexico, we illustrate how Western approaches to creating technical documentation, particularly in health-related contexts such as the COVID-19 pandemic, put communities at risk by failing to localize health messaging for Indigenous audiences. We then document our work intended to collaboratively design and translate COVID-19-related technical information alongside those Indigenous language speakers to benefit Indigenous language speakers in Gainesville and other parts of North Central Florida. Results: Through this discussion, we highlight how technical communicators can collaborate with Indigenous language speakers to create, translate, and share multilingual technical documents that can contribute to social justice efforts by enhancing language access. Conclusion: Through collaborations with Indigenous language speakers, translators, and interpreters, social/health justice projects in technical communication can be combined, localized, and adapted to better serve and represent the diversity of people, languages, and cultures that continue to increase in our world. IEEE

Ixamed's submission description for WMT20 Biomedical shared task: Benefits and limitations of using terminologies for domain adaptation

In this paper we describe the systems developed at Ixa for our participation in WMT20 Biomedical shared task in three language pairs, en-eu, en-es and es-en. When defining our approach, we have put the focus on making an efficient use of corpora recently compiled for training Machine Translation (MT) systems to translate Covid-19 related text, as well as reusing previously compiled corpora and developed systems for biomedical or clinical domain. Regarding the techniques used, we base on the findings from our previous works for translating clinical texts into Basque, making use of clinical terminology for adapting the MT systems to the clinical domain. However, after manually inspecting some of the outputs generated by our systems, for most of the submissions we end up using the system trained only with the basic corpus, since the systems including the clinical terminologies generated outputs shorter in length than the corresponding references. Thus, we present simple baselines for translating s between English and Spanish (en/es);while for translating s and terms from English into Basque (en-eu), we concatenate the best en-es system for each kind of text with our es-eu system. We present automatic evaluation results in terms of BLEU scores, and analyse the effect of including clinical terminology on the average sentence length of the generated outputs. Following the recent recommendations for a responsible use of GPUs for NLP research, we include an estimation of the generated CO2 emissions, based on the power consumed for training the MT systems. © 2020 Association for Computational Linguistics