Machine Translation (MT) Evaluation is an integral part of the MT development life cycle. Without analyzing the outputs of MT engines, it is impossible to evaluate the performance of an MT system. Through experiments, it has been identified that what works for English and other European languages does not work well with Indian languages. Thus, In this paper, we have introduced a reference-based MT evaluation metric for Gujarati which is based on supervised learning. We have trained two versions of the metric which uses 25 features for training. Among the two models, one model is trained using 6 hidden layers with 500 epochs while the other model is trained using 10 hidden layers with 500 epochs. To test the performance of the metric, we collected 1000 MT outputs of seven MT systems. These MT engine outputs were compared with 1 human reference translation. While comparing the developed metrics with other available metrics, it was found that the metrics produced better human correlations.

This work examines machine translation systems from a human perspective. The modern era of neural machine translation and its applicability. A scalar quality metric comprising 11 factors was employed in the study, and it was graded on an 0–4 scale by humans. IndicTrans2 and Facebook’s NLLB MT Engines were the two MT engines that were taken into consideration for the research. In order to determine the common mistakes that each MT engine encountered, the human evaluators conducted assessments using eleven parameters. The results showed that, of the two, IndicTrans2 MT Engine yielded superior outcomes.

The chapter shows the development of a part-of-speech (POS) tagger for Sindhi, which is a highly resource-poor language. For our study, we have used Sindhi in the Devanagari script. We have developed a corpus of thirty thousand POS-tagged sentences and developed Glove word embeddings for the monolingual Sindhi corpus of 1 lac sentences. We have used LSTM and Glove word embeddings for developing our POS tagger. The developed system was compared with a Hidden Markov Model (HMM)-based POS tagger developed for Sindhi (Nathani and Joshi, Part of speech tagging for a resource poor language: Sindhi in Devanagari script using HMM and CRF. In: Proceedings of the 18th international conference on natural language processing, 2021). The evaluation shows significant improvement over the previous study. The HMM POS tagger achieved an overall accuracy of 81.85

Language is a primary means of communication. It is a medium through which we can interact with society. Recognizing it, each language has its own set of grammatical rules. This study focused on the development of a rule-based chunker for a resource-poor language Sindhi using the Devanagari script. We have chosen a rule-based approach as language itself is a sequence of rules. This approach is fairly useful due to its ability to capture nuances of a language. The language rules were created and validated with the help of language experts. To develop the chunker, 50,000 sentences were used for the construction of rules. These sentences belonged to various domains like travel and tourism, health and administration. For this study, there was a requirement for a POS-tagged dataset. The data was annotated using Part of Speech tagger based on the Hidden Markov Model (HMM). 1,000 sentences were used to evaluate the system. The developed chunker showed an accuracy of 97.8

In this paper we have shown the development of English to Arabic Braille Neural Machine Translation (NMT) System. For our experiments we have developed two NMT systems. The first was the baseline NMT system which was trained only on the English-Arabic parallel corpus whereas the second NMT system has some sub modules for handling syntax transfer of English to Arabic and translation/transliteration of English Named Entities into Arabic equivalents. Then the system used this code-mixed text to train the NMT model. The English-Arabic parallel corpus was taken from Opus parallel corpus repository. For preprocessing of the English text Stanford's Stanza NLP Library was used. The trained NMT models were evaluated using standard automatic MT evaluation metrics and their results were also corelated with human judgements. It was found that the corpus augmented NMT system performed better than the baseline NMT system.

In this paper, we have shown the development of a Part of Speech (POS) tagger for Hadoti - a prominent language spoken in Rajasthan, India - despite its limited resources. For this, we manually tagged a corpus of 50,000 POS-tagged sentences and trained it using a Hidden Markov Model (HMM). Since no prior work had been reported in this area, we couldn't compare our results to any other system. This paper documents the efforts made to create an HMM POS tagger for Hadoti, to stimulate further research in this field. This work is expected to serve as a foundation for preserving the language and as a resource for aspiring researchers who wish to explore this area of Hadoti Language Processing. The system was evaluated for accuracy and produced 99.87

Machine Translation has been one of the main areas of research in the dawn of Artificial Intelligence. Although getting a human-level accuracy through a machine was considered as a task that is next to impossible, this cliché is on the verge of being broken. The paper discusses the development of an

Question Answering (QA) has proved to be an arduous challenge in the area of natural language processing (NLP) and artificial intelligence (AI). Many attempts have been made to develop complete solutions for QA as well as improving significant sub-modules of the QA systems to improve the overall performance through the course of time. Questions are the most important piece of QA, because knowing the question is equivalent to knowing what counts as an answer (Harrah in Philos Sci, 1961 [1]). In this work, we have attempted to understand questions in a better way by using Quantum Machine Learning (QML). The properties of Quantum Computing (QC) have enabled classically intractable data processing. So, in this paper, we have performed question classification on questions from two classes of SelQA (Selection-based Question Answering) dataset using quantum-based classifier algorithms-quantum support vector machine (QSVM) and variational quantum classifier (VQC) from Qiskit (Quantum Information Science toolKIT) for Python. We perform classification with both classifiers in almost similar environments and study the effects of circuit depths while comparing the results of both classifiers. We also use these classification results with our own rule-based QA system and observe significant performance improvement. Hence, this experiment has helped in improving the quality of QA in general.

In this paper, we have shown a script conversion (transliteration) technique that converts Sindhi text in the Devanagari script to the Perso-Arabic script. We showed this by incorporating a hybrid approach where some part of the text is converted using a rule base and in case an ambiguity arises then a probabilistic model is used to resolve the same. Using this approach, the system achieved an overall accuracy of 99.64%.

In this paper, we have shown a method of improving the quality of neural machine translation by translating/transliterating name entities as a preprocessing step. Through experiments we have shown the performance gain of our system. For evaluation we considered three types of name entities viz person names, location names and organization names. The system was able to correctly translate mostly all the name entities. For person names the accuracy was 99.86%, for location names the accuracy was 99.63% and for organization names the accuracy was 99.05%. Overall, the accuracy of the system was 99.52%

Since the beginning of machine translation (MT) research, MT evaluation has been an area of interest of researchers. In literature, one can find more papers on MT evaluation than on machine translation itself. This paper describes the work done on developing our MT evaluation metric which incorporat

In this paper, the expansion of English to Bharti Braille neural machine translation system has been carried out. It is shown how a baseline Neural Machine Translation (NMT) model can be improved by adding some linguistic knowledge to it. This was done for five language pairs where English sentences were translated into five Indian languages and then subsequently to corresponding Bharti Braille. This has been demonstrated by adding a sub-module for translating name entities. The approach shows promising results across language pair and improvement in the quality of NMT outputs. The least improvement was observed in English-Tamil language pair with 8.79

In this paper, we have shown the improvement of English to Bharti Braille machine translation system. We have shown how we can improve a baseline NMT model by adding some linguistic knowledge to it. This was done for five language pairs where English sentences were translated into five Indian languages and then subsequently to corresponding Bharti Braille. This has been demonstrated by adding a submodule for translating multi-word expressions. The approach shows promising results as across language pairs, we could see improvement in the quality of NMT outputs. The least improvement was observed in English-Nepali language pair with 22.08% and the most improvement was observed in the English-Hindi language pair with 23.30%

People who are visually impaired face a lot of difficulties while studying. One of the major causes to this is lack of available text in Bharti Braille script. In this paper, we have suggested a scheme to convert text in major Indian languages into Bharti Braille. The system uses a hybrid approach where at first the text in Indian language is given to a rule based system and in case if there is any ambiguity then it is resolved by applying a LSTM based model. The developed model has also been tested and found to have produced near accurate results.

Human mind is the palace of curious questions that seek answers. Computational resolution of this challenge is possible through Natural Language Processing techniques. Statistical techniques like machine learning and deep learning require a lot of data to train and despite that they fail to tap into the nuances of language. Such systems usually perform best on close-domain datasets. We have proposed development of a rule-based open-domain question-answering system which is capable of answering questions of any domain from a corresponding context passage. We have used 1000 questions from SQuAD 2.0 dataset for testing the developed system and it gives satisfactory results. In this paper, we have described the structure of the developed system and have analyzed the performance.

Hindi is the third most-spoken language in the world (615 million speakers) and has the fourth highest native speakers (341 million). It is an inflectionally rich and relatively free word-order language with an immense vocabulary set. Despite being such a celebrated language across the globe, very few Natural Language Processing (NLP) applications and tools have been developed to support it computationally. Moreover, most of the existing ones are not efficient enough due to the lack of semantic information (or contextual knowledge). Hindi grammar is based on Paninian grammar and derives most of its rules from it. Paninian grammar very aggressively highlights the role of karaka theory in free-word order languages. In this article, we present an application that extracts all possible karakas from simple Hindi sentences with an accuracy of 84.2% and an F1 score of 88.5%. We consider features such as Parts of Speech tags, post-position markers (vibhaktis), semantic tags for nouns and syntactic structure to grab the context in different-sized word windows within a sentence. With the help of these features, we built a rule-based inference engine to extract karakas from a sentence. The application takes in a text file with clean (without punctuation) simple Hindi sentences and gives back karaka tagged sentences in a separate text file as output.

Life is tough for people who are blind or deafblind. Braille is a medium through which they can read the text available in the world. Unfortunately, due to the lack of resources that can transcribe textual material into Braille, they are deprived of a lot of knowledge available in books. This paper is an attempt to address this issue. We have developed a system that can transcribe the text available in English into Hindi Braille. We have developed a hybrid machine assisted translation system, which can translate English text into Bharati(Hindi) Braille. This research would be helpful to people in our country who have a desire to learn and gain knowledge but are unable to do so due to the unavailability of literature in Bharati Braille. The developed system is critically important to the lives of visually impaired people because the ability to read and write in Braille opens the door to literacy, intellectual freedom, equal opportunity, and personal security for them. We have evaluated our MT system against standard MT evaluation metrics which provides promising results. We also did quality analysis of the MT output and found that for almost all the simple sentences, and the translations were good while for complex sentences. The system produced above average translations.

Machine learning and quantum computing fuse together to form quantum machine learning. Although the phenomenon is new, it has already proved its worth in various fields like finance and chemistry. The potential of quantum computing and its extraordinary properties enable us to process data in a way classical computer can never think of. When machine learning gets the power of quantum computing, information processing is enhanced significantly. In this paper, we have used variational quantum classifiers to classify questions from two domains of SelQA dataset. We keep the focus on the implications of circuit-depth in different experiments and analyze the results. VQC performs well with 11 features on lowest circuit depths and gives a testing accuracy of 58%.