Repetition in natural language generation reduces the informativeness of text and makes it less appealing. Various techniques have been proposed to alleviate it. In this work, we explore and propose techniques to reduce repetition in abstractive summarization. First, we explore the application of unlikelihood training and embedding matrix regularizers from previous work on language modeling to abstractive summarization. Next, we extend the coverage and temporal attention mechanisms to the token level to reduce repetition. In our experiments on the CNN/Daily Mail dataset, we observe that these techniques reduce the amount of repetition and increase the informativeness of the summaries, which we confirm via human evaluation.

Large scale pretrained models have demonstrated strong performances on several natural language generation and understanding benchmarks. However, introducing commonsense into them to generate more realistic text remains a challenge. Inspired from previous work on commonsense knowledge generation and generative commonsense reasoning, we introduce two methods to add commonsense reasoning skills and knowledge into abstractive summarization models. Both methods beat the baseline on ROUGE scores, demonstrating the superiority of our models over the baseline. Human evaluation results suggest that summaries generated by our methods are more realistic and have fewer commonsensical errors.

This paper reports a Machine Translation (MT) system submitted by the NLPRL team for the Bhojpuri–Hindi and Magahi–Hindi language pairs at LoResMT 2020 shared task. We used an unsupervised domain adaptation approach that gives promising results for zero or extremely low resource languages. Task organizers provide the development and the test sets for evaluation and the monolingual data for training. Our approach is a hybrid approach of domain adaptation and back-translation. Metrics used to evaluate the trained model are BLEU, RIBES, Precision, Recall and F-measure. Our approach gives relatively promising results, with a wide range, of 19.5, 13.71, 2.54, and 3.16 BLEU points for Bhojpuri to Hindi, Magahi to Hindi, Hindi to Bhojpuri and Hindi to Magahi language pairs, respectively.

Automated grammatical error correction has been explored as an important research problem within NLP, with the majority of the work being done on English and similar resource-rich languages. Grammar correction using neural networks is a data-heavy task, with the recent state of the art models requiring datasets with millions of annotated sentences for proper training. It is difficult to find such resources for Indic languages due to their relative lack of digitized content and complex morphology, compared to English. We address this problem by generating a large corpus of artificial inflectional errors for training GEC models. Moreover, to evaluate the performance of models trained on this dataset, we create a corpus of real Hindi errors extracted from Wikipedia edits. Analyzing this dataset with a modified version of the ERRANT error annotation toolkit, we find that inflectional errors are very common in this language. Finally, we produce the initial baseline results using state of the art methods developed for English.

Parsing news Headlines is one of the difficult tasks of Natural Language Processing. It is mostly because news Headlines (NHs) are not complete grammatical sentences. News editors use all sorts of tricks to grab readers’ attention, for instance, unusual capitalization as in the headline’ Ear SHOT ashok rajagopalan’; some are world knowledge demanding like ‘Church reformation celebrated’ where the ‘Church reformation’ refers to a historical event and not a piece of news about an ordinary church. The lack of transparency in NHs can be linguistic, cultural, social, or contextual. The lack of space provided for a news headline has led to creative liberty. Though many works like news value extraction, summary generation, emotion classification of NHs have been going on, parsing them had been a tough challenge. Linguists have also been interested in NHs for creativity in the language used by bending traditional grammar rules. Researchers have conducted studies on news reportage, discourse analysis of NHs, and many more. While the creativity seen in NHs is fascinating for language researchers, it poses a computational challenge for Natural Language Processing researchers. This paper presents an outline of the ongoing doctoral research on the parsing of Indian English NHs. The ultimate aim of this research is to provide a module that will generate correctly parsed NHs. The intention is to enhance the broad applicability of newspaper corpus for future Natural Language Processing applications.

This paper reports the results for the Machine Translation (MT) system submitted by the NLPRL team for the Hindi – Marathi Similar Translation Task at WMT 2020. We apply the Transformer-based Neural Machine Translation (NMT) approach on both translation directions for this language pair. The trained model is evaluated on the corpus provided by shared task organizers, using BLEU, RIBES, and TER scores. There were a total of 23 systems submitted for Marathi to Hindi and 21 systems submitted for Hindi to Marathi in the shared task. Out of these, our submission ranked 6th and 9th, respectively.

This paper describes the results of the system that we used for the WMT20 very low resource (VLR) supervised MT shared task. For our experiments, we use a byte-level version of BPE, which requires a base vocabulary of size 256 only. BPE based models are a kind of sub-word models. Such models try to address the Out of Vocabulary (OOV) word problem by performing word segmentation so that segments correspond to morphological units. They are also reported to work across different languages, especially similar languages due to their sub-word nature. Based on BLEU cased score, our NLPRL systems ranked ninth for HSB to GER and tenth in GER to HSB translation scenario.

The Coronavirus pandemic has been a dominating news on social media for the last many months. Efforts are being made to reduce its spread and reduce the casualties as well as new infections. For this purpose, the information about the infected people and their related symptoms, as available on social media, such as Twitter, can help in prevention and taking precautions. This is an example of using noisy text processing for disaster management. This paper discusses the NLPRL results in Shared Task-2 of WNUT-2020 workshop. We have considered this problem as a binary classification problem and have used a pre-trained ELMo embedding with GRU units. This approach helps classify the tweets with accuracy as 80.85% and 78.54% as F1-score on the provided test dataset. The experimental code is available online.

This paper describes the Machine Translation system for Tamil-English Indic Task organized at WAT 2019. We use Transformer- based architecture for Neural Machine Translation.

The contrast between the contextual and general meaning of a word serves as an important clue for detecting its metaphoricity. In this paper, we present a deep neural architecture for metaphor detection which exploits this contrast. Additionally, we also use cost-sensitive learning by re-weighting examples, and baseline features like concreteness ratings, POS and WordNet-based features. The best performing system of ours achieves an overall F1 score of 0.570 on All POS category and 0.605 on the Verbs category at the Metaphor Shared Task 2018.

This paper describes the best performing system for the shared task on Named Entity Recognition (NER) on code-switched data for the language pair Spanish-English (ENG-SPA). We introduce a gated neural architecture for the NER task. Our final model achieves an F1 score of 63.76%, outperforming the baseline by 10%.

Text language Identification is a Natural Language Processing task of identifying and recognizing a given language out of many different languages from a piece of text. This paper describes our submission to the ILI 2018 shared-task, which includes the identification of 5 closely related Indo-Aryan languages. We developed a word-level LSTM(Long Short-term Memory) model, a specific type of Recurrent Neural Network model, for this task. Given a sentence, our model embeds each word of the sentence and convert into its trainable word embedding, feeds them into our LSTM network and finally predict the language. We obtained an F1 macro score of 0.836, ranking 5th in the task.

An accurate language identification tool is an absolute necessity for building complex NLP systems to be used on code-mixed data. Lot of work has been recently done on the same, but there’s still room for improvement. Inspired from the recent advancements in neural network architectures for computer vision tasks, we have implemented multichannel neural networks combining CNN and LSTM for word level language identification of code-mixed data. Combining this with a Bi-LSTM-CRF context capture module, accuracies of 93.28% and 93.32% is achieved on our two testing sets.

Social media based micro-blogging sites like Twitter have become a common source of real-time information (impacting organizations and their strategies, and are used for expressing emotions and opinions. Automated analysis of such content therefore rises in importance. To this end, we explore the viability of using deep neural networks on the specific task of emotion intensity prediction in tweets. We propose a neural architecture combining convolutional and fully connected layers in a non-sequential manner - done for the first time in context of natural language based tasks. Combined with lexicon-based features along with transfer learning, our model achieves state-of-the-art performance, outperforming the previous system by 0.044 or 4.4% Pearson correlation on the WASSA’17 EmoInt shared task dataset. We investigate the performance of deep multi-task learning models trained for all emotions at once in a unified architecture and get encouraging results. Experiments performed on evaluating correlation between emotion pairs offer interesting insights into the relationship between them.

This paper describes our participation in SemEval 2018 Task 3 on Irony Detection in Tweets. We combine linguistic features with pre-trained activations of a neural network. The CNN is trained on the emoji prediction task. We combine the two feature sets and feed them into an XGBoost Classifier for classification. Subtask-A involves classification of tweets into ironic and non-ironic instances whereas Subtask-B involves classification of the tweet into - non-ironic, verbal irony, situational irony or other verbal irony. It is observed that combining features from these two different feature spaces improves our system results. We leverage the SMOTE algorithm to handle the problem of class imbalance in Subtask-B. Our final model achieves an F1-score of 0.65 and 0.47 on Subtask-A and Subtask-B respectively. Our system ranks 4th on both tasks respectively, outperforming the baseline by 6% on Subtask-A and 14% on Subtask-B.

This paper describes an ensemble system submitted as part of the LSDSem Shared Task 2017 - the Story Cloze Test. The main conclusion from our results is that an approach based on semantic similarity alone may not be enough for this task. We test various approaches and compare them with two ensemble systems. One is based on voting and the other on logistic regression based classifier. Our final system is able to outperform the previous state of the art for the Story Cloze test. Another very interesting observation is the performance of sentiment based approach which works almost as well on its own as our final ensemble system.

Unlike Entity Disambiguation in web search results, Opinion Disambiguation is a relatively unexplored topic. RevOpiD shared task at IJCNLP-2107 aimed to attract attention towards this research problem. In this paper, we summarize the first run of this task and introduce a new dataset that we have annotated for the purpose of evaluating Opinion Mining, Summarization and Disambiguation methods.

A treebank is an important resource for developing many NLP based tools. Errors in the treebank may lead to error in the tools that use it. It is essential to ensure the quality of a treebank before it can be deployed for other purposes. Automatic (or semi-automatic) detection of errors in the treebank can reduce the manual work required to find and remove errors. Usually, the errors found automatically are manually corrected by the annotators. There is not much work reported so far on error correction tools which helps the annotators in correcting errors efficiently. In this paper, we present such an error correction tool that is an extension of the error detection method described earlier (Ambati et al., 2010; Ambati et al., 2011; Agarwal et al., 2012).

The usefulness of annotated corpora is greatly increased if there is an associated tool that can allow various kinds of operations to be performed in a simple way. Different kinds of annotation frameworks and many query languages for them have been proposed, including some to deal with multiple layers of annotation. We present here an easy to learn query language for a particular kind of annotation framework based on ‘threaded trees', which are somewhere between the complete order of a tree and the anarchy of a graph. Through 'typed' threads, they can allow multiple levels of annotation in the same document. Our language has a simple, intuitive and concise syntax and high expressive power. It allows not only to search for complicated patterns with short queries but also allows data manipulation and specification of arbitrary return values. Many of the commonly used tasks that otherwise require writing programs, can be performed with one or more queries. We compare the language with some others and try to evaluate it.

Language resources can be classified under several categories. To be able to query and operate on all (or most of) these categories using a single digital tool would be very helpful for a large number of researchers working on languages. We describe such a tool in this paper. It is different from other such tools in that it allows querying and transformation on different kinds of resources (such as corpora, lexicon and language models) with the same framework. Search options can be given based on the kind of resource being queried. It is possible to select a matched resource and open it for editing in the specialized interfaces with which that resource is associated. The tool also allows the extracted or modified data to be saved separately, apart from having the usual facilities like displaying the results in KeyWord-In-Context (KWIC) format. We also present the notation used for querying and transformation, which is comparable to but different from the Corpus Query Language (CQL).

Grammars play an important role in many Natural Language Processing (NLP) applications. The traditional approach to creating grammars manually, besides being labor-intensive, has several limitations. With the availability of large scale syntactically annotated treebanks, it is now possible to automatically extract an approximate grammar of a language in any of the existing formalisms from a corresponding treebank. In this paper, we present a basic approach to extract grammars from dependency treebanks of two Indian languages, Hindi and Telugu. The process of grammar extraction requires a generalization mechanism. Towards this end, we explore an approach which relies on generalization of argument structure over the verbs based on their syntactic similarity. Such a generalization counters the effect of data sparseness in the treebanks. A grammar extracted using this system can not only expand already existing knowledge bases for NLP tasks such as parsing, but also aid in the creation of grammars for languages where none exist. Further, we show that the grammar extraction process can help in identifying annotation errors and thus aid in the task of the treebank validation.

Developing resources which can be used for Natural Language Processing is an extremely difficult task for any language, but is even more so for less privileged (or less computerized) languages. One way to overcome this difficulty is to adapt the resources of a linguistically close resource rich language. In this paper we discuss how the cost of such adaption can be estimated using subjective and objective measures of linguistic similarity for allocating financial resources, time, manpower etc. Since this is the first work of its kind, the method described in this paper should be seen as only a preliminary method, indicative of how better methods can be developed. Corpora of several less computerized languages had to be collected for the work described in the paper, which was difficult because for many of these varieties there is not much electronic data available. Even if it is, it is in non-standard encodings, which means that we had to build encoding converters for these varieties. The varieties we have focused on are some of the varieties spoken in the South Asian region.