Evaluating the reliability of news sources is a routine task for journalists and organizations committed to acquiring and disseminating accurate information.Recent research has shown that predicting sources’ reliability represents an important first-prior step in addressing additional challenges such as fake news detection and fact-checking.In this paper, we introduce a novel approach for source reliability estimation that leverages reinforcement learning strategies for estimating the reliability degree of news sources. Contrary to previous research, our proposed approach models the problem as the estimation of a reliability degree, and not a reliability label, based on how all the news media sources interact with each other on the Web.We validated the effectiveness of our method on a news media reliability dataset that is an order of magnitude larger than comparable existing datasets. Results show that the estimated reliability degrees strongly correlates with journalists-provided scores (Spearman=0.80) and can effectively predict reliability labels (macro-avg. F1 score=81.05).We release our implementation and dataset, aiming to provide a valuable resource for the NLP community working on information verification.

Automatic depression detection from conversational data has gained significant interest in recent years.The DAIC-WOZ dataset, interviews conducted by a human-controlled virtual agent, has been widely used for this task.Recent studies have reported enhanced performance when incorporating interviewer’s prompts into the model.In this work, we hypothesize that this improvement might be mainly due to a bias present in these prompts, rather than the proposed architectures and methods.Through ablation experiments and qualitative analysis, we discover that models using interviewer’s prompts learn to focus on a specific region of the interviews, where questions about past experiences with mental health issues are asked, and use them as discriminative shortcuts to detect depressed participants. In contrast, models using participant responses gather evidence from across the entire interview.Finally, to highlight the magnitude of this bias, we achieve a 0.90 F1 score by intentionally exploiting it, the highest result reported to date on this dataset using only textual information.Our findings underline the need for caution when incorporating interviewers’ prompts into models, as they may inadvertently learn to exploit targeted prompts, rather than learning to characterize the language and behavior that are genuinely indicative of the patient’s mental health condition.

We present Claim-Dissector: a novel latent variable model for fact-checking and analysis, which given a claim and a set of retrieved evidence jointly learns to identify: (i) the relevant evidences to the given claim (ii) the veracity of the claim. We propose to disentangle the per-evidence relevance probability and its contribution to the final veracity probability in an interpretable way — the final veracity probability is proportional to a linear ensemble of per-evidence relevance probabilities. In this way, the individual contributions of evidences towards the final predicted probability can be identified. In per-evidence relevance probability, our model can further distinguish whether each relevant evidence is supporting (S) or refuting (R) the claim. This allows to quantify how much the S/R probability contributes to final verdict or to detect disagreeing evidence. Despite its interpretable nature, our system achieves results competetive with state-of-the-art on the FEVER dataset, as compared to typical two-stage system pipelines, while using significantly fewer parameters. Furthermore, our analysis shows that our model can learn fine-grained relevance cues while using coarse-grained supervision and we demonstrate it in 2 ways. (i) We show that our model can achieve competitive sentence recall while using only paragraph-level relevance supervision. (ii) Traversing towards the finest granularity of relevance, we show that our model is capable of identifying relevance at the token level. To do this, we present a new benchmark TLR-FEVER focusing on token-level interpretability — humans annotate tokens in relevant evidences they considered essential when making their judgment. Then we measure how similar are these annotations to the tokens our model is focusing on.

With the increase of users on social media platforms, manipulating or provoking masses of people has become a piece of cake. This spread of hatred among people, which has become a loophole for freedom of speech, must be minimized. Hence, it is essential to have a system that automatically classifies the hatred content, especially on social media, to take it down. This paper presents a simple modular pipeline classifier with BERT embeddings and attention mechanism to classify hope speech content in the Hope Speech Detection shared task for Equality, Diversity, and Inclusion-ACL 2022. Our system submission ranks fourth with an F1-score of 0.84. We release our code-base here https://github.com/Deepanshu-beep/hope-speech-attention .

Social media platforms have been provoking masses of people. The individual comments affect a prevalent way of thinking by moving away from preoccupation with discrimination, loneliness, or influence in building confidence, support, and good qualities. This paper aims to identify hope in these social media posts. Hope significantly impacts the well-being of people, as suggested by health professionals. It reflects the belief to achieve an objective, discovers a new path, or become motivated to formulate pathways. In this paper we classify given a social media post, hope speech or not hope speech, using ensembled voting of BERT, ERNIE 2.0 and RoBERTa for English language with 0.54 macro F1-score (2^st rank). For non-English languages Malayalam, Spanish and Tamil we utilized XLM RoBERTA with 0.50, 0.81, 0.3 macro F1 score (1^st, 1^st,3^rd rank) respectively. For Kannada, we use Multilingual BERT with 0.32 F1 score(5^th)position. We release our code-base here: https://github.com/Muskaan-Singh/Hate-Speech-detection.git.

The increased expansion of abusive content on social media platforms negatively affects online users. Transphobic/homophobic content indicates hatred comments for lesbian, gay, transgender, or bisexual people. It leads to offensive speech and causes severe social problems that can make online platforms toxic and unpleasant to LGBT+people, endeavoring to eliminate equality, diversity, and inclusion. In this paper, we present our classification system; given comments, it predicts whether or not it contains any form of homophobia/transphobia with a Zero-Shot learning framework. Our system submission achieved 0.40, 0.85, 0.89 F1-score for Tamil and Tamil-English, English with (1^st, 1^st,8^th) ranks respectively. We release our codebase here: https://github.com/Muskaan-Singh/Homophobia-and-Transphobia-ACL-Submission.git.

Depression is a common illness involving sadness and lack of interest in all day-to-day activities. It is important to detect depression at an early stage as it is treated at an early stage to avoid consequences. In this paper, we present our system submission of ARGUABLY for DepSign-LT-EDI@ACL-2022. We aim to detect the signs of depression of a person from their social media postings wherein people share their feelings and emotions. The proposed system is an ensembled voting model with fine-tuned BERT, RoBERTa, and XLNet. Given social media postings in English, the submitted system classify the signs of depression into three labels, namely “not depressed,” “moderately depressed,” and “severely depressed.” Our best model is ranked 3^rd position with 0.54% accuracy . We make our codebase accessible here.

In this paper, we describe our participation in the subtask 1 of CASE-2022, Event Causality Identification with Casual News Corpus. We address the Causal Relation Identification (CRI) task by exploiting a set of simple yet complementary techniques for fine-tuning language models (LMs) on a few annotated examples (i.e., a few-shot configuration).We follow a prompt-based prediction approach for fine-tuning LMs in which the CRI task is treated as a masked language modeling problem (MLM). This approach allows LMs natively pre-trained on MLM tasks to directly generate textual responses to CRI-specific prompts. We compare the performance of this method against ensemble techniques trained on the entire dataset. Our best-performing submission was fine-tuned with only 256 instances per class, 15.7% of the all available data, and yet obtained the second-best precision (0.82), third-best accuracy (0.82), and an F1-score (0.85) very close to what was reported by the winner team (0.86).

In this paper, we describe our shared task submissions for Subtask 2 in CASE-2022, Event Causality Identification with Casual News Corpus. The challenge focused on the automatic detection of all cause-effect-signal spans present in the sentence from news-media. We detect cause-effect-signal spans in a sentence using T5 — a pre-trained autoregressive language model. We iteratively identify all cause-effect-signal span triplets, always conditioning the prediction of the next triplet on the previously predicted ones. To predict the triplet itself, we consider different causal relationships such as cause→effect→signal. Each triplet component is generated via a language model conditioned on the sentence, the previous parts of the current triplet, and previously predicted triplets. Despite training on an extremely small dataset of 160 samples, our approach achieved competitive performance, being placed second in the competition. Furthermore, we show that assuming either cause→effect or effect→cause order achieves similar results.

In this paper the authors detail the various legal and ethical issues faced during the ATCO2 project. This project is aimed at developing tools to automatically collect and transcribe air traffic conversations, especially conversations between pilots and air controls towers. In this paper the authors will develop issues related to intellectual property, public data, privacy, and general ethics issues related to the collection of air-traffic control speech.

This work represents the system proposed by team Innovators for SemEval 2022 Task 8: Multilingual News Article Similarity. Similar multilingual news articles should match irrespective of the style of writing, the language of conveyance, and subjective decisions and biases induced by medium/outlet. The proposed architecture includes a machine translation system that translates multilingual news articles into English and presents a multitask learning model trained simultaneously on three distinct datasets. The system leverages the PageRank algorithm for Long-form text alignment. Multitask learning approach allows simultaneous training of multiple tasks while sharing the same encoder during training, facilitating knowledge transfer between tasks. Our best model is ranked 16 with a Pearson score of 0.733.

This paper documents our approach for the Creative-Summ 2022 shared task for Automatic Summarization of Creative Writing. For this purpose, we develop an automatic summarization pipeline where we leverage a denoising autoencoder for pretraining sequence-to-sequence models and fine-tune it on a large-scale abstractive screenplay summarization dataset to summarize TV transcripts from primetime shows. Our pipeline divides the input transcript into smaller conversational blocks, removes redundant text, summarises the conversational blocks, obtains the block-wise summaries, cleans, structures, and then integrates the summaries to create the meeting minutes. Our proposed system achieves some of the best scores across multiple metrics(lexical, semantical) in the Creative-Summ shared task.

This paper presents our submission for the shared task on isometric neural machine translation in International Conference on Spoken Language Translation (IWSLT). There are numerous state-of-art models for translation problems. However, these models lack any length constraint to produce short or long outputs from the source text. In this paper, we propose a hierarchical approach to generate isometric translation on MUST-C dataset, we achieve a BERTscore of 0.85, a length ratio of 1.087, a BLEU score of 42.3, and a length range of 51.03%. On the blind dataset provided by the task organizers, we obtain a BERTscore of 0.80, a length ratio of 1.10 and a length range of 47.5%. We have made our code public here https://github.com/aakash0017/Machine-Translation-ISWLT

This paper describes the team (“Tamalli”)’s submission to AmericasNLP2021 shared task on Open Machine Translation for low resource South American languages. Our goal was to evaluate different Machine Translation (MT) techniques, statistical and neural-based, under several configuration settings. We obtained the second-best results for the language pairs “Spanish-Bribri”, “Spanish-Asháninka”, and “Spanish-Rarámuri” in the category “Development set not used for training”. Our performed experiments will serve as a point of reference for researchers working on MT with low-resource languages.

This paper provides the description of shared tasks to the WAT 2021 by our team “NLPHut”. We have participated in the English→Hindi Multimodal translation task, English→Malayalam Multimodal translation task, and Indic Multi-lingual translation task. We have used the state-of-the-art Transformer model with language tags in different settings for the translation task and proposed a novel “region-specific” caption generation approach using a combination of image CNN and LSTM for the Hindi and Malayalam image captioning. Our submission tops in English→Malayalam Multimodal translation task (text-only translation, and Malayalam caption), and ranks second-best in English→Hindi Multimodal translation task (text-only translation, and Hindi caption). Our submissions have also performed well in the Indic Multilingual translation tasks.

Multimodal Machine Translation (MMT) systems utilize additional information from other modalities beyond text to improve the quality of machine translation (MT). The additional modality is typically in the form of images. Despite proven advantages, it is indeed difficult to develop an MMT system for various languages primarily due to the lack of a suitable multimodal dataset. In this work, we develop an MMT for English-> Bengali using a recently published Bengali Visual Genome (BVG) dataset that contains images with associated bilingual textual descriptions. Through a comparative study of the developed MMT system vis-a-vis a Text-to-text translation, we demonstrate that the use of multimodal data not only improves the translation performance improvement in BLEU score of +1.3 on the development set, +3.9 on the evaluation test, and +0.9 on the challenge test set but also helps to resolve ambiguities in the pure text description. As per best of our knowledge, our English-Bengali MMT system is the first attempt in this direction, and thus, can act as a baseline for the subsequent research in MMT for low resource languages.

The preparation of parallel corpora is a challenging task, particularly for languages that suffer from under-representation in the digital world. In a multi-lingual country like India, the need for such parallel corpora is stringent for several low-resource languages. In this work, we provide an extended English-Odia parallel corpus, OdiEnCorp 2.0, aiming particularly at Neural Machine Translation (NMT) systems which will help translate English↔Odia. OdiEnCorp 2.0 includes existing English-Odia corpora and we extended the collection by several other methods of data acquisition: parallel data scraping from many websites, including Odia Wikipedia, but also optical character recognition (OCR) to extract parallel data from scanned images. Our OCR-based data extraction approach for building a parallel corpus is suitable for other low resource languages that lack in online content. The resulting OdiEnCorp 2.0 contains 98,302 sentences and 1.69 million English and 1.47 million Odia tokens. To the best of our knowledge, OdiEnCorp 2.0 is the largest Odia-English parallel corpus covering different domains and available freely for non-commercial and research purposes.

Identifying the author of a given text can be useful in historical literature, plagiarism detection, or police investigations. Authorship Attribution (AA) has been well studied and mostly relies on a large feature engineering work. More recently, deep learning-based approaches have been explored for Authorship Attribution (AA). In this paper, we introduce BertAA, a fine-tuning of a pre-trained BERT language model with an additional dense layer and a softmax activation to perform authorship classification. This approach reaches competitive performances on Enron Email, Blog Authorship, and IMDb (and IMDb62) datasets, up to 5.3% (relative) above current state-of-the-art approaches. We performed an exhaustive analysis allowing to identify the strengths and weaknesses of the proposed method. In addition, we evaluate the impact of including additional features (e.g. stylometric and hybrid features) in an ensemble approach, improving the macro-averaged F1-Score by 2.7% (relative) on average.

Language detection is considered a difficult task especially for similar languages, varieties, and dialects. With the growing number of online content in different languages, the need for reliable and robust language detection tools also increased. In this work, we use supervised autoencoders with a bayesian optimizer for language detection and highlights its efficiency in detecting similar languages with dialect variance in comparison to other state-of-the-art techniques. We evaluated our approach on multiple datasets (Ling10, Discriminating between Similar Language (DSL), and Indo-Aryan Language Identification (ILI)). Obtained results demonstrate that SAE are higly effective in detecting languages, up to a 100% accuracy in the Ling10. Similarly, we obtain a competitive performance in identifying similar languages, and dialects, 92% and 85% for DSL ans ILI datasets respectively.

This paper describes the ODIANLP submission to WAT 2020. We have participated in the English-Hindi Multimodal task and Indic task. We have used the state-of-the-art Transformer model for the translation task and InceptionResNetV2 for the Hindi Image Captioning task. Our submission tops in English->Hindi Multimodal task in its track and Odia<->English translation tasks. Also, our submissions performed well in the Indic Multilingual tasks.

Text summarization is considered as a challenging task in the NLP community. The availability of datasets for the task of multilingual text summarization is rare, and such datasets are difficult to construct. In this work, we build an abstract text summarizer for the German language text using the state-of-the-art “Transformer” model. We propose an iterative data augmentation approach which uses synthetic data along with the real summarization data for the German language. To generate synthetic data, the Common Crawl (German) dataset is exploited, which covers different domains. The synthetic data is effective for the low resource condition and is particularly helpful for our multilingual scenario where availability of summarizing data is still a challenging issue. The data are also useful in deep learning scenarios where the neural models require a large amount of training data for utilization of its capacity. The obtained summarization performance is measured in terms of ROUGE and BLEU score. We achieve an absolute improvement of +1.5 and +16.0 in ROUGE1 F1 (R1_F1) on the development and test sets, respectively, compared to the system which does not rely on data augmentation.

This paper describes the Idiap submission to WAT 2019 for the English-Hindi Multi-Modal Translation Task. We have used the state-of-the-art Transformer model and utilized the IITB English-Hindi parallel corpus as an additional data source. Among the different tracks of the multi-modal task, we have participated in the “Text-Only” track for the evaluation and challenge test sets. Our submission tops in its track among the competitors in terms of both automatic and manual evaluation. Based on automatic scores, our text-only submission also outperforms systems that consider visual information in the “multi-modal translation” task.

The multi-level adaptive networks (MLAN) technique is a cross-lingual adaptation framework where a bottleneck (BN) layer in a deep neural network (DNN) trained in a source language is used for producing BN features to be exploited in a second DNN in a target language. We investigate how the correlation (in the sense of phonetic similarity) of the source and target languages and the amount of data of the source language affect the efficiency of the MLAN schemes. We experiment with three different scenarios using, i) French, as a source language uncorrelated to the target language, ii) Ukrainian, as a source language correlated to the target one and finally iii) English as a source language uncorrelated to the target language using a relatively large amount of data in respect to the other two scenarios. In all cases Russian is used as target language. GLOBALPHONE data is used, except for English, where a mixture of LIBRISPEECH, TEDLIUM and AMIDA is available. The results have shown that both of these two factors are important for the MLAN schemes. Specifically, on the one hand, when a modest amount of data from the source language is used, the correlation of the source and target languages is very important. On the other hand, the correlation of the two languages seems to be less important when a relatively large amount of data, from the source language, is used. The best performance in word error rate (WER), was achieved when the English language was used as the source one in the multi-task MLAN scheme, achieving a relative improvement of 9.4% in respect to the baseline DNN model.

This paper describes the data collection and annotation carried out within the DBOX project ( Eureka project, number E! 7152). This project aims to develop interactive games based on spoken natural language human-computer dialogues, in 3 European languages: English, German and French. We collect the DBOX data continuously. We first start with human-human Wizard of Oz experiments to collect human-human data in order to model natural human dialogue behaviour, for better understanding of phenomena of human interactions and predicting interlocutors actions, and then replace the human Wizard by an increasingly advanced dialogue system, using evaluation data for system improvement. The designed dialogue system relies on a Question-Answering (QA) approach, but showing truly interactive gaming behaviour, e.g., by providing feedback, managing turns and contact, producing social signals and acts, e.g., encouraging vs. downplaying, polite vs. rude, positive vs. negative attitude towards players or their actions, etc. The DBOX dialogue corpus has required substantial investment. We expect it to have a great impact on the rest of the project. The DBOX project consortium will continue to maintain the corpus and to take an interest in its growth, e.g., expand to other languages. The resulting corpus will be publicly released.