Accented speech classification plays a vital role in the advancement of high-quality automatic speech recognition (ASR) technology. For certain applications, like multi-accented speech classification, it is not always viable to obtain data with accent variation, especially for resource-poor languages. This is one of the major reasons that contributes to the underperformance of the speech classification systems. Therefore, in order to handle speech variability in Indian language speaker accents, we propose a few-shot learning paradigm in this study. It learns generic feature embeddings using an encoder from a pre-trained whisper model and a classification head for classification. The model is refined using LLM’s fine-tuning techniques, such as LoRA and QLoRA, for the six Indian English accents in the Indic Accent Dataset. The experimental findings show that the accuracy of the model is greatly increased by the few-shot learning paradigm’s effectiveness combined with LLM’s fine-tuning techniques. In optimal settings, the model’s accuracy can reach 94% when the trainable parameters are set to 5%.

Identifying fake news hidden as real news is crucial to fight misinformation and ensure reliable information, especially in resource-scarce languages like Malayalam. To recognize the unique challenges of fake news in languages like Malayalam, we present a dataset curated specifically for classifying fake news in Malayalam. This fake news is categorized based on the degree of misinformation, marking the first of its kind in this language. Further, we propose baseline models employing multilingual BERT and diverse machine learning classifiers. Our findings indicate that logistic regression trained on LaBSE features demonstrates promising initial performance with an F1 score of 0.3393. However, addressing the significant data imbalance remains essential for further improvement in model accuracy.

This paper examines the submissions of various participating teams to the task on Hate and Offensive Language Detection in Telugu Codemixed Text (HOLD-Telugu) organized as part of DravidianLangTech 2024. In order to identify the contents containing harmful information in Telugu codemixed social media text, the shared task pushes researchers and academicians to build models. The dataset for the task was created by gathering YouTube comments and annotated manually. A total of 23 teams participated and submitted their results to the shared task. The rank list was created by assessing the submitted results using the macro F1-score.

This paper presents the findings of the shared task on multimodal sentiment analysis, abusive language detection and hate speech detection in Dravidian languages. Through this shared task, researchers worldwide can submit models for three crucial social media data analysis challenges in Dravidian languages: sentiment analysis, abusive language detection, and hate speech detection. The aim is to build models for deriving fine-grained sentiment analysis from multimodal data in Tamil and Malayalam, identifying abusive and hate content from multimodal data in Tamil. Three modalities make up the multimodal data: text, audio, and video. YouTube videos were gathered to create the datasets for the tasks. Thirty-nine teams took part in the competition. However, only two teams, though, turned in their findings. The macro F1-score was used to assess the submissions

The rise of online social media has revolutionized communication, offering users a convenient way to share information and stay updated on current events. However, this surge in connectivity has also led to the proliferation of misinformation, commonly known as fake news. This misleading content, often disguised as legitimate news, poses a significant challenge as it can distort public perception and erode trust in reliable sources. This shared task consists of two subtasks such as task 1 and task 2. Task 1 aims to classify a given social media text into original or fake. The goal of the FakeDetect-Malayalam task2 is to encourage participants to develop effective models capable of accurately detecting and classifying fake news articles in the Malayalam language into different categories like False, Half True, Mostly False, Partly False, and Mostly True. For this shared task, 33 participants submitted their results.

Speech recognition is known to be a specialized application of speech processing. Automatic speech recognition (ASR) systems are designed to perform the speech-to-text task. Although ASR systems have been the subject of extensive research, they still encounter certain challenges when speech variations arise. The speaker’s age, gender, vulnerability, and other factors are the main causes of the variations in speech. In this work, we propose a fine-tuned speech recognition model for recognising the spoken words of vulnerable individuals in Tamil. This research utilizes a dataset sourced from the LT-EDI@EACL2024 shared task. We trained and tested pre-trained ASR models, including XLS-R and Whisper. The findings highlight that the fine-tuned Whisper ASR model surpasses the XLSR, achieving a word error rate (WER) of 24.452, signifying its superior performance in recognizing speech from diverse individuals.

This paper focuses on identifying hate and offensive keywords from codemix Malayalam social media text. As part of this work, a dataset for hate and offensive keyword extraction for codemix Malayalam language was created. Two different methods were experimented to extract Hate and Offensive language (HOL) keywords from social media text. In the first method, intrinsic evaluation was performed on the dataset to identify the hate and offensive keywords. Three different approaches namely – unigram approach, bigram approach and trigram approach were performed to extract the HOL keywords, sequence of HOL words and the sequence that contribute HOL meaning even in the absence of a HOL word. Five different transformer models were used in each of the pproaches for extracting the embeddings for the ngrams. Later, HOL keywords were extracted based on the similarity score obtained using the cosine similarity. Out of the five transformer models, the best results were obtained with multilingual BERT. In the second method, multilingual BERT transformer model was fine tuned with the dataset to develop a HOL keyword tagger model. This work is a new beginning for HOL keyword identification in Dravidian language – Malayalam.

This paper summarizes the shared task on multimodal abusive language detection and sentiment analysis in Dravidian languages as part of the third Workshop on Speech and Language Technologies for Dravidian Languages at RANLP 2023. This shared task provides a platform for researchers worldwide to submit their models on two crucial social media data analysis problems in Dravidian languages - abusive language detection and sentiment analysis. Abusive language detection identifies social media content with abusive information, whereas sentiment analysis refers to the problem of determining the sentiments expressed in a text. This task aims to build models for detecting abusive content and analyzing fine-grained sentiment from multimodal data in Tamil and Malayalam. The multimodal data consists of three modalities - video, audio and text. The datasets for both tasks were prepared by collecting videos from YouTube. Sixty teams participated in both tasks. However, only two teams submitted their results. The submissions were evaluated using macro F1-score.

This paper discusses the submissions to the shared task on abusive comment detection in Tamil and Telugu codemixed social media text conducted as part of the third Workshop on Speech and Language Technologies for Dravidian Languages at RANLP 20239. The task encourages researchers to develop models to detect the contents containing abusive information in Tamil and Telugu codemixed social media text. The task has three subtasks - abusive comment detection in Tamil, Tamil-English and Telugu-English. The dataset for all the tasks was developed by collecting comments from YouTube. The submitted models were evaluated using macro F1-score, and prepared the rank list accordingly.

Part-of-speech (POS) tagging is a fundamental task in natural language processing (NLP) that involves assigning grammatical categories to words in a sentence. In this study, we investigate the application of deep sequential models for POS tagging of Telugu, a low-resource Dravidian language with rich morphology. We use the Universal dependencies dataset for this research and explore various deep learning architectures, including Recurrent Neural Networks (RNNs), Long Short-Term Memory (LSTM) networks, Gated Recurrent Units (GRUs), and their stacked variants for POS tagging. Additionally, we utilize multilingual BERT embeddings and indicBERT embeddings to capture contextual information from the input sequences. Our experiments demonstrate that stacked LSTM with multilingual BERT embeddings achieves the highest performance, outperforming other approaches and attaining an F1 score of 0.8812. These findings suggest that deep sequential models, particularly stacked LSTMs with multilingual BERT embeddings, are effective tools for POS tagging in Telugu.

Dependency parsing is a method for doing surface-level syntactic analysis on natural language texts. The scarcity of any viable tools for doing these tasks in Dravidian Languages has introduced a new line of research into these topics. This paper focuses on a novel approach that uses word-to-word dependency tagging using BERT models to improve the malt parser performance. We used Tamil, a morphologically rich and free word language. The individual words are tokenized using BERT models and the dependency relations are recognized using Machine Learning Algorithms. Oversampling algorithms such as SMOTE (Chawla et al., 2002) and ADASYN (He et al., 2008) are used to tackle data imbalance and consequently improve parsing results. The results obtained are used in the malt parser and this can be accustomed to further highlight that feature-based approaches can be used for such tasks.

This paper describes the approach of team CEN-Tamil used for abusive comment detection in Tamil. This task aims to identify whether a given comment contains abusive comments. We used TF-IDF with char-wb analyzers with Random Kitchen Sink (RKS) algorithm to create feature vectors and the Support Vector Machine (SVM) classifier with polynomial kernel for classification. We used this method for both Tamil and Tamil-English datasets and secured first place with an f1-score of 0.32 and seventh place with an f1-score of 0.25, respectively. The code for our approach is shared in the GitHub repository.

This paper presents the findings of the shared task on Multimodal Sentiment Analysis and Troll meme classification in Dravidian languages held at ACL 2022. Multimodal sentiment analysis deals with the identification of sentiment from video. In addition to video data, the task requires the analysis of corresponding text and audio features for the classification of movie reviews into five classes. We created a dataset for this task in Malayalam and Tamil. The Troll meme classification task aims to classify multimodal Troll memes into two categories. This task assumes the analysis of both text and image features for making better predictions. The performance of the participating teams was analysed using the F1-score. Only one team submitted their results in the Multimodal Sentiment Analysis task, whereas we received six submissions in the Troll meme classification task. The only team that participated in the Multimodal Sentiment Analysis shared task obtained an F1-score of 0.24. In the Troll meme classification task, the winning team achieved an F1-score of 0.596.

This paper narrates the work of the team Amrita_CEN for the shared task on Patronizing and Condescending Language Detection at SemEval 2022. We implemented machine learning algorithms such as Support Vector Machine (SVV), Logistic regression, Naive Bayes, XG Boost and Random Forest for modelling the tasks. At the same time, we also applied a feature engineering method to solve the class imbalance problem with respect to training data. Among all the models, the logistic regression model outperformed all other models and we have submitted results based upon the same.

This paper describes the submission of the team Amrita_CEN to the shared task on iSarcasm Eval: Intended Sarcasm Detection in English and Arabic at SemEval 2022. We employed machine learning algorithms towards sarcasm detection. Here, we used K-Nearest Neighbor (KNN), Support Vector Machine (SVM), Naïve Bayes, Logistic Regression, and Decision Tree along with the Random Forest ensemble method. Additionally, feature engineering techniques were applied to deal with the problems of class imbalance during training. Among the models considered, our study shows that the SVM, logistic regression and ensemble model Random Forest exhibited the best performance, which was submitted to the shared task.

The purpose and influence of a citation are important in understanding the quality of a publication. The 3c citation context classification shared task at the Second Workshop on Scholarly Document Processing aims at addressing this problem. This paper is the submission of the team Amrita_CEN_NLP to the shared task. We employed Bi-directional Long Short Term Memory (LSTM) networks and a Random Forest classifier for modelling the aforementioned problems by considering the class imbalance problem in the data.

This paper describes the submission of the team Amrita_CEN_NLP to the shared task on Offensive Language Identification in Dravidian Languages at EACL 2021. We implemented three deep neural network architectures such as a hybrid network with a Convolutional layer, a Bidirectional Long Short-Term Memory network (Bi-LSTM) layer and a hidden layer, a network containing a Bi-LSTM and another with a Bidirectional Recurrent Neural Network (Bi-RNN). In addition to that, we incorporated a cost-sensitive learning approach to deal with the problem of class imbalance in the training data. Among the three models, the hybrid network exhibited better training performance, and we submitted the predictions based on the same.

Natural Language Processing (NLP), is the field of artificial intelligence that gives the computer the ability to interpret, perceive and extract appropriate information from human languages. Contemporary NLP is predominantly a data driven process. It employs machine learning and statistical algorithms to learn language structures from textual corpus. While application of NLP in English, certain European languages such as Spanish, German, etc. and Chinese, Arabic has been tremendous, it is not so, in many Indian languages. There are obvious advantages in creating aligned bilingual and multilingual corpora. Machine translation, cross-lingual information retrieval, content availability and linguistic comparison are a few of the most sought after applications of such parallel corpora. This paper explains and validates a parallel corpus we created for English-Tamil bilingual pair.

Identification of the purpose and influence of citation is significant in assessing the impact of a publication. ‘3C’ Citation Context Classification Task in Workshop on Mining Scientific Publication is a shared task to address the abovementioned problems. This working note describes the submissions of Amrita_CEN_NLP team to the shared task. We used Random Forest with cost-sensitive learning for classification of sentences encoded into a vector of dimension 300.

This working note presents the methodology used in deepCybErNet submission to the shared task on Emotion Intensities in Tweets (EmoInt) WASSA-2017. The goal of the task is to predict a real valued score in the range [0-1] for a particular tweet with an emotion type. To do this, we used Bag-of-Words and embedding based on recurrent network architecture. We have developed two systems and experiments are conducted on the Emotion Intensity shared Task 1 data base at WASSA-2017. A system which uses word embedding based on recurrent network architecture has achieved highest 5 fold cross-validation accuracy. This has used embedding with recurrent network to extract optimal features at tweet level and logistic regression for prediction. These methods are highly language independent and experimental results shows that the proposed methods are apt for predicting a real valued score in than range [0-1] for a given tweet with its emotion type.