This document contains the instructions for preparing a manuscript for the proceedings of RANLP 2023. The document itself conforms to its own specifications and is therefore an example of what your manuscript should look like. These instructions should be used for both papers submitted for review and for final versions of accepted papers. Authors are asked to conform to all the directions reported in this document.

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.

We present an overview of the second shared task on homophobia/transphobia Detection in social media comments. Given a comment, a system must predict whether or not it contains any form of homophobia/transphobia. The shared task included five languages: English, Spanish, Tamil, Hindi, and Malayalam. The data was given for two tasks. Task A was given three labels, and Task B fine-grained seven labels. In total, 75 teams enrolled for the shared task in Codalab. For task A, 12 teams submitted systems for English, eight teams for Tamil, eight teams for Spanish, and seven teams for Hindi. For task B, nine teams submitted for English, 7 teams for Tamil, 6 teams for Malayalam. We present and analyze all submissions in this paper.

The rise of social media has led to a drastic surge in the dissemination of hostile and toxic content, fostering an alarming proliferation of hate speech, inflammatory remarks, and abusive language. The exponential growth of social media has facilitated the widespread circulation of hostile and toxic content, giving rise to an unprecedented influx of hate speech, incendiary language, and abusive rhetoric. The study utilized different techniques to represent the text data in a numerical format. Word embedding techniques aim to capture the semantic and syntactic information of the text data, which is essential in text classification tasks. The study utilized various techniques such as CNN, BERT, and N-gram to classify social media posts into depression and non-depression categories. Text classification tasks often rely on deep learning techniques such as Convolutional Neural Networks (CNN), while the BERT model, which is pre-trained, has shown exceptional performance in a range of natural language processing tasks. To assess the effectiveness of the suggested approaches, the research employed multiple metrics, including accuracy, precision, recall, and F1-score. The outcomes of the investigation indicate that the suggested techniques can identify symptoms of depression with an average accuracy rate of 56%.

Many people may find motivation in their lives by spreading content on social media that is encouraging or hopeful. Creating an effective model that helps in accurately predicting the target class is a challenging task. The problem of Hope speech identification is dealt with in this work using machine learning and deep learning methods. This paper presents the description of the system submitted by our team(VEL) to the Hope Speech Detection for Equality, Diversity, and Inclusion(HSD-EDI) LT-EDI-RANLP 2023 shared task for the Bulgarian language. The main goal of this shared task is to identify the given text into the Hope speech or Non-Hope speech category. The proposed method used the H2O deep learning model with MPNet embeddings and achieved the second rank for the Bulgarian language with the Macro F1 score of 0.69.