Ireland

Online social networks are ubiquitous and user-friendly. Nevertheless, it is vital to detect and moderate offensive content to maintain decency and empathy. However, mining social media texts is a complex task since users don&#39;t adhere to any fixed patterns. Comments can be written in any combination of languages and many of them may be low-resource.

In this paper, we present our system for the LT-EDI shared task on detecting homophobia and transphobia in social media comments. We experiment with a number of monolingual and multilingual transformer based models such as mBERT along with a data augmentation technique for tackling class imbalance. Such pretrained large models have recently shown tremendous success on a variety of benchmark tasks in natural language processing. We observe their performance on a carefully annotated, real life dataset of YouTube comments in English as well as Tamil.

Our submission achieved ranks 9, 6 and 3 with a macro-averaged F1-score of 0.42, 0.64 and 0.58 in the English, Tamil and Tamil-English subtasks respectively. The code for the system has been open sourced.

ACL 2022

Leveraging Pretrained Language Models for Detecting Homophobia and Transphobia in Social Media Comments

sentiment analysis

computational social science

hate speech detection

# Welcome everyone to ACL 2022!

The 60th Annual Meeting of the Association for Computational Linguistics is taking place May 22-27, 2022 as a hybrid event, in Dublin and online. We are happy to welcome all of you to this anniversary edition with an almost 50-50 in-person and virtual participation. 
The main conference program features oral presentations, in-person and virtual posters and demo sessions, a plenary session for our best paper presentations and awards, three amazing keynote events and two new initiatives of invited talks: Spotlight Talks for Young Rising Stars (STIRS) and The Next Big Idea Talks. Posters (including Findings of ACL 2022) and demos are grouped by areas for both the in-person and the virtual sessions. For the virtual component, the talks will be on Zoom and the posters and the demos will be in GatherTown. The Student Research Workshop will have an oral session and a poster session as part of Poster Session 1. The program also features eight Tutorials and 28 Workshops. 

 
We wish you a wonderful conference! 
[**The ACL 2022 Organizing Committee**](https://www.2022.aclweb.org/organisers)
 
[**Conference Handbook**](https://drive.google.com/file/d/1_BUCMfhMVrjG9E2e71aHdHeE28KSje0l/view?usp=sharing) 
[**Mini Handbook**](https://drive.google.com/file/d/1qlBKl0wzmlVF1oCeMQl3BahLd9nLP5Ce/view?usp=sharing) 
[**Posters and Demo guides**](https://drive.google.com/file/d/1UucMAoCNncIOaH1rMMDa0owuG9qgvJTG/view?usp=sharing)

The Association for Computational Linguistics (ACL) is the premier international scientific and professional society for people working on computational problems involving human language, a field often referred to as either computational linguistics or natural language processing (NLP). 

Online social networks are ubiquitous and user-friendly. Nevertheless, it is vital to detect and moderate offensive content to maintain decency and empathy. However, mining social media texts is a complex task since users don't adhere to any fixed patterns. Comments can be written in any combination of languages and many of them may be low-resource.

In this paper, we present our system for the LT-EDI shared task on detecting homophobia and transphobia in social media comments. We experiment with a number of monolingual and multilingual transformer based models such as mBERT along with a data augmentation technique for tackling class imbalance. Such pretrained large models have recently shown tremendous success on a variety of benchmark tasks in natural language processing. We observe their performance on a carefully annotated, real life dataset of YouTube comments in English as well as Tamil.

Our submission achieved ranks 9, 6 and 3 with a macro-averaged F1-score of 0.42, 0.64 and 0.58 in the English, Tamil and Tamil-English subtasks respectively. The code for the system has been open sourced.

workshop paper

This paper describes our system that participated in LT-EDI-ACL2022- Homophobia/Transphobia Detection in Social Media. Sexual minorities face a lot of unfair treatment and discrimination in our world. This creates enormous stress and many psychological problems for sexual minorities. There is a lot of hate speech on the internet, and homophobia/transphobia is one against sexual minorities. Identifying and processing homophobia/ transphobia through natural language processing technology can improve the efficiency of processing it, and can quickly screen out it on the Internet. The organizer of LT-EDI-ACL2022- Homophobia/Transphobia Detection in Social Media constructs a homophobia/ transphobia detection dataset based on YouTube comments for English and Tamil. We use a Roberta -based approach to conduct homophobia/transphobia detection experiments on the dataset of the competition, and get better results.

A Roberta based Approach for Homophobia/Transphobia Detection in Social Media

Spreading positive vibes or hope content on social media may help many people to get motivated in their life. To address Hope Speech detection in YouTube comments, this paper presents the description of the models submitted by our team - MUCIC, to the Hope Speech Detection for Equality, Diversity, and Inclusion (HopeEDI) shared task at Association for Computational Linguistics (ACL) 2022. This shared task consists of texts in five languages, namely: English, Spanish (in Latin scripts), and Tamil, Malayalam, and Kannada (in code-mixed native and Roman scripts) with the aim of classifying the YouTube comment into "Hope", "Not-Hope" or "Not-Intended" categories. The proposed methodology uses the re-sampling technique to deal with imbalanced data in the corpus and obtained 1st rank for English language with a macro-averaged F1-score of 0.550 and weighted-averaged F1-score of 0.860. The code to reproduce this work is available in https://github.com/anushamdgowda/Hope-speech.

Hope Speech Detection using Data Re-Sampling and 1D Conv-LSTM

In recent years social media has become one of the major forums for expressing human views and emotions. With the help of smartphones and high-speed internet, anyone can express their views on Social media. However, this can also lead to the spread of hatred and violence in society. Therefore it is necessary to build a method to find and support helpful social media content. In this paper, we studied Natural Language Processing approach for detecting Hope speech in a given sentence. The task was to classify the sentences into ‘Hope speech’ and ‘Non-hope speech’. The dataset was provided by LT-EDI organizers with text from Youtube comments. Based on the task description, we developed a system using the pre-trained language model BERT to complete this task. Our model achieved 1st rank in the Kannada language with a weighted average F1 score of 0.750, 2nd rank in the Malayalam language with a weighted average F1 score of 0.740, 3rd rank in the Tamil language with a weighted average F1 score of 0.390 and 6th rank in the English language with a weighted average F1 score of 0.880.

Hope Speech Detection for Equality, Diversity, and Inclusion Using ALBERT model

An Automatic Speech Recognition System is developed for addressing the Tamil conversational speech data of the elderly people and transgender. The speech corpus used in this system is collected from the people who adhere their communication in Tamil at some primary places like bank, hospital, vegetable markets. Our ASR system is designed with pre-trained model which is used to recognize the speech data. WER(Word Error Rate) calculation is used to analyse the performance of the ASR system. This evaluation could help to make a comparison of utterances between the elderly people and others. Similarly, the comparison between the transgender and other people is also done. Our proposed ASR system achieves the word error rate as 39.65

Transformer based Approach for Speech Recognition for Vulnerable Individuals in Tamil

In this paper, we report the solution of the team BERT 4EVER for the LT-EDI-2022 shared task2: Homophobia/Transphobia Detection in social media comments in ACL 2022, which aims to classify Youtube comments into one of the following categories: no, moderate, or severe depression. We model the problem as a text classification task and a text generation task and respectively propose two different models for the tasks. To combine the knowledge learned from these two different models, we softly fuse the predicted probabilities of the models above and then select the label with the highest probability as the final output. In addition, multiple augmentation strategies are leveraged to improve the model generalization capability, such as back translation and adversarial training. Experimental results demonstrate the effectiveness of the proposed models and two augmented strategies.

Detecting signs of Depression from Social Media：Detecting Depression in Social Media using Prompt-Learning and Word-Emotion Cluster

Hope is an inherent part of human life and essential for improving the quality of life. Hope increases happiness and reduces stress and feelings of helplessness. Hope speech is the desired outcome for better and can be studied using text from various online sources where people express their desires and outcomes. In this paper, we address a deep-learning approach with a combination of linguistic and psycho-linguistic features for hope-speech detection. We report our best results submitted to LT-EDI-2022 which ranked 2nd and 3rd in English and Spanish respectively.

Are transformers the only hope? Hope speech detection for Spanish and English comments

The DepSign-LT-EDI-ACL2022 shared task focuses on early prediction of severity of depression over social media posts. The BioNLP group at Department of Data Science and Engineering in Indian Institute of Science Education and Research Bhopal (IISERB) has participated in this challenge and submitted three runs based on three different text mining models. The severity of depression were categorized into three classes, viz., no depression, moderate, and severe and the data to build models were released as part of this shared task. The objective of this work is to identify relevant features from the given social media texts for effective text classification. As part of our investigation, we explored features derived from text data using document embeddings technique and simple bag of words model following different weighting schemes. Subsequently, adaptive boosting, logistic regression, random forest and support vector machine (SVM) classifiers were used to identify the scale of depression from the given texts. The experimental analysis on the given validation data show that the SVM classifier using the bag of words model following term frequency and inverse document frequency weighting scheme outperforms the other models for identifying depression. However, this framework could not achieve a place among the top ten runs of the shared task. Here we will describe the potential of the proposed framework as well as the possible reasons behind mediocre performance on the given data.

A Bag of Words and Document Embeddings Based Framework to Identify Severity of Depression Over Social Media

Over the years, there has been a slow but steady change in the attitude of society towards different kinds of sexuality. However, on social media platforms, where people have the license to be anonymous, toxic comments targeted at homosexuals, transgenders and the LGBTQ+ community are not uncommon. Detection of homophobic comments on social media can be useful in making the internet a safer place for everyone. For this task, we used a combination of word embeddings and SVM Classifiers as well as some BERT-based transformers. We achieved a weighted F1-score of 0.93 on the English dataset, 0.75 on the Tamil dataset and 0.87 on the Tamil-English Code-Mixed dataset.

Homophobia/Transphobia Detection in Multiple Languages using SVM Classifiers and BERT-based Transformers

In this presentation I present our model for the shared task.

Detecting Signs of Depression from Social Media Text

Depression is a mental illness that negatively affects a person's well-being and can, if left untreated, lead to serious consequences such as suicide. Therefore, it is important to recognize the signs of depression early. In the last decade, social media has become one of the most common places to express one's feelings. Hence, there is a possibility of text processing and applying machine learning techniques to detect possible signs of depression. In this paper, we present our approaches to solving the shared task titled Detecting Signs of Depression from Social Media Text. We explore three different approaches to solve the challenge: fine-tuning BERT model, leveraging AutoML for the construction of features and classifier selection and finally, we explore latent spaces derived from the combination of textual and knowledge-based representations. We ranked 9th out of 31 teams in the competition. Our best solution, based on knowledge graph and textual representations, was 4.9% behind the best model in terms of Macro F1, and only 1.9% behind in terms of Recall.

Leveraging Pretrained Language Models for Detecting Homophobia and Transphobia in Social Media Comments

Downloads

Next from ACL 2022

A Roberta based Approach for Homophobia/Transphobia Detection in Social Media

Similar lecture

Ensembled Transformers Against Homophobia and Transphobia

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES