Sabit Hassan

Meng Ye, Karan Sikka, Katherine Atwell et al.

Content moderation is the process of flagging content based on pre-defined platform rules. There has been a growing need for AI moderators to safeguard users as well as protect the mental health of human moderators from traumatic content. While prior works have focused on identifying hateful/offensive language, they are not adequate for meeting the challenges of content moderation since 1) moderation decisions are based on violation of rules, which subsumes detection of offensive speech, and 2) such rules often differ across communities which entails an adaptive solution. We propose to study the challenges of content moderation by introducing a multilingual dataset of 1.8 Million Reddit comments spanning 56 subreddits in English, German, Spanish and French. We perform extensive experimental analysis to highlight the underlying challenges and suggest related research problems such as cross-lingual transfer, learning under label noise (human biases), transfer of moderation models, and predicting the violated rule. Our dataset and analysis can help better prepare for the challenges and opportunities of auto moderation.

Katherine Atwell, Sabit Hassan, Malihe Alikhani

Using style-transfer models to reduce offensiveness of social media comments can help foster a more inclusive environment. However, there are no sizable datasets that contain offensive texts and their inoffensive counterparts, and fine-tuning pretrained models with limited labeled data can lead to the loss of original meaning in the style-transferred text. To address this issue, we provide two major contributions. First, we release the first publicly-available, parallel corpus of offensive Reddit comments and their style-transferred counterparts annotated by expert sociolinguists. Then, we introduce the first discourse-aware style-transfer models that can effectively reduce offensiveness in Reddit text while preserving the meaning of the original text. These models are the first to examine inferential links between the comment and the text it is replying to when transferring the style of offensive Reddit text. We propose two different methods of integrating discourse relations with pretrained transformer models and evaluate them on our dataset of offensive comments from Reddit and their inoffensive counterparts. Improvements over the baseline with respect to both automatic metrics and human evaluation indicate that our discourse-aware models are better at preserving meaning in style-transferred text when compared to the state-of-the-art discourse-agnostic models.

Sabit Hassan, Malihe Alikhani

Counterspeech can be an effective method for battling hateful content on social media. Automated counterspeech generation can aid in this process. Generated counterspeech, however, can be viable only when grounded in the context of topic, audience and sensitivity as these factors influence both the efficacy and appropriateness. In this work, we propose a novel framework based on theories of discourse to study the inferential links that connect counter speeches to the hateful comment. Within this framework, we propose: i) a taxonomy of counterspeech derived from discourse frameworks, and ii) discourse-informed prompting strategies for generating contextually-grounded counterspeech. To construct and validate this framework, we present a process for collecting an in-the-wild dataset of counterspeech from Reddit. Using this process, we manually annotate a dataset of 3.9k Reddit comment pairs for the presence of hatespeech and counterspeech. The positive pairs are annotated for 10 classes in our proposed taxonomy. We annotate these pairs with paraphrased counterparts to remove offensiveness and first-person references. We show that by using our dataset and framework, large language models can generate contextually-grounded counterspeech informed by theories of discourse. According to our human evaluation, our approaches can act as a safeguard against critical failures of discourse-agnostic models.

Mert İnan, Yang Zhong, Sabit Hassan et al.

End-to-end sign language generation models do not accurately represent the prosody in sign language. A lack of temporal and spatial variations leads to poor-quality generated presentations that confuse human interpreters. In this paper, we aim to improve the prosody in generated sign languages by modeling intensification in a data-driven manner. We present different strategies grounded in linguistics of sign language that inform how intensity modifiers can be represented in gloss annotations. To employ our strategies, we first annotate a subset of the benchmark PHOENIX-14T, a German Sign Language dataset, with different levels of intensification. We then use a supervised intensity tagger to extend the annotated dataset and obtain labels for the remaining portion of it. This enhanced dataset is then used to train state-of-the-art transformer models for sign language generation. We find that our efforts in intensification modeling yield better results when evaluated with automatic metrics. Human evaluation also indicates a higher preference of the videos generated using our model.

Yan Wang, Partho Hassan, Samiha Sadeka et al.

We introduce Lunara Aesthetic II, a publicly released, ethically sourced image dataset designed to support controlled evaluation and learning of contextual consistency in modern image generation and editing systems. The dataset comprises 2,854 anchor-linked variation pairs derived from original art and photographs created by Moonworks. Each variation pair applies contextual transformations, such as illumination, weather, viewpoint, scene composition, color tone, or mood; while preserving a stable underlying identity. Lunara Aesthetic II operationalizes identity-preserving contextual variation as a supervision signal while also retaining Lunara's signature high aesthetic scores. Results show high identity stability, strong target attribute realization, and a robust aesthetic profile that exceeds large-scale web datasets. Released under the Apache 2.0 license, Lunara Aesthetic II is intended for benchmarking, fine-tuning, and analysis of contextual generalization, identity preservation, and edit robustness in image generation and image-to-image systems with interpretable, relational supervision. The dataset is publicly available at: https://huggingface.co/datasets/moonworks/lunara-aesthetic-image-variations.

Sabit Hassan, Anthony Sicilia, Malihe Alikhani

Ensuring robust safety measures across a wide range of scenarios is crucial for user-facing systems. While Large Language Models (LLMs) can generate valuable data for safety measures, they often exhibit distributional biases, focusing on common scenarios and neglecting rare but critical cases. This can undermine the effectiveness of safety protocols developed using such data. To address this, we propose a novel framework that integrates active learning with clustering to guide LLM generation, enhancing their representativeness and robustness in safety scenarios. We demonstrate the effectiveness of our approach by constructing a dataset of 5.4K potential safety violations through an iterative process involving LLM generation and an active learner model's feedback. Our results show that the proposed framework produces a more representative set of safety scenarios without requiring prior knowledge of the underlying data distribution. Additionally, data acquired through our method improves the accuracy and F1 score of both the active learner model as well models outside the scope of active learning process, highlighting its broad applicability.

Yuya Asano, Sabit Hassan, Paras Sharma et al.

General-purpose automatic speech recognition (ASR) systems do not always perform well in goal-oriented dialogue. Existing ASR correction methods rely on prior user data or named entities. We extend correction to tasks that have no prior user data and exhibit linguistic flexibility such as lexical and syntactic variations. We propose a novel context augmentation with a large language model and a ranking strategy that incorporates contextual information from the dialogue states of a goal-oriented conversational AI and its tasks. Our method ranks (1) n-best ASR hypotheses by their lexical and semantic similarity with context and (2) context by phonetic correspondence with ASR hypotheses. Evaluated in home improvement and cooking domains with real-world users, our method improves recall and F1 of correction by 34% and 16%, respectively, while maintaining precision and false positive rate. Users rated .8-1 point (out of 5) higher when our correction method worked properly, with no decrease due to false positives.

Sabit Hassan, Anthony Sicilia, Malihe Alikhani

Ensuring that Large Language Models (LLMs) generate text representative of diverse sub-populations is essential, particularly when key concepts related to under-represented groups are scarce in the training data. We address this challenge with a novel clustering-based active learning framework, enhanced with knowledge distillation. The proposed framework transforms the intermediate outputs of the learner model, enabling effective active learning for generative tasks for the first time. Integration of clustering and knowledge distillation yields more representative models without prior knowledge of underlying data distribution and overbearing human efforts. We validate our approach in practice through case studies in counter-narration and style transfer. We construct two new datasets in tandem with model training, showing a performance improvement of 2%-10% over baseline models. Our results also show more consistent performance across various data subgroups and increased lexical diversity, underscoring our model's resilience to skewness in available data. Further, our results show that the data acquired via our approach improves the performance of secondary models not involved in the learning loop, showcasing practical utility of the framework.

Sabit Hassan, Hye-Young Chung, Xiang Zhi Tan et al.

When assisting people in daily tasks, robots need to accurately interpret visual cues and respond effectively in diverse safety-critical situations, such as sharp objects on the floor. In this context, we present M-CoDAL, a multimodal-dialogue system specifically designed for embodied agents to better understand and communicate in safety-critical situations. The system leverages discourse coherence relations to enhance its contextual understanding and communication abilities. To train this system, we introduce a novel clustering-based active learning mechanism that utilizes an external Large Language Model (LLM) to identify informative instances. Our approach is evaluated using a newly created multimodal dataset comprising 1K safety violations extracted from 2K Reddit images. These violations are annotated using a Large Multimodal Model (LMM) and verified by human annotators. Results with this dataset demonstrate that our approach improves resolution of safety situations, user sentiment, as well as safety of the conversation. Next, we deploy our dialogue system on a Hello Robot Stretch robot and conduct a within-subject user study with real-world participants. In the study, participants role-play two safety scenarios with different levels of severity with the robot and receive interventions from our model and a baseline system powered by OpenAI's ChatGPT. The study results corroborate and extend the findings from the automated evaluation, showing that our proposed system is more persuasive in a real-world embodied agent setting.

Yan Wang, M M Sayeef Abdullah, Partho Hassan et al.

The dataset spans diverse artistic styles, including regionally grounded aesthetics from the Middle East, Northern Europe, East Asia, and South Asia, alongside general categories such as sketch and oil painting. All images are generated using the Moonworks Lunara model and intentionally crafted to embody distinct, high-quality aesthetic styles, yielding a first-of-its-kind dataset with substantially higher aesthetic scores, exceeding even aesthetics-focused datasets, and general-purpose datasets by a larger margin. Each image is accompanied by a human-refined prompt and structured annotations that jointly describe salient objects, attributes, relationships, and stylistic cues. Unlike large-scale web-derived datasets that emphasize breadth over precision, the Lunara Aesthetic Dataset prioritizes aesthetic quality, stylistic diversity, and licensing transparency, and is released under the Apache 2.0 license to support research and unrestricted academic and commercial use.

Yan Wang, Heidi Ann Scharf Donovan, Sabit Hassan et al.

Patients who effectively manage their symptoms often demonstrate higher levels of engagement in conversations and interventions with healthcare practitioners. This engagement is multifaceted, encompassing cognitive and socio-affective dimensions. Consequently, it is crucial for AI systems to understand the engagement in natural conversations between patients and practitioners to better contribute toward patient care. In this paper, we present a novel dataset (MedNgage), which consists of patient-nurse conversations about cancer symptom management. We manually annotate the dataset with a novel framework of categories of patient engagement from two different angles, namely: i) socio-affective (3.1K spans), and ii) cognitive use of language (1.8K spans). Through statistical analysis of the data that is annotated using our framework, we show a positive correlation between patient symptom management outcomes and their engagement in conversations. Additionally, we demonstrate that pre-trained transformer models fine-tuned on our dataset can reliably predict engagement classes in patient-nurse conversations. Lastly, we use LIME (Ribeiro et al., 2016) to analyze the underlying challenges of the tasks that state-of-the-art transformer models encounter. The de-identified data is available for research purposes upon request.

Sabit Hassan, Malihe Alikhani

Despite recent advancements, NLP models continue to be vulnerable to bias. This bias often originates from the uneven distribution of real-world data and can propagate through the annotation process. Escalated integration of these models in our lives calls for methods to mitigate bias without overbearing annotation costs. While active learning (AL) has shown promise in training models with a small amount of annotated data, AL's reliance on the model's behavior for selective sampling can lead to an accumulation of unwanted bias rather than bias mitigation. However, infusing clustering with AL can overcome the bias issue of both AL and traditional annotation methods while exploiting AL's annotation efficiency. In this paper, we propose a novel adaptive clustering-based active learning algorithm, D-CALM, that dynamically adjusts clustering and annotation efforts in response to an estimated classifier error-rate. Experiments on eight datasets for a diverse set of text classification tasks, including emotion, hatespeech, dialog act, and book type detection, demonstrate that our proposed algorithm significantly outperforms baseline AL approaches with both pretrained transformers and traditional Support Vector Machines. D-CALM showcases robustness against different measures of information gain and, as evident from our analysis of label and error distribution, can significantly reduce unwanted model bias.

Hamdy Mubarak, Sabit Hassan, Shammur Absar Chowdhury

We introduce a generic, language-independent method to collect a large percentage of offensive and hate tweets regardless of their topics or genres. We harness the extralinguistic information embedded in the emojis to collect a large number of offensive tweets. We apply the proposed method on Arabic tweets and compare it with English tweets - analysing key cultural differences. We observed a constant usage of these emojis to represent offensiveness throughout different timespans on Twitter. We manually annotate and publicly release the largest Arabic dataset for offensive, fine-grained hate speech, vulgar and violence content. Furthermore, we benchmark the dataset for detecting offensiveness and hate speech using different transformer architectures and perform in-depth linguistic analysis. We evaluate our models on external datasets - a Twitter dataset collected using a completely different method, and a multi-platform dataset containing comments from Twitter, YouTube and Facebook, for assessing generalization capability. Competitive results on these datasets suggest that the data collected using our method captures universal characteristics of offensive language. Our findings also highlight the common words used in offensive communications, common targets for hate speech, specific patterns in violence tweets; and pinpoint common classification errors that can be attributed to limitations of NLP models. We observe that even state-of-the-art transformer models may fail to take into account culture, background and context or understand nuances present in real-world data such as sarcasm.

Hamdy Mubarak, Sabit Hassan, Shammur Absar Chowdhury et al.

The emergence of the COVID-19 pandemic and the first global infodemic have changed our lives in many different ways. We relied on social media to get the latest information about the COVID-19 pandemic and at the same time to disseminate information. The content in social media consisted not only health related advises, plans, and informative news from policy makers, but also contains conspiracies and rumors. It became important to identify such information as soon as they are posted to make actionable decisions (e.g., debunking rumors, or taking certain measures for traveling). To address this challenge, we develop and publicly release the first largest manually annotated Arabic tweet dataset, ArCovidVac, for the COVID-19 vaccination campaign, covering many countries in the Arab region. The dataset is enriched with different layers of annotation, including, (i) Informativeness (more vs. less importance of the tweets); (ii) fine-grained tweet content types (e.g., advice, rumors, restriction, authenticate news/information); and (iii) stance towards vaccination (pro-vaccination, neutral, anti-vaccination). Further, we performed in-depth analysis of the data, exploring the popularity of different vaccines, trending hashtags, topics and presence of offensiveness in the tweets. We studied the data for individual types of tweets and temporal changes in stance towards vaccine. We benchmarked the ArCovidVac dataset using transformer architectures for informativeness, content types, and stance detection.

Sabit Hassan, Shaden Shaar, Kareem Darwish

Emotion detection can provide us with a window into understanding human behavior. Due to the complex dynamics of human emotions, however, constructing annotated datasets to train automated models can be expensive. Thus, we explore the efficacy of cross-lingual approaches that would use data from a source language to build models for emotion detection in a target language. We compare three approaches, namely: i) using inherently multilingual models; ii) translating training data into the target language; and iii) using an automatically tagged parallel corpus. In our study, we consider English as the source language with Arabic and Spanish as target languages. We study the effectiveness of different classification models such as BERT and SVMs trained with different features. Our BERT-based monolingual models that are trained on target language data surpass state-of-the-art (SOTA) by 4% and 5% absolute Jaccard score for Arabic and Spanish respectively. Next, we show that using cross-lingual approaches with English data alone, we can achieve more than 90% and 80% relative effectiveness of the Arabic and Spanish BERT models respectively. Lastly, we use LIME to analyze the challenges of training cross-lingual models for different language pairs

Ahmed Abdelali, Sabit Hassan, Hamdy Mubarak et al.

Pretraining Bidirectional Encoder Representations from Transformers (BERT) for downstream NLP tasks is a non-trival task. We pretrained 5 BERT models that differ in the size of their training sets, mixture of formal and informal Arabic, and linguistic preprocessing. All are intended to support Arabic dialects and social media. The experiments highlight the centrality of data diversity and the efficacy of linguistically aware segmentation. They also highlight that more data or more training step do not necessitate better models. Our new models achieve new state-of-the-art results on several downstream tasks. The resulting models are released to the community under the name QARiB.

Hamdy Mubarak, Sabit Hassan

Over the past few months, there were huge numbers of circulating tweets and discussions about Coronavirus (COVID-19) in the Arab region. It is important for policy makers and many people to identify types of shared tweets to better understand public behavior, topics of interest, requests from governments, sources of tweets, etc. It is also crucial to prevent spreading of rumors and misinformation about the virus or bad cures. To this end, we present the largest manually annotated dataset of Arabic tweets related to COVID-19. We describe annotation guidelines, analyze our dataset and build effective machine learning and transformer based models for classification.

Ahmed Abdelali, Hamdy Mubarak, Younes Samih et al.

We present QADI, an automatically collected dataset of tweets belonging to a wide range of country-level Arabic dialects -covering 18 different countries in the Middle East and North Africa region. Our method for building this dataset relies on applying multiple filters to identify users who belong to different countries based on their account descriptions and to eliminate tweets that are either written in Modern Standard Arabic or contain inappropriate language. The resultant dataset contains 540k tweets from 2,525 users who are evenly distributed across 18 Arab countries. Using intrinsic evaluation, we show that the labels of a set of randomly selected tweets are 91.5% accurate. For extrinsic evaluation, we are able to build effective country-level dialect identification on tweets with a macro-averaged F1-score of 60.6% across 18 classes.