Abstract

The COVID-19 pandemic has so far accounted for reported 5.5𝑀 deaths worldwide, with 8.7% of these coming from India. The pandemic exacerbated the weakness of the Indian healthcare system. As of January 20, 2022, India is the second worst affected country with 38.2𝑀 reported cases and 487𝐾 deaths. According to epidemiologists, vaccines are an essential tool to prevent the spread of the pandemic. India’s vaccination drive began on January 16, 2021 with governmental policies being introduced to prioritize different populations of the society. Through the course of the vaccination drive, multiple new policies were also introduced to ensure that vaccines are readily available and vaccination coverage is increased. However, at the same time, some of the government policies introduced led to unintended inequities in the populations being targeted. In this report, we enumerate and analyze the inequities that existed in India’s vaccination policy drive, and also compute the effect of the new policies that were introduced. We analyze these potential inequities not only qualitatively but also quantitatively by leveraging the data that was made available through the government portals. Specifically, (a) we discover inequities that might exist in the policies, (b) we quantify the effect of new policies introduced to increase vaccination coverage, and (c) we also point the data discrepancies that exist across different data sources

Abstract

Social media in present times has a significant and growing influence. Fake news being spread on these platforms have a disruptive and damaging impact on our lives. Furthermore, as multimedia content improves the visibility of posts more than text data, it has been observed that often multi- media is being used for creating fake content. A plethora of previous multimodal-based work has tried to address the problem of modeling heteroge- neous modalities in identifying fake content. How- ever, these works have the following limitations: (1) inefficient encoding of inter-modal relations by utilizing a simple concatenation operator on the modalities at a later stage in a model, which might result in information loss; (2) training very deep neural networks with a disproportionate number of parameters on small but complex real-life multi- modal datasets result in higher chances of over- fitting. To address these limitations, we propose GAME-ON, a Graph Neural Network based end- to-end trainable framework that allows granular in- teractions within and across different modalities to learn more robust data representations for multi- modal fake news detection. We use two publicly available fake news datasets, Twitter and Weibo, for evaluations. Our model outperforms on Twitter by an average of 11% and keeps competitive perfor- mance on Weibo, within a 2.6% margin, while us- ing 65% fewer parameters than the best comparable state-of-the-art baseline.

Abstract

Data has become an integral part of day-to-day human life. Users leave behind a trail of digital footprint that includes their personal and non-personal information. A normal user puts 1.7 megabytes of data every second into the hand of service providers and trusts them to keep it safe. However, researchers have found out that in the name of improving the quality of service, the service providers, knowing or accidentally, put users’ personal information at risk of getting into the hands of an adversary. The service providers usually apply masking or anonymization before releasing the users’ data. Anonymization techniques do not guarantee privacy preservation and are proven to be prone to cross-linking attacks. In the past, researchers were able to successfully cross-link multiple datasets to leak the sensitive information of various users. Cross-linking attacks are always possible on anonymized datasets, and therefore, service providers must use a technique that guarantees privacy preservation. Differential privacy is superior for publishing sensitive information while pro- tecting privacy. It provides mathematical guarantees and prevents background knowledge attacks such that infor- mation remains private regardless of whatever information an adversary might have. This paper discusses how differential privacy can help achieve privacy guarantees for the release of sensitive heterogeneous datasets while pre- serving its utility.

Abstract

Online Social Media platforms (OSMs) have become an essential source of information. The high speed at which OSM users submit data makes moderation extremely hard. Consequently, besides offering online networking to users, the OSMs have also become carriers for spreading fake news. Knowingly or unknowingly, users circulate fake news on OSMs, adversely affecting an individual’s offline activity. To counter fake news, several dedicated websites (referred to as fact-checkers) have sprung up whose sole purpose is to identify and report fake news incidents. There are well-known datasets of fake news; however, not much work has been done regarding credible datasets of fake news in India. Therefore, we design an automated data collection pipeline to collect fake incidents reported by fact-checkers in this work. We gather 4,803 fake news incidents from June 2016 to December 2019 reported by six popular fact-checking websites in India and make this dataset (FakeNewsIndia) available to the research community. We find 5,031 tweets on Twitter and 866 videos on YouTube mentioned in these 4,803 fake news incidents. Further, we evaluate the impact of fake new incidents on the two prominent OSM platforms, namely, Twitter and YouTube. We use popularity metrics based on engagement rate and likes ratio to measure impact and categorize impact into three levels — low, medium, and high. Our learning models use features extracted from text, images, and videos present in the fake news incident articles written by fact-checking websites. Experiments show that we can predict the impact (popularity) of videos (appearing on fake news incident articles) on YouTube more accurately (with baseline accuracy ranging from 86% to 92%) as compared to the impact (popularity) of tweets on Twitter (with baseline accuracy of 37% to 41%). We need to build more intelligent models that predict tweets’ impact, appearing in fact-checking incident articles on Twitter as future work. 1.

Abstract

Hashtag segmentation is the task of breaking a hashtag into its constituent tokens. Hashtags often encode the essence of usergenerated posts, along with information like topic and sentiment, which are useful in downstream tasks. Hashtags prioritize brevity and are written in unique ways - transliterating and mixing languages, spelling variations, creative named entities. Benchmark datasets used for the hashtag segmentation task - STAN, BOUN - are small in size and extracted from a single set of tweets. However, datasets should reflect the variations in writing styles of hashtags and also account for domain and language specificity, failing which the results will misrepresent model performance. We argue that model performance should be assessed on a wider variety of hashtags, and datasets should be carefully curated. To this end, we propose HashSet, a dataset comprising of: a) 1.9k manually annotated dataset; b) 3.3M loosely supervised dataset. HashSet dataset is sampled from a different set of tweets when compared to existing datasets and provides an alternate distribution of hashtags to build and validate hashtag segmentation models. We show that the performance of SOTA models for Hashtag Segmentation drops substantially on proposed dataset, indicating that the proposed dataset provides an alternate set of hashtags to train and assess models. Datasets and results are released publicly and can be accessed

Abstract

Fraudulent jobs are an emerging threat over online recruitment platforms such as LinkedIn, Glassdoor. Fraudulent job postings affect the platform’s trustworthiness and have a negative impact on user experience. Therefore, these platforms need to detect and remove these fraudulent jobs. Generally, fraudulent job postings contain untenable facts about domain-specific entities such as mismatch in skills, industries, offered compensation, etc. However, existing approaches focus on studying writing styles, linguistics, and context-based features, and ignore the relationships among domain-specific entities. To bridge this gap, we propose an approach based on the Knowledge Graph (KG) of domain-specific entities to detect fraudulent jobs. In this paper, we present a multi-tier novel end-to-end framework called FRaudulent Jobs Detection (FRJD) Engine, which considers a) fact validation module using KGs, b) contextual module using deep neural networks c) meta-data module to capture the semantics of job postings. We conduct our experiments using a fact validation dataset containing 4 million facts extracted from job postings. Extensive evaluation shows that FRJD yields a 0.96 F1-score on the curated dataset of 157,880 job postings. Finally, we provide insights on the performance of different fact-checking algorithms on recruitment domain

Abstract

The COVID-19 pandemic has disrupted people's lives driving them to act in fear, anxiety, and anger, leading to worldwide racist events in the physical world and online social networks. Though there are works focusing on Sinophobia during the COVID-19 pandemic, less attention has been given to the recent surge in Islamophobia. A large number of positive cases arising out of the religious Tablighi Jamaat gathering has driven people towards forming anti-Muslim communities around hashtags like #coronajihad, #tablighijamaatvirus on Twitter. In addition to the online spaces, the rise in Islamophobia has also resulted in increased hate crimes in the real world. Hence, an investigation is required to create interventions. To the best of our knowledge, we present the first large-scale quantitative study linking

Abstract

The exponential rise of online social media has enabled the creation, distribution, and consumption of information at an unprecedented rate. However, it has also led to the burgeoning of various forms of online abuse. Increasing cases of online antisemitism have be- come one of the major concerns because of its socio-political con- sequences. Unlike other major forms of online abuse like racism, sexism, etc., online antisemitism has not been studied much from a machine learning perspective. To the best of our knowledge, we present the first work in the direction of automated multimodal de- tection of online antisemitism. The task poses multiple challenges that include extracting signals across multiple modalities, contex- tual references, and handling multiple aspects of antisemitism. Un- fortunately, there does not exist any publicly available benchmark corpus for this critical task. Hence, we collect and label two datasets with 3,102 and 3,509 social media posts from Twitter and Gab re- spectively. Further, we present a multimodal deep learning system that detects the presence of antisemitic content and its specific anti- semitism category using text and images from posts. We perform an extensive set of experiments on the two datasets to evaluate the ef- ficacy of the proposed system. Finally, we also present a qualitative analysis of our study.

Abstract

Social media has grown exponentially in a short period, coming to the forefront of communications and online interactions. Despite their rapid growth, social media platforms have been unable to scale to different languages globally and remain inaccessible to many. In this paper, we characterize Koo, a multilingual micro-blogging site that rose in popularity in 2021, as an Indian alternative to Twitter. We collected a dataset of 4.07 million users, 163.12 million follower-following relationships, and their content and activity across 12 languages. We study the user demographic along the lines of language, location, gender, and profession. The prominent presence of Indian languages in the discourse on Koo indicates the platform's success in promoting regional languages. We observe Koo's follower-following network to be much denser than Twitter's, comprising of closely-knit linguistic communities. An N-gram analysis of posts on Koo shows a #KooVsTwitter rhetoric, revealing the debate comparing the two platforms. Our characterization highlights the dynamics of the multilingual social network and its diverse Indian user base.

Abstract

COVID-19 infodemic has been spreading faster than the pandemic itself. The misinformation riding upon the infodemic wave poses a major threat to people’s health and governance systems. Managing this infodemic not only requires mitigating misinformation but also an early understanding of underlying psychological patterns. In this study, we present a novel epidemic response management strategy. We analyze the psychometric impact and coupling of COVID19 infodemic with official COVID-19 bulletins at the national and state level in India. We looked at them from the psycholinguistic lens of emotions and quantified the extent and coupling between them. We modified Empath, a deep skipgram-based lexicon builder, for effective capture of health-related emotions. Using this, we analyzed the lead-lag relationships between the time-evolution of these emotions in social media and official bulletins using Granger’s causality. It showed that state bulletins led the social media for some emotions such as Medical Emergency. In contrast, social media led the government bulletins for some topics such as hygiene, government, fun, and leisure. Further insights potentially relevant for policymakers and communicators engaged in mitigating misinformation are also discussed. We also introduce CoronaIndiaDataset, the first social-media-based Indian COVID-19 dataset at the national and state levels with over 5.6 million national and 2.6 million state-level tweets for the first wave of COVID-19 in India and 1.2 million national tweets for the second wave of COVID-19 in India.