Source author record

Ponnurangam Kumaraguru

Ponnurangam Kumaraguru appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record

Catalog footprint

What is connected

55works
12topics
4close collaborators

Actions

Connect this record

Log in to claim

Research graph

See the researcher in context

Open full explorer

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

Published work

55 published item(s)

preprint2026arXiv

Intrinsic Guardrails: How Semantic Geometry of Personality Interacts with Emergent Misalignment in LLMs

Fine-tuning Large Language Models (LLMs) on benign narrow data can sometimes induce broad harmful behaviors, a vulnerability termed emergent misalignment (EM). While prior work links these failures to specific directions in the activation space, their relationship to the model's broader persona remains unexplored. We map the latent personality space of LLMs through established psychometric profiles like the Big Five, Dark Triad, and LLM-specific behaviors (e.g. evil, sycophancy), and show that the semantic geometry is highly stable across aligned models and their corrupted fine-tunes. Through causal interventions, we find that directions isolating social valence, such as the 'Evil' persona vector, and a Semantic Valence Vector (SVV) that we introduce, function as intrinsic guardrails: ablating them drives the misalignment rates above $40$%, while amplifying them suppresses the failure mode to less than $3$%. Leveraging the structural stability of the personality space, we show that vectors extracted $\textit{a priori}$ from an instruct-tuned model transfer zero-shot to successfully regulate EM in corrupted fine-tunes. Overall, our findings suggest that harmful fine-tuning does not overwrite a model's internal representation of personality, allowing conserved representations to serve as robust, cross-distribution guardrails.

preprint2026arXiv

MACA: A Framework for Distilling Trustworthy LLMs into Efficient Retrievers

Modern enterprise retrieval systems must handle short, underspecified queries such as ``foreign transaction fee refund'' and ``recent check status''. In these cases, semantic nuance and metadata matter but per-query large language model (LLM) re-ranking and manual labeling are costly. We present Metadata-Aware Cross-Model Alignment (MACA), which distills a calibrated metadata aware LLM re-ranker into a compact student retriever, avoiding online LLM calls. A metadata-aware prompt verifies the teacher's trustworthiness by checking consistency under permutations and robustness to paraphrases, then supplies listwise scores, hard negatives, and calibrated relevance margins. The student trains with MACA's MetaFusion objective, which combines a metadata conditioned ranking loss with a cross model margin loss so it learns to push the correct answer above semantically similar candidates with mismatched topic, sub-topic, or entity. On a proprietary consumer banking FAQ corpus and BankFAQs, the MACA teacher surpasses a MAFA baseline at Accuracy@1 by five points on the proprietary set and three points on BankFAQs. MACA students substantially outperform pretrained encoders; e.g., on the proprietary corpus MiniLM Accuracy@1 improves from 0.23 to 0.48, while keeping inference free of LLM calls and supporting retrieval-augmented generation.

preprint2026arXiv

Shadow Unlearning: A Neuro-Semantic Approach to Fidelity-Preserving Faceless Forgetting in LLMs

Machine unlearning aims to selectively remove the influence of specific training samples to satisfy privacy regulations such as the GDPR's 'Right to be Forgotten'. However, many existing methods require access to the data being removed, exposing it to membership inference attacks and potential misuse of Personally Identifiable Information (PII). We address this critical challenge by proposing Shadow Unlearning, a novel paradigm of approximate unlearning, that performs machine unlearning on anonymized forget data without exposing PII. We further propose a novel privacy-preserving framework, Neuro-Semantic Projector Unlearning (NSPU) to achieve Shadow unlearning. To evaluate our method, we compile Multi-domain Fictitious Unlearning (MuFU) forget set across five diverse domains and introduce an evaluation stack to quantify the trade-off between knowledge retention and unlearning effectiveness. Experimental results on various LLMs show that NSPU achieves superior unlearning performance, preserves model utility, and enhances user privacy. Additionally, the proposed approach is at least 10 times more computationally efficient than standard unlearning approaches. Our findings foster a new direction for privacy-aware machine unlearning that balances data protection and model fidelity.

preprint2025arXiv

PrivacyBench: A Conversational Benchmark for Evaluating Privacy in Personalized AI

Personalized AI agents rely on access to a user's digital footprint, which often includes sensitive data from private emails, chats and purchase histories. Yet this access creates a fundamental societal and privacy risk: systems lacking social-context awareness can unintentionally expose user secrets, threatening digital well-being. We introduce PrivacyBench, a benchmark with socially grounded datasets containing embedded secrets and a multi-turn conversational evaluation to measure secret preservation. Testing Retrieval-Augmented Generation (RAG) assistants reveals that they leak secrets in up to 26.56% of interactions. A privacy-aware prompt lowers leakage to 5.12%, yet this measure offers only partial mitigation. The retrieval mechanism continues to access sensitive data indiscriminately, which shifts the entire burden of privacy preservation onto the generator. This creates a single point of failure, rendering current architectures unsafe for wide-scale deployment. Our findings underscore the urgent need for structural, privacy-by-design safeguards to ensure an ethical and inclusive web for everyone.

preprint2022arXiv

Co-WIN: Really Winning? Analysing Inequity in India's Vaccination Response

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.

preprint2022arXiv

Diagnosing Data from ICTs to Provide Focused Assistance in Agricultural Adoptions

In the last two decades, ICTs have played a pivotal role in empowering rural populations in India by making knowledge more accessible. Digital Green (DG) is one such ICT that employs a participatory approach with smallholder farmers to produce instructional videos that encompass content specific to them. With help of human mediators, they disseminate these videos using projectors to improve the adoption of agricultural practices. DG's web-based data tracker stores attendance and adoption logs of millions of farmers, videos screened and their demographic information. We leverage this data for a period of ten years between 2010-2020 across five states in India and use it to conduct a holistic evaluation of the ICT. First, we find disparities in adoption rates of farmers, following which we use statistical tests to identify different factors that lead to these disparities and gender-based inequalities. Second, to provide assistance to farmers facing challenges, we model the adoption of practices from a video as a prediction problem and experiment with different model architectures. Our classifier achieves accuracies ranging from 79% to 90% across the five states, demonstrating its potential for assisting future ethnographic investigations. Third, we use SHAP values in conjunction with our model for explaining the impact of various network, content and demographic features on adoption. Our research finds that farmers greatly benefit from past adopters of a video from their group and village. We also discover that videos with a low content-specificity benefit some farmers more than others. Next, we highlight the implications of our findings by translating them into recommendations for community building, revisiting participatory approach and mitigating inequalities. We conclude with a discussion on how our work can assist future investigations into the lived experiences of farmers.

preprint2022arXiv

Erasing Labor with Labor: Dark Patterns and Lockstep Behaviors on Google Play

Google Play's policy forbids the use of incentivized installs, ratings, and reviews to manipulate the placement of apps. However, there still exist apps that incentivize installs for other apps on the platform. To understand how install-incentivizing apps affect users, we examine their ecosystem through a socio-technical lens and perform a mixed-methods analysis of their reviews and permissions. Our dataset contains 319K reviews collected daily over five months from 60 such apps that cumulatively account for over 160.5M installs. We perform qualitative analysis of reviews to reveal various types of dark patterns that developers incorporate in install-incentivizing apps, highlighting their normative concerns at both user and platform levels. Permissions requested by these apps validate our discovery of dark patterns, with over 92% apps accessing sensitive user information. We find evidence of fraudulent reviews on install-incentivizing apps, following which we model them as an edge stream in a dynamic bipartite graph of apps and reviewers. Our proposed reconfiguration of a state-of-the-art microcluster anomaly detection algorithm yields promising preliminary results in detecting this fraud. We discover highly significant lockstep behaviors exhibited by reviews that aim to boost the overall rating of an install-incentivizing app. Upon evaluating the 50 most suspicious clusters of boosting reviews detected by the algorithm, we find (i) near-identical pairs of reviews across 94% (47 clusters), and (ii) over 35% (1,687 of 4,717 reviews) present in the same form near-identical pairs within their cluster. Finally, we conclude with a discussion on how fraud is intertwined with labor and poses a threat to the trust and transparency of Google Play.

preprint2022arXiv

FaIRCoP: Facial Image Retrieval using Contrastive Personalization

Retrieving facial images from attributes plays a vital role in various systems such as face recognition and suspect identification. Compared to other image retrieval tasks, facial image retrieval is more challenging due to the high subjectivity involved in describing a person's facial features. Existing methods do so by comparing specific characteristics from the user's mental image against the suggested images via high-level supervision such as using natural language. In contrast, we propose a method that uses a relatively simpler form of binary supervision by utilizing the user's feedback to label images as either similar or dissimilar to the target image. Such supervision enables us to exploit the contrastive learning paradigm for encapsulating each user's personalized notion of similarity. For this, we propose a novel loss function optimized online via user feedback. We validate the efficacy of our proposed approach using a carefully designed testbed to simulate user feedback and a large-scale user study. Our experiments demonstrate that our method iteratively improves personalization, leading to faster convergence and enhanced recommendation relevance, thereby, improving user satisfaction. Our proposed framework is also equipped with a user-friendly web interface with a real-time experience for facial image retrieval.

preprint2022arXiv

HashSet -- A Dataset For Hashtag Segmentation

Hashtag segmentation is the task of breaking a hashtag into its constituent tokens. Hashtags often encode the essence of user-generated 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.

preprint2022arXiv

Learning to Automate Follow-up Question Generation using Process Knowledge for Depression Triage on Reddit Posts

Conversational Agents (CAs) powered with deep language models (DLMs) have shown tremendous promise in the domain of mental health. Prominently, the CAs have been used to provide informational or therapeutic services to patients. However, the utility of CAs to assist in mental health triaging has not been explored in the existing work as it requires a controlled generation of follow-up questions (FQs), which are often initiated and guided by the mental health professionals (MHPs) in clinical settings. In the context of depression, our experiments show that DLMs coupled with process knowledge in a mental health questionnaire generate 12.54% and 9.37% better FQs based on similarity and longest common subsequence matches to questions in the PHQ-9 dataset respectively, when compared with DLMs without process knowledge support. Despite coupling with process knowledge, we find that DLMs are still prone to hallucination, i.e., generating redundant, irrelevant, and unsafe FQs. We demonstrate the challenge of using existing datasets to train a DLM for generating FQs that adhere to clinical process knowledge. To address this limitation, we prepared an extended PHQ-9 based dataset, PRIMATE, in collaboration with MHPs. PRIMATE contains annotations regarding whether a particular question in the PHQ-9 dataset has already been answered in the user's initial description of the mental health condition. We used PRIMATE to train a DLM in a supervised setting to identify which of the PHQ-9 questions can be answered directly from the user's post and which ones would require more information from the user. Using performance analysis based on MCC scores, we show that PRIMATE is appropriate for identifying questions in PHQ-9 that could guide generative DLMs towards controlled FQ generation suitable for aiding triaging. Dataset created as a part of this research: https://github.com/primate-mh/Primate2022

preprint2022arXiv

PreCogIIITH at HinglishEval : Leveraging Code-Mixing Metrics & Language Model Embeddings To Estimate Code-Mix Quality

Code-Mixing is a phenomenon of mixing two or more languages in a speech event and is prevalent in multilingual societies. Given the low-resource nature of Code-Mixing, machine generation of code-mixed text is a prevalent approach for data augmentation. However, evaluating the quality of such machine generated code-mixed text is an open problem. In our submission to HinglishEval, a shared-task collocated with INLG2022, we attempt to build models factors that impact the quality of synthetically generated code-mix text by predicting ratings for code-mix quality.

preprint2022arXiv

TweetBoost: Influence of Social Media on NFT Valuation

NFT or Non-Fungible Token is a token that certifies a digital asset to be unique. A wide range of assets including, digital art, music, tweets, memes, are being sold as NFTs. NFT-related content has been widely shared on social media sites such as Twitter. We aim to understand the dominant factors that influence NFT asset valuation. Towards this objective, we create a first-of-its-kind dataset linking Twitter and OpenSea (the largest NFT marketplace) to capture social media profiles and linked NFT assets. Our dataset contains 245,159 tweets posted by 17,155 unique users, directly linking 62,997 NFT assets on OpenSea worth 19 Million USD. We have made the dataset public. We analyze the growth of NFTs, characterize the Twitter users promoting NFT assets, and gauge the impact of Twitter features on the virality of an NFT. Further, we investigate the effectiveness of different social media and NFT platform features by experimenting with multiple machine learning and deep learning models to predict an asset's value. Our results show that social media features improve the accuracy by 6% over baseline models that use only NFT platform features. Among social media features, count of user membership lists, number of likes and retweets are important features.

preprint2022arXiv

What's Kooking? Characterizing India's Emerging Social Network, Koo

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.

preprint2022arXiv

Zero-shot Entity and Tweet Characterization with Designed Conditional Prompts and Contexts

Online news and social media have been the de facto mediums to disseminate information globally from the beginning of the last decade. However, bias in content and purpose of intentions are not regulated, and managing bias is the responsibility of content consumers. In this regard, understanding the stances and biases of news sources towards specific entities becomes important. To address this problem, we use pretrained language models, which have been shown to bring about good results with no task-specific training or few-shot training. In this work, we approach the problem of characterizing Named Entities and Tweets as an open-ended text classification and open-ended fact probing problem.We evaluate the zero-shot language model capabilities of Generative Pretrained Transformer 2 (GPT-2) to characterize Entities and Tweets subjectively with human psychology-inspired and logical conditional prefixes and contexts. First, we fine-tune the GPT-2 model on a sufficiently large news corpus and evaluate subjective characterization of popular entities in the corpus by priming with prefixes. Second, we fine-tune GPT-2 with a Tweets corpus from a few popular hashtags and evaluate characterizing tweets by priming the language model with prefixes, questions, and contextual synopsis prompts. Entity characterization results were positive across measures and human evaluation.

preprint2021arXiv

Capitol (Pat)riots: A comparative study of Twitter and Parler

On 6 January 2021, a mob of right-wing conservatives stormed the USA Capitol Hill interrupting the session of congress certifying 2020 Presidential election results. Immediately after the start of the event, posts related to the riots started to trend on social media. A social media platform which stood out was a free speech endorsing social media platform Parler; it is being claimed as the platform on which the riots were planned and talked about. Our report presents a contrast between the trending content on Parler and Twitter around the time of riots. We collected data from both platforms based on the trending hashtags and draw comparisons based on what are the topics being talked about, who are the people active on the platforms and how organic is the content generated on the two platforms. While the content trending on Twitter had strong resentments towards the event and called for action against rioters and inciters, Parler content had a strong conservative narrative echoing the ideas of voter fraud similar to the attacking mob. We also find a disproportionately high manipulation of traffic on Parler when compared to Twitter.

preprint2021arXiv

Factorization of Fact-Checks for Low Resource Indian Languages

The advancement in technology and accessibility of internet to each individual is revolutionizing the real time information. The liberty to express your thoughts without passing through any credibility check is leading to dissemination of fake content in the ecosystem. It can have disastrous effects on both individuals and society as a whole. The amplification of fake news is becoming rampant in India too. Debunked information often gets republished with a replacement description, claiming it to depict some different incidence. To curb such fabricated stories, it is necessary to investigate such deduplicates and false claims made in public. The majority of studies on automatic fact-checking and fake news detection is restricted to English only. But for a country like India where only 10% of the literate population speak English, role of regional languages in spreading falsity cannot be undermined. In this paper, we introduce FactDRIL: the first large scale multilingual Fact-checking Dataset for Regional Indian Languages. We collect an exhaustive dataset across 7 months covering 11 low-resource languages. Our propose dataset consists of 9,058 samples belonging to English, 5,155 samples to Hindi and remaining 8,222 samples are distributed across various regional languages, i.e. Bangla, Marathi, Malayalam, Telugu, Tamil, Oriya, Assamese, Punjabi, Urdu, Sinhala and Burmese. We also present the detailed characterization of three M's (multi-lingual, multi-media, multi-domain) in the FactDRIL accompanied with the complete list of other varied attributes making it a unique dataset to study. Lastly, we present some potential use cases of the dataset. We expect this dataset will be a valuable resource and serve as a starting point to fight proliferation of fake news in low resource languages.

preprint2020arXiv

Don't cross that stop line: Characterizing Traffic Violations in Metropolitan Cities

In modern metropolitan cities, the task of ensuring safe roads is of paramount importance. Automated systems of e-challans (Electronic traffic-violation receipt) are now being deployed across cities to record traffic violations and to issue fines. In the present study, an automated e-challan system established in Ahmedabad (Gujarat, India) has been analyzed for characterizing user behaviour, violation types as well as finding spatial and temporal patterns in the data. We describe a method of collecting e-challan data from the e-challan portal of Ahmedabad traffic police and create a dataset of over 3 million e-challans. The dataset was first analyzed to characterize user behaviour with respect to repeat offenses and fine payment. We demonstrate that a lot of users repeat their offenses (traffic violation) frequently and are less likely to pay fines of higher value. Next, we analyze the data from a spatial and temporal perspective and identify certain spatio-temporal patterns present in our dataset. We find that there is a drastic increase/decrease in the number of e-challans issued during the festival days and identify a few hotspots in the city that have high intensity of traffic violations. In the end, we propose a set of 5 features to model recidivism in traffic violations and train multiple classifiers on our dataset to evaluate the effectiveness of our proposed features. The proposed approach achieves 95% accuracy on the dataset.

preprint2020arXiv

Hashtags are (not) judgemental: The untold story of Lok Sabha elections 2019

Hashtags in online social media have become a way for users to build communities around topics, promote opinions, and categorize messages. In the political context, hashtags on Twitter are used by users to campaign for their parties, spread news, or to get followers and get a general idea by following a discussion built around a hashtag. In the past, researchers have studied certain types and specific properties of hashtags by utilizing a lot of data collected around hashtags. In this paper, we perform a large-scale empirical analysis of elections using only the hashtags shared on Twitter during the 2019 Lok Sabha elections in India. We study the trends and events unfolded on the ground, the latent topics to uncover representative hashtags and semantic similarity to relate hashtags with the election outcomes. We collect over 24 million hashtags to perform extensive experiments. First, we find the trending hashtags to cross-reference them with the tweets in our dataset to list down notable events. Second, we use Latent Dirichlet Allocation to find topic patterns in the dataset. In the end, we use skip-gram word embedding model to find semantically similar hashtags. We propose popularity and an influence metric to predict election outcomes using just the hashtags. Empirical results show that influence is a good measure to predict the election outcome.

preprint2020arXiv

Modeling Citation Trajectories of Scientific Papers

Several network growth models have been proposed in the literature that attempt to incorporate properties of citation networks. Generally, these models aim at retaining the degree distribution observed in real-world networks. In this work, we explore whether existing network growth models can realize the diversity in citation growth exhibited by individual papers - a new node-centric property observed recently in citation networks across multiple domains of research. We theoretically and empirically show that the network growth models which are solely based on degree and/or intrinsic fitness cannot realize certain temporal growth behaviors that are observed in real-world citation networks. To this end, we propose two new growth models that localize the influence of papers through an appropriate attachment mechanism. Experimental results on the real-world citation networks of Computer Science and Physics domains show that our proposed models can better explain the temporal behavior of citation networks than existing models.

preprint2020arXiv

Psychometric Analysis and Coupling of Emotions Between State Bulletins and Twitter in India during COVID-19 Infodemic

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. Since social media is the largest source of information, managing the infodemic not only requires mitigating of misinformation but also an early understanding of psychological patterns resulting from it. During the COVID-19 crisis, Twitter alone has seen a sharp 45% increase in the usage of its curated events page, and a 30% increase in its direct messaging usage, since March 6th 2020. In this study, we analyze the psychometric impact and coupling of the COVID-19 infodemic with the official bulletins related to COVID-19 at the national and state level in India. We look at these two sources with a psycho-linguistic lens of emotions and quantified the extent and coupling between the two. We modified path, a deep skip-gram based open-sourced lexicon builder for effective capture of health-related emotions. We were then able to capture the time-evolution of health-related emotions in social media and official bulletins. An analysis of lead-lag relationships between the time series of extracted emotions from official bulletins and social media using Granger's causality showed that state bulletins were leading the social media for some emotions such as Medical Emergency. Further insights that are potentially relevant for the policymaker and the communicators actively engaged in mitigating misinformation are also discussed. Our paper also introduces CoronaIndiaDataset2, the first social media based COVID-19 dataset at national and state levels from India with over 5.6 million national and 2.6 million state-level tweets. Finally, we present our findings as COVibes, an interactive web application capturing psychometric insights captured upon the CoronaIndiaDataset, both at a national and state level.

preprint2020arXiv

Trawling for Trolling: A Dataset

The ability to accurately detect and filter offensive content automatically is important to ensure a rich and diverse digital discourse. Trolling is a type of hurtful or offensive content that is prevalent in social media, but is underrepresented in datasets for offensive content detection. In this work, we present a dataset that models trolling as a subcategory of offensive content. The dataset was created by collecting samples from well-known datasets and reannotating them along precise definitions of different categories of offensive content. The dataset has 12,490 samples, split across 5 classes; Normal, Profanity, Trolling, Derogatory and Hate Speech. It encompasses content from Twitter, Reddit and Wikipedia Talk Pages. Models trained on our dataset show appreciable performance without any significant hyperparameter tuning and can potentially learn meaningful linguistic information effectively. We find that these models are sensitive to data ablation which suggests that the dataset is largely devoid of spurious statistical artefacts that could otherwise distract and confuse classification models.

preprint2020arXiv

WorkerRep: Immutable Reputation System For Crowdsourcing Platform Based on Blockchain

Crowdsourcing is a process wherein an individual or an organisation utilizes the talent pool present over the Internet to accomplish their task. The existing crowdsourcing platforms and their reputation computation are centralised and hence prone to various attacks or malicious manipulation of the data by the central entity. A few distributed crowdsourcing platforms have been proposed but they lack a robust reputation mechanism. So we propose a decentralised crowdsourcing platform having an immutable reputation mechanism to tackle these problems. It is built on top of Ethereum network and does not require the user to trust a third party for a non malicious experience. It also utilizes IOTAs consensus mechanism which reduces the cost for task evaluation significantly.

preprint2016arXiv

Emotions, Demographics and Sociability in Twitter Interactions

The social connections people form online affect the quality of information they receive and their online experience. Although a host of socioeconomic and cognitive factors were implicated in the formation of offline social ties, few of them have been empirically validated, particularly in an online setting. In this study, we analyze a large corpus of geo-referenced messages, or tweets, posted by social media users from a major US metropolitan area. We linked these tweets to US Census data through their locations. This allowed us to measure emotions expressed in the tweets posted from an area, the structure of social connections, and also use that area's socioeconomic characteristics in analysis. %We extracted the structure of online social interactions from the people mentioned in tweets from that area. We find that at an aggregate level, places where social media users engage more deeply with less diverse social contacts are those where they express more negative emotions, like sadness and anger. Demographics also has an impact: these places have residents with lower household income and education levels. Conversely, places where people engage less frequently but with diverse contacts have happier, more positive messages posted from them and also have better educated, younger, more affluent residents. Results suggest that cognitive factors and offline characteristics affect the quality of online interactions. Our work highlights the value of linking social media data to traditional data sources, such as US Census, to drive novel analysis of online behavior.

preprint2016arXiv

KidsTube: Detection, Characterization and Analysis of Child Unsafe Content & Promoters on YouTube

YouTube draws large number of users who contribute actively by uploading videos or commenting on existing videos. However, being a crowd sourced and large content pushed onto it, there is limited control over the content. This makes malicious users push content (videos and comments) which is inappropriate (unsafe), particularly when such content is placed around cartoon videos which are typically watched by kids. In this paper, we focus on presence of unsafe content for children and users who promote it. For detection of child unsafe content and its promoters, we perform two approaches, one based on supervised classification which uses an extensive set of video-level, user-level and comment-level features and another based Convolutional Neural Network using video frames. Detection accuracy of 85.7% is achieved which can be leveraged to build a system to provide a safe YouTube experience for kids. Through detailed characterization studies, we are able to successfully conclude that unsafe content promoters are less popular and engage less as compared with other users. Finally, using a network of unsafe content promoters and other users based on their engagements (likes, subscription and playlist addition) and other factors, we find that unsafe content is present very close to safe content and unsafe content promoters form very close knit communities with other users, thereby further increasing the likelihood of a child getting getting exposed to unsafe content.

preprint2016arXiv

Me, Myself and My Killfie: Characterizing and Preventing Selfie Deaths

Over the past couple of years, clicking and posting selfies has become a popular trend. However, since March 2014, 127 people have died and many have been injured while trying to click a selfie. Researchers have studied selfies for understanding the psychology of the authors, and understanding its effect on social media platforms. In this work, we perform a comprehensive analysis of the selfie-related casualties and infer various reasons behind these deaths. We use inferences from incidents and from our understanding of the features, we create a system to make people more aware of the dangerous situations in which these selfies are taken. We use a combination of text-based, image-based and location-based features to classify a particular selfie as dangerous or not. Our method ran on 3,155 annotated selfies collected on Twitter gave 73% accuracy. Individually the image-based features were the most informative for the prediction task. The combination of image-based and location-based features resulted in the best accuracy. We have made our code and dataset available at http://labs.precog.iiitd.edu.in/killfie.

preprint2016arXiv

PicHunt: Social Media Image Retrieval for Improved Law Enforcement

First responders are increasingly using social media to identify and reduce crime for well-being and safety of the society. Images shared on social media hurting religious, political, communal and other sentiments of people, often instigate violence and create law & order situations in society. This results in the need for first responders to inspect the spread of such images and users propagating them on social media. In this paper, we present a comparison between different hand-crafted features and a Convolutional Neural Network (CNN) model to retrieve similar images, which outperforms state-of-art hand-crafted features. We propose an Open-Source-Intelligent (OSINT) real-time image search system, robust to retrieve modified images that allows first responders to analyze the current spread of images, sentiments floating and details of users propagating such content. The system also aids officials to save time of manually analyzing the content by reducing the search space on an average by 67%.

preprint2016arXiv

Visual Themes and Sentiment on Social Networks To Aid First Responders During Crisis Events

Online Social Networks explode with activity whenever a crisis event takes place. Most content generated as part of this activity is a mixture of text and images, and is particularly useful for first responders to identify popular topics of interest and gauge the pulse and sentiment of citizens. While multiple researchers have used text to identify, analyze and measure themes and public sentiment during such events, little work has explored visual themes floating on networks in the form of images, and the sentiment inspired by them. Given the potential of visual content for influencing users' thoughts and emotions, we perform a large scale analysis to compare popular themes and sentiment across images and textual content posted on Facebook during the terror attacks that took place in Paris in 2015. Using state-of-the-art image summarization techniques, we discovered multiple visual themes which were popular in images, but were not identifiable through text. We uncovered instances of misinformation and false flag (conspiracy) theories among popular image themes, which were not prominent in user generated textual content, and can be of particular inter- est to first responders. Our analysis also revealed that while textual content posted after the attacks reflected negative sentiment, images inspired positive sentiment. To the best of our knowledge, this is the first large scale study of images posted on social networks during a crisis event.

preprint2015arXiv

Characterising Behavior and Emotions on Social Media for Safety: Exploring Online Communication between Police and Citizens

Increased use of social media by police to connect with citizens has encouraged researchers to study different aspects of information exchange (e.g. type of information, credibility and propagation) during emergency and crisis situation. Research studies lack understanding of human behavior such as engagement, emotions and social interaction between citizen and police department on social media. Several social media studies explore and show technological implications of human behavioral aspects in various contexts such as workplace interaction and depression in young mothers. In this paper, we study online interactions between citizens and Indian police in context of day-to-day policing, including safety concerns, advisories, etc. Indian police departments use Facebook to issue advisories, send alerts and receive citizen complaints and suggestions regarding safety issues and day-to-day policing. We explore how citizens express their emotions and social support on Facebook. Our work discusses technological implications of behavioral aspects on social well being of citizens.

preprint2015arXiv

Detecting Malicious Content on Facebook

Online Social Networks (OSNs) witness a rise in user activity whenever an event takes place. Malicious entities exploit this spur in user-engagement levels to spread malicious content that compromises system reputation and degrades user experience. It also generates revenue from advertisements, clicks, etc. for the malicious entities. Facebook, the world's biggest social network, is no exception and has recently been reported to face much abuse through scams and other type of malicious content, especially during news making events. Recent studies have reported that spammers earn $200 million just by posting malicious links on Facebook. In this paper, we characterize malicious content posted on Facebook during 17 events, and discover that existing efforts to counter malicious content by Facebook are not able to stop all malicious content from entering the social graph. Our findings revealed that malicious entities tend to post content through web and third party applications while legitimate entities prefer mobile platforms to post content. In addition, we discovered a substantial amount of malicious content generated by Facebook pages. Through our observations, we propose an extensive feature set based on entity profile, textual content, metadata, and URL features to identify malicious content on Facebook in real time and at zero-hour. This feature set was used to train multiple machine learning models and achieved an accuracy of 86.9%. The intent is to catch malicious content that is currently evading Facebook's detection techniques. Our machine learning model was able to detect more than double the number of malicious posts as compared to existing malicious content detection techniques. Finally, we built a real world solution in the form of a REST based API and a browser plug-in to identify malicious Facebook posts in real time.

preprint2015arXiv

Hiding in Plain Sight: The Anatomy of Malicious Facebook Pages

Facebook is the world's largest Online Social Network, having more than 1 billion users. Like most other social networks, Facebook is home to various categories of hostile entities who abuse the platform by posting malicious content. In this paper, we identify and characterize Facebook pages that engage in spreading URLs pointing to malicious domains. We used the Web of Trust API to determine domain reputations of URLs published by pages, and identified 627 pages publishing untrustworthy information, misleading content, adult and child unsafe content, scams, etc. which are deemed as "Page Spam" by Facebook, and do not comply with Facebook's community standards. Our findings revealed dominant presence of politically polarized entities engaging in spreading content from untrustworthy web domains. Anger and religion were the most prominent topics in the textual content published by these pages. We found that at least 8% of all malicious pages were dedicated to promote a single malicious domain. Studying the temporal posting activity of pages revealed that malicious pages were more active than benign pages. We further identified collusive behavior within a set of malicious pages spreading adult and pornographic content. We believe our findings will enable technologists to devise efficient automated solutions to identify and curb the spread of malicious content through such pages. To the best of our knowledge, this is the first attempt in literature, focused exclusively on characterizing malicious Facebook pages.

preprint2015arXiv

TweetCred: Real-Time Credibility Assessment of Content on Twitter

During sudden onset crisis events, the presence of spam, rumors and fake content on Twitter reduces the value of information contained on its messages (or "tweets"). A possible solution to this problem is to use machine learning to automatically evaluate the credibility of a tweet, i.e. whether a person would deem the tweet believable or trustworthy. This has been often framed and studied as a supervised classification problem in an off-line (post-hoc) setting. In this paper, we present a semi-supervised ranking model for scoring tweets according to their credibility. This model is used in TweetCred, a real-time system that assigns a credibility score to tweets in a user's timeline. TweetCred, available as a browser plug-in, was installed and used by 1,127 Twitter users within a span of three months. During this period, the credibility score for about 5.4 million tweets was computed, allowing us to evaluate TweetCred in terms of response time, effectiveness and usability. To the best of our knowledge, this is the first research work to develop a real-time system for credibility on Twitter, and to evaluate it on a user base of this size.

preprint2014arXiv

@I to @Me: An Anatomy of Username Changing Behavior on Twitter

An identity of a user on an online social network (OSN) is defined by her profile, content and network attributes. OSNs allow users to change their online attributes with time, to reflect changes in their real-life. Temporal changes in users' content and network attributes have been well studied in literature, however little research has explored temporal changes in profile attributes of online users. This work makes the first attempt to study changes to a unique profile attribute of a user - username and on a popular OSN which allows users to change usernames multiple times - Twitter. We collect, monitor and analyze 8.7 million Twitter users at macroscopic level and 10,000 users at microscopic level to understand username changing behavior. We find that around 10% of monitored Twitter users opt to change usernames for possible reasons such as space gain, followers gain, and username promotion. Few users switch back to any of their past usernames, however prefer recently dropped usernames to switch back to. Users who change usernames are more active and popular than users who don't. In-degree, activity and account creation year of users are weakly correlated with their frequency of username change. We believe that past usernames of a user and their associated benefits inferred from the past, can help Twitter to suggest its users a set of suitable usernames to change to. Past usernames may also help in other applications such as searching and linking multiple OSN accounts of a user and correlating multiple Twitter profiles to a single user.

preprint2014arXiv

Analyzing Social and Stylometric Features to Identify Spear phishing Emails

Spear phishing is a complex targeted attack in which, an attacker harvests information about the victim prior to the attack. This information is then used to create sophisticated, genuine-looking attack vectors, drawing the victim to compromise confidential information. What makes spear phishing different, and more powerful than normal phishing, is this contextual information about the victim. Online social media services can be one such source for gathering vital information about an individual. In this paper, we characterize and examine a true positive dataset of spear phishing, spam, and normal phishing emails from Symantec's enterprise email scanning service. We then present a model to detect spear phishing emails sent to employees of 14 international organizations, by using social features extracted from LinkedIn. Our dataset consists of 4,742 targeted attack emails sent to 2,434 victims, and 9,353 non targeted attack emails sent to 5,912 non victims; and publicly available information from their LinkedIn profiles. We applied various machine learning algorithms to this labeled data, and achieved an overall maximum accuracy of 97.76% in identifying spear phishing emails. We used a combination of social features from LinkedIn profiles, and stylometric features extracted from email subjects, bodies, and attachments. However, we achieved a slightly better accuracy of 98.28% without the social features. Our analysis revealed that social features extracted from LinkedIn do not help in identifying spear phishing emails. To the best of our knowledge, this is one of the first attempts to make use of a combination of stylometric features extracted from emails, and social features extracted from an online social network to detect targeted spear phishing emails.

preprint2014arXiv

bit.ly/malicious: Deep Dive into Short URL based e-Crime Detection

Existence of spam URLs over emails and Online Social Media (OSM) has become a massive e-crime. To counter the dissemination of long complex URLs in emails and character limit imposed on various OSM (like Twitter), the concept of URL shortening has gained a lot of traction. URL shorteners take as input a long URL and output a short URL with the same landing page (as in the long URL) in return. With their immense popularity over time, URL shorteners have become a prime target for the attackers giving them an advantage to conceal malicious content. Bitly, a leading service among all shortening services is being exploited heavily to carry out phishing attacks, work-from-home scams, pornographic content propagation, etc. This imposes additional performance pressure on Bitly and other URL shorteners to be able to detect and take a timely action against the illegitimate content. In this study, we analyzed a dataset of 763,160 short URLs marked suspicious by Bitly in the month of October 2013. Our results reveal that Bitly is not using its claimed spam detection services very effectively. We also show how a suspicious Bitly account goes unnoticed despite of a prolonged recurrent illegitimate activity. Bitly displays a warning page on identification of suspicious links, but we observed this approach to be weak in controlling the overall propagation of spam. We also identified some short URL based features and coupled them with two domain specific features to classify a Bitly URL as malicious or benign and achieved an accuracy of 86.41%. The feature set identified can be generalized to other URL shortening services as well. To the best of our knowledge, this is the first large scale study to highlight the issues with the implementation of Bitly's spam detection policies and proposing suitable countermeasures.

preprint2014arXiv

Broker Bots: Analyzing automated activity during High Impact Events on Twitter

Twitter is now an established and a widely popular news medium. Be it normal banter or a discussion on high impact events like Boston marathon blasts, February 2014 US Icestorm, etc., people use Twitter to get updates. Twitter bots have today become very common and acceptable. People are using them to get updates about emergencies like natural disasters, terrorist strikes, etc. Twitter bots provide these users a means to perform certain tasks on Twitter that are both simple and structurally repetitive. During high impact events these Twitter bots tend to provide time critical and comprehensive information. We present how bots participate in discussions and augment them during high impact events. We identify bots in high impact events for 2013: Boston blasts, February 2014 US Icestorm, Washington Navy Yard Shooting, Oklahoma tornado, and Cyclone Phailin. We identify bots among top tweeters by getting all such accounts manually annotated. We then study their activity and present many important insights. We determine the impact bots have on information diffusion during these events and how they tend to aggregate and broker information from various sources to different users. We also analyzed their tweets, list down important differentiating features between bots and non bots (normal or human accounts) during high impact events. We also show how bots are slowly moving away from traditional API based posts towards web automation platforms like IFTTT, dlvr.it, etc. Using standard machine learning, we proposed a methodology to identify bots/non bots in real time during high impact events. This study also looks into how the bot scenario has changed by comparing data from high impact events from 2013 with data from similar type of events from 2011. Lastly, we also go through an in-depth analysis of Twitter bots who were active during 2013 Boston Marathon Blast.

preprint2014arXiv

Emerging Phishing Trends and Effectiveness of the Anti-Phishing Landing Page

Each month, more attacks are launched with the aim of making web users believe that they are communicating with a trusted entity which compels them to share their personal, financial information. Phishing costs Internet users billions of dollars every year. Researchers at Carnegie Mellon University (CMU) created an anti-phishing landing page supported by Anti-Phishing Working Group (APWG) with the aim to train users on how to prevent themselves from phishing attacks. It is used by financial institutions, phish site take down vendors, government organizations, and online merchants. When a potential victim clicks on a phishing link that has been taken down, he / she is redirected to the landing page. In this paper, we present the comparative analysis on two datasets that we obtained from APWG's landing page log files; one, from September 7, 2008 - November 11, 2009, and other from January 1, 2014 - April 30, 2014. We found that the landing page has been successful in training users against phishing. Forty six percent users clicked lesser number of phishing URLs from January 2014 to April 2014 which shows that training from the landing page helped users not to fall for phishing attacks. Our analysis shows that phishers have started to modify their techniques by creating more legitimate looking URLs and buying large number of domains to increase their activity. We observed that phishers are exploiting ICANN accredited registrars to launch their attacks even after strict surveillance. We saw that phishers are trying to exploit free subdomain registration services to carry out attacks. In this paper, we also compared the phishing e-mails used by phishers to lure victims in 2008 and 2014. We found that the phishing e-mails have changed considerably over time. Phishers have adopted new techniques like sending promotional e-mails and emotionally targeting users in clicking phishing URLs.

preprint2014arXiv

Exploration of gaps in Bitly's spam detection and relevant counter measures

Existence of spam URLs over emails and Online Social Media (OSM) has become a growing phenomenon. To counter the dissemination issues associated with long complex URLs in emails and character limit imposed on various OSM (like Twitter), the concept of URL shortening gained a lot of traction. URL shorteners take as input a long URL and give a short URL with the same landing page in return. With its immense popularity over time, it has become a prime target for the attackers giving them an advantage to conceal malicious content. Bitly, a leading service in this domain is being exploited heavily to carry out phishing attacks, work from home scams, pornographic content propagation, etc. This imposes additional performance pressure on Bitly and other URL shorteners to be able to detect and take a timely action against the illegitimate content. In this study, we analyzed a dataset marked as suspicious by Bitly in the month of October 2013 to highlight some ground issues in their spam detection mechanism. In addition, we identified some short URL based features and coupled them with two domain specific features to classify a Bitly URL as malicious / benign and achieved a maximum accuracy of 86.41%. To the best of our knowledge, this is the first large scale study to highlight the issues with Bitly's spam detection policies and proposing a suitable countermeasure.

preprint2014arXiv

Followers or Phantoms? An Anatomy of Purchased Twitter Followers

Online Social Media (OSM) is extensively used by contemporary Internet users to communicate, socialize and disseminate information. This has led to the creation of a distinct online social identity which in turn has created the need of online social reputation management techniques. A significant percentage of OSM users utilize various methods to drive and manage their reputation on OSM. This has given rise to underground markets which buy/sell fraudulent accounts, `likes', `comments' (Facebook, Instagram) and `followers' (Twitter) to artificially boost their social reputation. In this study, we present an anatomy of purchased followers on Twitter and their behaviour. We illustrate in detail the profile characteristics, content sharing and behavioural patterns of purchased follower accounts. Previous studies have analyzed the purchased follower markets and customers. Ours is the first study which analyzes the anatomy of purchased followers accounts. Some of the key insights of our study show that purchased followers have a very high unfollow entropy rate and low social engagement with their friends. In addition, we noticed that purchased follower accounts have significant difference in their interaction and content sharing patterns in comparison to random Twitter users. We also found that underground markets do not follow their service policies and guarantees they provide to customer. Our study highlights the key identifiers for suspicious follow behaviour. We then built a supervised learning mechanism to predict suspicious follower behaviour with 88.2% accuracy. We believe that understanding the anatomy and characteristics of purchased followers can help detect suspicious follower behaviour and fraudulent accounts to a larger extent.

preprint2014arXiv

Indian Premier League (IPL), Cricket, Online Social Media

In recent past online social media has played a pivotal role in sharing of information and opinions on real time events. Events in physical space are reflected in digital world through online social networks. Event based studies for such content have been widely done on Twitter in the computer science community. In this report, we performed a detailed analysis of a sports event called the Indian Premier League (IPL'13) for both Facebook and Twitter. IPL is the most popular cricket league in India with players from across the world. We analysed more than 2.6 million tweets and 700 thousand Facebook posts for temporal activity, text quality, geography of users and the spot-fixing scandal which came up during the league. We were able to draw strong correlations between the brand value of teams and how much they were talked about on social media across Facebook and Twitter. Analysis of geo-tagged data showed major activity from metropolitan suburbs however activity was not restricted to the regions geographically associated with each team. We present a decay calculation methodology, using which we derive that activity died down on both Twitter and Facebook in a very similar manner. Such analysis can be used to model events and study their penetration in social networks. We analysed text for spot-fixing and found that user response to allegations about matches being fixed was cold. The complete analysis presented in this report, can be particularly useful for studying events involving crisis or events of political importance having similar structure.

preprint2014arXiv

It Doesn't Break Just on Twitter. Characterizing Facebook content During Real World Events

Multiple studies in the past have analyzed the role and dynamics of the Twitter social network during real world events. However, little work has explored the content of other social media services, or compared content across two networks during real world events. We believe that social media platforms like Facebook also play a vital role in disseminating information on the Internet during real world events. In this work, we study and characterize the content posted on the world's biggest social network, Facebook, and present a comparative analysis of Facebook and Twitter content posted during 16 real world events. Contrary to existing notion that Facebook is used mostly as a private network, our findings reveal that more than 30% of public content that was present on Facebook during these events, was also present on Twitter. We then performed qualitative analysis on the content spread by the most active users during these events, and found that over 10% of the most active users on both networks post spam content. We used stylometric features from Facebook posts and tweet text to classify this spam content, and were able to achieve an accuracy of over 99% for Facebook, and over 98% for Twitter. This work is aimed at providing researchers with an overview of Facebook content during real world events, and serve as basis for more in-depth exploration of its various aspects like information quality, and credibility during real world events.

preprint2014arXiv

Online Social Media and Police in India: Behavior, Perceptions, Challenges

Police agencies across the globe are increasingly using Online Social Media (OSM) to acquire intelligence and connect with citizens. Developed nations have well thought of strategies to use OSM for policing. However, developing nations like India are exploring and evolving OSM as a policing solution. India, in recent years, experienced many events where rumors and fake content on OSM instigated communal violence. In contrast to traditional media (e.g. television and print media) used by Indian police departments, OSM offers velocity, variety, veracity and large volume of information. These introduce new challenges for police like platforms selection, secure usage strategy, developing trust, handling offensive comments, and security / privacy implication of information shared through OSM. Success of police initiatives on OSM to maintain law and order depends both on their understanding of OSM and citizen's acceptance / participation on these platforms. This study provides multidimensional understanding of behavior, perceptions, interactions, and expectation regarding policing through OSM. First, we examined recent updates from four different police pages- Delhi, Bangalore, Uttar Pradesh and Chennai to comprehend various dimensions of police interaction with citizens on OSM. Second, we conducted 20 interviews with IPS officers (Indian Police Service) and 17 interviews with citizens to understand decision rationales and expectation gaps between two stakeholders (police and citizens); this was followed up with 445 policemen surveys and 204 citizen surveys. We also present differences between police expectations of Indian and police departments in developed countries.

preprint2014arXiv

Privacy4ICTD in India: Exploring Perceptions, Attitudes and Awareness about ICT Use

Several ICT studies give anecdotal evidences showing privacy to be an area of concern that can influence adoption of technology in the developing world. However, in-depth understanding of end users' privacy attitudes and awareness is largely unexplored in developing countries such as India. We conducted a survey with 10,427 Indian citizens to bring forth various insights on privacy expectations and perceptions of end users. Our study explores end-users' privacy expectations on three ICT platforms - mobile phones, OSN (Online Social Network), and government projects. Our results, though preliminary, show that users disproportionately consider financial details as personal information in comparison to medical records. Users heavily use mobile phones to store personal details and show high trust in mobile service providers for protecting the private data. However, users show concerns that mobile service provider may allow improper access of their personal information to third parties and government. We also find that female participants in the study were marginally more conscious of their privacy than males. To the best of our knowledge, this work presents the largest privacy study which benchmarks privacy perceptions among Indian citizens. Understanding users' privacy perceptions can help improve technology adoption and develop policies and laws for improving technology experience and enabling development for a better life in India.

preprint2014arXiv

What They Do in Shadows: Twitter Underground Follower Market

Internet users and businesses are increasingly using online social networks (OSN) to drive audience traffic and increase their popularity. In order to boost social presence, OSN users need to increase the visibility and reach of their online profile, like - Facebook likes, Twitter followers, Instagram comments and Yelp reviews. For example, an increase in Twitter followers not only improves the audience reach of the user but also boosts the perceived social reputation and popularity. This has led to a scope for an underground market that provides followers, likes, comments, etc. via a network of fraudulent and compromised accounts and various collusion techniques. In this paper, we landscape the underground markets that provide Twitter followers by studying their basic building blocks - merchants, customers and phony followers. We charecterize the services provided by merchants to understand their operational structure and market hierarchy. Twitter underground markets can operationalize using a premium monetary scheme or other incentivized freemium schemes. We find out that freemium market has an oligopoly structure with few merchants being the market leaders. We also show that merchant popularity does not have any correlation with the quality of service provided by the merchant to its customers. Our findings also shed light on the characteristics and quality of market customers and the phony followers provided. We draw comparison between legitimate users and phony followers, and find out key identifiers to separate such users. With the help of these differentiating features, we build a supervised learning model to predict suspicious following behaviour with an accuracy of 89.2%.

preprint2013arXiv

Call Me MayBe: Understanding Nature and Risks of Sharing Mobile Numbers on Online Social Networks

There is a great concern about the potential for people to leak private information on OSNs, but few quantitative studies on this. This research explores the activity of sharing mobile numbers on OSNs, via public profiles and posts. We attempt to understand the characteristics and risks of mobile numbers sharing behaviour on OSNs and focus on Indian mobile numbers. We collected 76,347 unique mobile numbers posted by 85905 users on Twitter and Facebook and analysed 2997 numbers, prefixed with +91. We observed, most users shared their own mobile numbers to spread urgent information; and to market products and escort business. Fewer female users shared mobile numbers on OSNs. Users utilized other OSN platforms and third party applications like Twitterfeed, to post mobile numbers on multiple OSNs. In contrast to the user's perception of numbers spreading quickly on OSN, we observed that except for emergency, most numbers did not diffuse deep. To assess risks associated with mobile numbers exposed on OSNs, we used numbers to gain sensitive information about their owners (e.g. name, Voter ID) by collating publicly available data from OSNs, Truecaller, OCEAN. On using the numbers on WhatApp, we obtained a myriad of sensitive details (relationship status, BBM pins) of the number owner. We communicated the observed risks to the owners by calling. Few users were surprised to know about the online presence of their number, while a few others intentionally posted it online for business purposes. We observed, 38.3% of users who were unaware of the online presence of their number have posted their number themselves on the social network. With these observations, we highlight that there is a need to monitor leakage of mobile numbers via profile and public posts. To the best of our knowledge, this is the first exploratory study to critically investigate the exposure of Indian mobile numbers on OSNs.

preprint2013arXiv

ChaMAILeon: Exploring the Usability of a Privacy Preserving Email Sharing System

While passwords, by definition, are meant to be secret, recent trends have witnessed an increasing number of people sharing their email passwords with friends, colleagues, and significant others. However, leading websites like Google advise their users not to share their passwords with anyone, to avoid security and privacy breaches. To understand users' general password sharing behavior and practices, we conducted an online survey with 209 Indian participants and found that 64.35% of the participants felt a need to share their email passwords. Further, about 77% of the participants said that they would want to use a system which could provide them access control features, to maintain their privacy while sharing emails. To address the privacy concerns of users who need to share emails, we propose ChaMAILeon, a system which enables users to share their email passwords while maintaining their privacy. ChaMAILeon allows users to create multiple passwords for their email account. Each such password corresponds to a different set of access control rules, and gives a different view of the same email account. We conducted a controlled experiment with 30 participants to evaluate the usability of the system. Each participant was required to perform 5 tasks. Each task corresponded to different access control rules, which the participant was required to set, for a dummy email account. We found that, with a reasonable number of multiple attempts, all 30 participants were able to perform all 5 tasks given to them. The system usability score was found out to be 75.42. Moreover, 56.6% of the participants said that they would like to use ChaMAILeon frequently.

preprint2013arXiv

ChaMAILeon: Simplified email sharing like never before!

While passwords, by definition, are meant to be secret, recent trends in the Internet usage have witnessed an increasing number of people sharing their email passwords for both personal and professional purposes. As sharing passwords increases the chances of your passwords being compromised, leading websites like Google strongly advise their users not to share their passwords with anyone. To cater to this conflict of usability versus security and privacy, we introduce ChaMAILeon, an experimental service, which allows users to share their email passwords while maintaining their privacy and not compromising their security. In this report, we discuss the technical details of the implementation of ChaMAILeon.

preprint2013arXiv

Cross-Pollination of Information in Online Social Media: A Case Study on Popular Social Networks

Owing to the popularity of Online Social Media (OSM), Internet users share a lot of information (including personal) on and across OSM services every day. For example, it is common to find a YouTube video embedded in a blog post with an option to share the link on Facebook. Users recommend, comment, and forward information they receive from friends, contributing in spreading the information in and across OSM services. We term this information diffusion process from one OSM service to another as Cross-Pollination, and the network formed by users who participate in Cross-Pollination and content produced in the network as \emph{Cross-Pollinated network}. Research has been done about information diffusion within one OSM service, but little is known about Cross-Pollination. In this paper, we aim at filling this gap by studying how information (video, photo, location) from three popular OSM services (YouTube, Flickr and Foursquare) diffuses on Twitter, the most popular microblogging service. Our results show that Cross-Pollinated networks follow temporal and topological characteristics of the diffusion OSM (Twitter in our study). Furthermore, popularity of information on source OSM (YouTube, Flickr and Foursquare) does not imply its popularity on Twitter. Our results also show that Cross-Pollination helps Twitter in terms of traffic generation and user involvement, but only a small fraction of videos and photos gain a significant number of views from Twitter. We believe this is the first research work which explicitly characterizes the diffusion of information across different OSM services.

preprint2013arXiv

Limited Attention and Centrality in Social Networks

How does one find important or influential people in an online social network? Researchers have proposed a variety of centrality measures to identify individuals that are, for example, often visited by a random walk, infected in an epidemic, or receive many messages from friends. Recent research suggests that a social media users' capacity to respond to an incoming message is constrained by their finite attention, which they divide over all incoming information, i.e., information sent by users they follow. We propose a new measure of centrality --- limited-attention version of Bonacich's Alpha-centrality --- that models the effect of limited attention on epidemic diffusion. The new measure describes a process in which nodes broadcast messages to their out-neighbors, but the neighbors' ability to receive the message depends on the number of in-neighbors they have. We evaluate the proposed measure on real-world online social networks and show that it can better reproduce an empirical influence ranking of users than other popular centrality measures.

preprint2013arXiv

OCEAN: Open-source Collation of eGovernment data And Networks - Understanding Privacy Leaks in Open Government Data

The awareness and sense of privacy has increased in the minds of people over the past few years. Earlier, people were not very restrictive in sharing their personal information, but now they are more cautious in sharing it with strangers, either in person or online. With such privacy expectations and attitude of people, it is difficult to embrace the fact that a lot of information is publicly available on the web. Information portals in the form of the e-governance websites run by Delhi Government in India provide access to such PII without any anonymization. Several databases e.g., Voterrolls, Driving Licence number, MTNL phone directory, PAN card serve as repositories of personal information of Delhi residents. This large amount of available personal information can be exploited due to the absence of proper written law on privacy in India. PII can also be collected from various social networking sites like Facebook, Twitter, GooglePlus etc. where the users share some information about them. Since users themselves put this information, it may not be considered as a privacy breach, but if the information is aggregated, it may give out much more information resulting in a bigger threat. To bring such issues to public notice, we developed Open-source Collation of eGovernment data And Networks (OCEAN), a system where the user enters little information (e.g. Name) about a person and gets large amount of personal information about him / her like name, age, address, date of birth, mother's name, father's name, voter ID, driving licence number, PAN.

preprint2013arXiv

PhishAri: Automatic Realtime Phishing Detection on Twitter

With the advent of online social media, phishers have started using social networks like Twitter, Facebook, and Foursquare to spread phishing scams. Twitter is an immensely popular micro-blogging network where people post short messages of 140 characters called tweets. It has over 100 million active users who post about 200 million tweets everyday. Phishers have started using Twitter as a medium to spread phishing because of this vast information dissemination. Further, it is difficult to detect phishing on Twitter unlike emails because of the quick spread of phishing links in the network, short size of the content, and use of URL obfuscation to shorten the URL. Our technique, PhishAri, detects phishing on Twitter in realtime. We use Twitter specific features along with URL features to detect whether a tweet posted with a URL is phishing or not. Some of the Twitter specific features we use are tweet content and its characteristics like length, hashtags, and mentions. Other Twitter features used are the characteristics of the Twitter user posting the tweet such as age of the account, number of tweets, and the follower-followee ratio. These Twitter specific features coupled with URL based features prove to be a strong mechanism to detect phishing tweets. We use machine learning classification techniques and detect phishing tweets with an accuracy of 92.52%. We have deployed our system for end-users by providing an easy to use Chrome browser extension which works in realtime and classifies a tweet as phishing or safe. We show that we are able to detect phishing tweets at zero hour with high accuracy which is much faster than public blacklists and as well as Twitter's own defense mechanism to detect malicious content. To the best of our knowledge, this is the first realtime, comprehensive and usable system to detect phishing on Twitter.

preprint2013arXiv

Studying User Footprints in Different Online Social Networks

With the growing popularity and usage of online social media services, people now have accounts (some times several) on multiple and diverse services like Facebook, LinkedIn, Twitter and YouTube. Publicly available information can be used to create a digital footprint of any user using these social media services. Generating such digital footprints can be very useful for personalization, profile management, detecting malicious behavior of users. A very important application of analyzing users' online digital footprints is to protect users from potential privacy and security risks arising from the huge publicly available user information. We extracted information about user identities on different social networks through Social Graph API, FriendFeed, and Profilactic; we collated our own dataset to create the digital footprints of the users. We used username, display name, description, location, profile image, and number of connections to generate the digital footprints of the user. We applied context specific techniques (e.g. Jaro Winkler similarity, Wordnet based ontologies) to measure the similarity of the user profiles on different social networks. We specifically focused on Twitter and LinkedIn. In this paper, we present the analysis and results from applying automated classifiers for disambiguating profiles belonging to the same user from different social networks. UserID and Name were found to be the most discriminative features for disambiguating user profiles. Using the most promising set of features and similarity metrics, we achieved accuracy, precision and recall of 98%, 99%, and 96%, respectively.

preprint2013arXiv

The Pin-Bang Theory: Discovering The Pinterest World

Pinterest is an image-based online social network, which was launched in the year 2010 and has gained a lot of traction, ever since. Within 3 years, Pinterest has attained 48.7 million unique users. This stupendous growth makes it interesting to study Pinterest, and gives rise to multiple questions about it's users, and content. We characterized Pinterest on the basis of large scale crawls of 3.3 million user profiles, and 58.8 million pins. In particular, we explored various attributes of users, pins, boards, pin sources, and user locations, in detail and performed topical analysis of user generated textual content. The characterization revealed most prominent topics among users and pins, top image sources, and geographical distribution of users on Pinterest. We then investigated this social network from a privacy and security standpoint, and found traces of malware in the form of pin sources. Instances of Personally Identifiable Information (PII) leakage were also discovered in the form of phone numbers, BBM (Blackberry Messenger) pins, and email addresses. Further, our analysis demonstrated how Pinterest is a potential venue for copyright infringement, by showing that almost half of the images shared on Pinterest go uncredited. To the best of our knowledge, this is the first attempt to characterize Pinterest at such a large scale.

preprint2013arXiv

Three-Way Dissection of a Game-CAPTCHA: Automated Attacks, Relay Attacks, and Usability

Existing captcha solutions on the Internet are a major source of user frustration. Game captchas are an interesting and, to date, little-studied approach claiming to make captcha solving a fun activity for the users. One broad form of such captchas -- called Dynamic Cognitive Game (DCG) captchas -- challenge the user to perform a game-like cognitive task interacting with a series of dynamic images. We pursue a comprehensive analysis of a representative category of DCG captchas. We formalize, design and implement such captchas, and dissect them across: (1) fully automated attacks, (2) human-solver relay attacks, and (3) usability. Our results suggest that the studied DCG captchas exhibit high usability and, unlike other known captchas, offer some resistance to relay attacks, but they are also vulnerable to our novel dictionary-based automated attack.

preprint2012arXiv

Characterizing Pedophile Conversations on the Internet using Online Grooming

Cyber-crime targeting children such as online pedophile activity are a major and a growing concern to society. A deep understanding of predatory chat conversations on the Internet has implications in designing effective solutions to automatically identify malicious conversations from regular conversations. We believe that a deeper understanding of the pedophile conversation can result in more sophisticated and robust surveillance systems than majority of the current systems relying only on shallow processing such as simple word-counting or key-word spotting. In this paper, we study pedophile conversations from the perspective of online grooming theory and perform a series of linguistic-based empirical analysis on several pedophile chat conversations to gain useful insights and patterns. We manually annotated 75 pedophile chat conversations with six stages of online grooming and test several hypothesis on it. The results of our experiments reveal that relationship forming is the most dominant online grooming stage in contrast to the sexual stage. We use a widely used word-counting program (LIWC) to create psycho-linguistic profiles for each of the six online grooming stages to discover interesting textual patterns useful to improve our understanding of the online pedophile phenomenon. Furthermore, we present empirical results that throw light on various aspects of a pedophile conversation such as probability of state transitions from one stage to another, distribution of a pedophile chat conversation across various online grooming stages and correlations between pre-defined word categories and online grooming stages.

preprint2012arXiv

Finding Nemo: Searching and Resolving Identities of Users Across Online Social Networks

An online user joins multiple social networks in order to enjoy different services. On each joined social network, she creates an identity and constitutes its three major dimensions namely profile, content and connection network. She largely governs her identity formulation on any social network and therefore can manipulate multiple aspects of it. With no global identifier to mark her presence uniquely in the online domain, her online identities remain unlinked, isolated and difficult to search. Earlier research has explored the above mentioned dimensions, to search and link her multiple identities with an assumption that the considered dimensions have been least disturbed across her identities. However, majority of the approaches are restricted to exploitation of one or two dimensions. We make a first attempt to deploy an integrated system (Finding Nemo) which uses all the three dimensions of an identity to search for a user on multiple social networks. The system exploits a known identity on one social network to search for her identities on other social networks. We test our system on two most popular and distinct social networks - Twitter and Facebook. We show that the integrated system gives better accuracy than the individual algorithms. We report experimental findings in the report.