Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileAcademic lineage
Possible advisorsa guess from early papers, not confirmed
- Sivaji BandyopadhyaySuggested from co-authorship
- Animesh MukherjeeSuggested from co-authorship
Is this you? Claim this profile to confirm or dismiss it.
Works384 from public data
- 556
- Factuality challenges in the era of large language models and opportunities for fact-checking319
By analysing the factuality constraints within LLMs and their impact on fact-checking, this work aims to contribute to a path towards maintaining accuracy at a time of confluence of generative artificial intelligence and misinformation.
- Metrics for Community Analysis300
A survey of the start-of-the-art metrics used for the detection and the evaluation of community structure and a comparative analysis of these metrics in measuring the goodness of the underlying community structure is presented.
- SpotFake+: A Multimodal Framework for Fake News Detection via Transfer Learning (Student Abstract)286
This work presents SpotFake+, a multimodal approach that leverages transfer learning to capture semantic and contextual information from the news articles and its associated images and achieves the better accuracy for fake news detection.
- SemEval-2020 Task 8: Memotion Analysis- the Visuo-Lingual Metaphor!255
The objective of this proposal is to bring the attention of the research community towards the automatic processing of Internet memes by classifying intensity of meme emotion.
- MOMENTA: A Multimodal Framework for Detecting Harmful Memes and Their Targets231
This work proposes MOMENTA (MultimOdal framework for detecting harmful memes, a novel multimodal deep neural network that uses global and local perspectives to detect harmful memes and outperforms several strong rivaling approaches.
- SemEval-2020 Task 9: Overview of Sentiment Analysis of Code-Mixed Tweets208
The SemEval-2020 Task 9 on Sentiment Analysis of Code-Mixed Tweets (SentiMix 2020) is presented and it is observed that BERT-like models and ensemble methods are the most common and successful approaches among the participants.
- Detecting Harmful Memes and Their Targets202
This work presents HarMeme, the first benchmark dataset, containing 3,544 memes related to COVID-19, and proposes two novel problem formulations: detecting harmful memes and the social entities that these harmful memes target.
- Cross-SEAN: A cross-stitch semi-supervised neural attention model for COVID-19 fake news detection154
CTF, a large-scale COVID-19 Twitter dataset with labelled genuine and fake tweets is introduced and Cross-SEAN, a cross-stitch based semi-supervised end-to-end neural attention model which leverages the large amount of unlabelled data is proposed.
- Handling Bias in Toxic Speech Detection: A Survey145
A systematic study of the limitations and challenges of existing methods for mitigating bias in toxicity detection and a case study to introduce the concept of bias shift due to knowledge-based bias mitigation are conducted.
- On the permanence of vertices in network communities139
Compared to other metrics, permanence provides a more accurate estimate of a derived community structure to the ground-truth community and is more sensitive to perturbations in the network.
- We Used Neural Networks to Detect Clickbaits: You Won’t Believe What Happened Next!130
A neural network architecture based on Recurrent Neural Networks for detecting clickbaits is introduced, which relies on distributed word representations learned from a large unannotated corpora, and character embeddings learned via Convolutional Neural Networks.
- 116
- Overview of CONSTRAINT 2021 Shared Tasks: Detecting English COVID-19 Fake News and Hindi Hostile Posts111
The findings of the shared tasks conducted at the CONSTRAINT Workshop at AAAI 2021 are presented and the most successful models were BERT or its variations.
- Multi-Modal Sarcasm Detection and Humor Classification in Code-Mixed Conversations109
A Hindi-English code-mixed dataset is developed for the multi-modal sarcasm detection and humor classification in conversational dialog, and a novel attention-rich neural architecture for the utterance classification is proposed.
- 109
- 103
- POSIX: A Prompt Sensitivity Index For Large Language Models92
POSIX - a novel PrOmpt Sensitivity IndeX as a reliable measure of prompt sensitivity is proposed, thereby offering a more comprehensive evaluation of LLM performance and is used to measure and thereby compare prompt sensitivity of various open-source LLMs.
- Unsupervised Fake News Detection77
GTUT, a graph-based approach for unsupervised fake news detection, is developed which operates in three phases and establishes the improved effectiveness of the method over state-of-the-art techniques for the task.
- Speaker and Time-aware Joint Contextual Learning for Dialogue-act Classification in Counselling Conversations71
This work develops a novel dataset, named HOPE, to provide a platform for the dialogue-act classification in counselling conversations, identifies the requirement of such conversation, and proposes SPARTA, a transformer-based architecture with a novel speaker- and time-aware contextual learning for the Dialogue-Act classification.
- 63
- How to think step-by-step: A mechanistic understanding of chain-of-thought reasoning61
This work investigates the neural sub-structures within LLMs that manifest CoT reasoning from a mechanistic point of view, and demonstrates that LLMs deploy multiple parallel pathways of answer generation for step-by-step reasoning.
- Recent Advances in Online Hate Speech Moderation: Multimodality and the Role of Large Models58
This comprehensive survey delves into the recent strides in HS moderation, spotlighting the burgeoning role of large language models (LLMs) and large multimodal models (LMMs) and the development of more nuanced, context-aware systems.
- Exploitation of Social Media for Emergency Relief and Preparedness: Recent Research and Trends58
A special issue on “Exploitation of Social Media for Emergency Relief and Preparedness” was conducted for the journal Information Systems Frontiers to present a platform for dissemination of the empirical results of various technologies for extracting vital and actionable information from social media content in disaster situations.
- 56
- 55
- 54
- Nice Perfume. How Long Did You Marinate in It? Multimodal Sarcasm Explanation53
This paper proposes a novel problem -- Multimodal Sarcasm Explanation (MuSE) -- given a multimodal sarcastic post containing an image and a caption, it aims to generate a natural language explanation to reveal the intended sarcasm.
- Hate is the New Infodemic: A Topic-aware Modeling of Hate Speech Diffusion on Twitter52
This work crawls a large-scale dataset of tweets, retweets, user activity history, and follower networks, comprising over 161 million tweets from more than 41 million unique users, and proposes RETINA, a novel neural architecture that incorporates exogenous influence using scaled dot-product attention to predict the retweet dynamics of hateful content.
- Model-Agnostic Meta-Learning for Multilingual Hate Speech Detection51
HateMAML is proposed, a model-agnostic meta-learning (MAML)-based framework that effectively performs hate speech detection in low-resource languages and outperforms the state-of-the-art baselines by more than 3% in the cross-domain multilingual transfer setting.
- 51
- See, hear, read: Leveraging multimodality with guided attention for abstractive text summarization50
This paper introduces AVIATE, the first large-scale dataset for abstractive text summarization with videos of diverse duration, compiled from presentations in well-known academic conferences like NDSS, ICML, NeurIPS, etc, and proposes a factorized multi-modal Transformer based decoder-only language model, which performs significantly better than traditional encoder-decoder based networks.
- Hide and Seek: Outwitting Community Detection Algorithms50
Network deception using permanence loss (NEURAL), a novel method that greedily optimizes a node-centric objective function to determine the rewiring strategy, is introduced and captures important metainformation about edges that otherwise could not be inferred by observing only their topological structures.
- Nurse is Closer to Woman than Surgeon? Mitigating Gender-Biased Proximities in Word Embeddings49
RAN-Debias is proposed, a novel gender debiasing methodology that not only eliminates the bias present in a word vector but also alters the spatial distribution of its neighboring vectors, achieving a bias-free setting while maintaining minimal semantic offset.
- HawkesEye: Detecting Fake Retweeters Using Hawkes Process and Topic Modeling48
This work introduces an up-to-date, temporally diverse, trend-oriented labeled dataset, and develops a novel classifier, called HawkesEye, which makes predictions based on a temporal window, in contrast to existing approaches which require a graph-like relationship between tweet entities, or the presence of the entire retweeting timeline of a retweeter.
- Constant Communities in Complex Networks48
This work identifies and study the properties of invariant groups of vertices (constant communities) whose assignment to communities are, quite remarkably, not affected by vertex ordering.
- Identification of Reduplication in Bengali Corpus and their Semantic Analysis: A Rule Based Approach47
Rule-based approach is used to identify reduplications from the Bengali corpus of the articles of Rabindranath Tagore and their structural and semantics classifications are analyzed.
- Changing views: Persuasion modeling and argument extraction from online discussions45
This work proposes a deep LSTM model to classify whether a conversation leads to a successful persuasion or not, and uses this model to predict whether a certain chain of arguments can lead to persuasion.
- Incorporating Network Embedding into Markov Random Field for Better Community Detection44
This work proposes a general Markov Random Field (MRF) framework to incorporate coupling in network embedding which allows better detecting network communities, and improves the accuracy of existing embedding methods, and corrects most wrongly-divided statistically-significant nodes.
- Role of interdisciplinarity in computer sciences: quantification, impact and life trajectory44
How the extent of interdisciplinarity in computer science domain has changed over the last 50 years is investigated and general benchmarks for future studies of interdisciplinary research activities in other domains of science are provided.
- 43
- Retweet Us, We will Retweet You: Spotting Collusive Retweeters Involved in Blackmarket Services43
A first attempt to investigate blackmarket customers engaged in producing fake retweets of Twitter users using state-of-the-art supervised models, and designs a browser extension which spots its fake retweeters in real-time.
- On the categorization of scientific citation profiles in computer science42
A new dynamic growth model reveals how citation networks evolve over time, pointing the way toward reformulated scientometrics.
- Discovering emotion and reasoning its flip in multi-party conversations using masked memory network and transformer41
A novel task – Emotion Flip Reasoning (EFR) that aims to identify past utterances which have triggered one’s emotion state to flip at a certain time and a masked memory network to address the former and a Transformer-based network for the latter task.
- Spotting Collective Behaviour of Online Frauds in Customer Reviews41
DeFrauder is proposed, an unsupervised method to detect online fraud reviewer groups that first detects candidate fraud groups by leveraging the underlying product review graph and incorporating several behavioral signals which model multi-faceted collaboration among reviewers.
- Ensemble-based overlapping community detection using disjoint community structures41
This work proposes an ensemble-based approach, called EnCoD, that leverages the solutions produced by various disjoint community detection algorithms to discover the overlapping community structure and shows that it is generic enough to be applied to networks where the vertices are associated with explicit semantic features.
- Collective Classification of Spam Campaigners on Twitter: A Hierarchical Meta-Path Based Approach41
This paper proposes a simple yet effective metric, called Hierarchical Meta-Path Score (HMPS), to measure the proximity of an unknown user to the other known pool of spammers, and designs a feedback-based active learning strategy that significantly outperforms three state-of-the-art baselines for the task of spam detection.
- DyPerm: Maximizing Permanence for Dynamic Community Detection40
DyPerm is the first dynamic community detection method which optimizes a novel community scoring metric, called permanence, and turns out to be 15 times faster than its static counterpart.
- 40
- Emotion-Aware Multimodal Fusion for Meme Emotion Detection38
A novel multimodal neural framework that explicitly models emotion-enriched visual cues, and employs an efficient cross-modal fusion via a gating mechanism that competes strongly with previous best approaches on the challenging meme analysis task.
- 36
- LESA: Linguistic Encapsulation and Semantic Amalgamation Based Generalised Claim Detection from Online Content36
LESA is proposed, a framework which aims at advancing headfirst into expunging the issue of labeled unstructured text for experimentation by assembling a source-independent generalized model that captures syntactic features through part-of-speech and dependency embeddings, as well as contextual features through a fine-tuned language model.
- EC3: Combining Clustering and Classification for Ensemble Learning36
This work proposes EC3, a novel algorithm that merges classification and clustering together in order to support both binary and multi-class classification and shows that its methods are faster, more resilient to noise and class imbalance than the best baseline method.
- GenPerm: A Unified Method for Detecting Non-Overlapping and Overlapping Communities36
This work presents a generalized framework that can identify both non-overlapping and overlapping communities, without any prior input about the network or its community distribution, using a vertex-based metric, GenPerm, that quantifies by how much a vertex belongs to each of its constituent communities.
- Proactively Reducing the Hate Intensity of Online Posts via Hate Speech Normalization35
NACL, a simple yet efficient hate speech normalization model that operates in three stages - first, it measures the hate intensity of the original sample; second, it identifies the hate span(s) within it; and finally, it reduces hate intensity by paraphrasing the hate spans.
- Hatemongers ride on echo chambers to escalate hate speech diffusion34
This study analyzes more than 32 million posts from over 6.8 million users across three popular online social networks to investigate the interrelations between hateful behavior, information dissemination, and polarized organization mediated by echo chambers and finds that hatemongers play a more crucial role in governing the spread of information.
- Sleeping beauties in Computer Science: characterization and early identification33
This work identifies more than 5,000 sleeping beauties in Computer Science, and characterise them based on their sub-field and their citation profile after awakening, and proposes and develops a machine learning-based classification approach that attempts to classify publications based on whether they are likely to be SBs.
- 32
- Findings of the CONSTRAINT 2022 Shared Task on Detecting the Hero, the Villain, and the Victim in Memes32
The task aims to delve deeper into the domain of meme comprehension by deciphering the connotations behind the entities present in a meme by determining the victimizing, glorifying, and vilifying intentions embedded in meme entities to explicate their connotations.
- CoReRank32
CoReRank is the first unsupervised method to detect collusive users and suspicious tweets simultaneously with theoretical guarantees and is guaranteed to converge in a finite number of iterations and has linear time complexity.
- 32
- FeRoSA: A Faceted Recommendation System for Scientific Articles32
A framework of faceted recommendation for scientific articles abbreviated as FeRoSA which apart from ensuring quality retrieval of scientific articles for a query paper, also efficiently arranges the recommended papers into different facets categories is proposed.
- Leveraging disjoint communities for detecting overlapping community structure31
Permanence based Vertex-replication algorithm for Overlapping Community detection, a new post-processing technique that by combining with any existing disjoint community detection algorithm, can suitably process each vertex using a new vertex-based metric, called permanence, and thereby finds out overlapping candidates with their community memberships.
- When did you become so smart, oh wise one?! Sarcasm Explanation in Multi-modal Multi-party Dialogues30
- Counseling Summarization Using Mental Health Knowledge Guided Utterance Filtering30
The aim is mental health counseling summarization to build upon domain knowledge and to help clinicians quickly glean meaning in the real world and to propose ConSum, a novel counseling-component guided summarization model.
- Neural Abstractive Summarization with Structural Attention30
This work presents a hierarchical encoder based on structural attention to model such inter-sentence and inter-document dependencies within the sentences of a document and shows that the proposed model achieves significant improvement over the baseline in both single and multi-document summarization settings.
- 29
- Would Your Tweet Invoke Hate on the Fly? Forecasting Hate Intensity of Reply Threads on Twitter29
DESSERT, a novel deep state-space model that leverages the function approximation capability of deep neural networks with the capacity to quantify the uncertainty of statistical signal processing models is proposed and deployed in an advanced AI platform designed to monitor real-world problematic hateful content.
- Understanding and modeling diverse scientific careers of researchers29
This paper analyzes the diverse scientific careers of researchers in order to understand the key factors that could lead to a successful career and proposes two entropy-based metrics to measure a researcher's choice of research topics.
- 27
- A Comprehensive Understanding of Code-Mixed Language Semantics Using Hierarchical Transformer26
A hierarchical transformer (HIT)-based architecture to learn the semantics of code-mixed languages and demonstrates the generalizability of the HIT architecture using masked language modeling (MLM)-based pretraining, zero-shot learning (ZSL), and transfer learning approaches.
- 26
- 26
- 26
- The Role Of Citation Context In Predicting Long-Term Citation Profiles: An Experimental Study Based On A Massive Bibliographic Text Dataset26
Analyzing a massive dataset of nearly 1.5 million computer science articles and more than 26 million citation contexts, it is shown that average countX and average citeWords discriminate between various citation ranges as well as citation categories.
- Exercise? I thought you said 'Extra Fries’: Leveraging Sentence Demarcations and Multi-hop Attention for Meme Affect Analysis25
A multi-hop attention-based deep neural network framework, called MHA-Meme, whose prime objective is to leverage the spatial-domain correspondence between the visual modality and various textual segments to extract fine-grained feature representations for classification, which outperforms baselines in all the tasks on average.
- 25
- Detecting and Characterizing Extremist Reviewer Groups in Online Product Reviews25
Surprisingly, there are a lot of verified reviewers showing extreme sentiment on Amazon, which, on further investigation, leads to ways to circumvent the existing mechanisms in place to prevent unofficial incentives on Amazon.
- 24
- 24
- Heterogeneous Edge Embedding for Friend Recommendation24
This work proposes a method to mine network representation that effectively exploits edge heterogeneity in multi-graphs and outperforms various state-of-the-art baselines on Hike’s social network in terms of accuracy metrics as well as user satisfaction.
- Blackmarket-Driven Collusion Among Retweeters–Analysis, Detection, and Characterization24
This paper is the first attempt to provide a thorough study of the collusive users involved in two types of blackmarket services – Premium and Freemium, and provides network-centric, profile- centric, timeline-centric and retweet-centric characteristics of these users and shows how usersinvolved in premium black market services exhibit diverse behavior as compared to those involved in freemium services.
- CQASUMM: Building References for Community Question Answering Summarization Corpora24
OdinioSumm is introduced, a new MDS which outperforms the best baseline by 4.6% w.r.t ROUGE-1 score and is compared to the popular MDS techniques and evaluated their performance on the CQA corpora.
- Determining Sentiment in Citation Text and Analyzing Its Impact on the Proposed Ranking Index24
A new index is proposed which takes into account both the quantitative and qualitative factors while scoring a paper and a ranking of research papers based on the M-index is developed.
- 23
- MalReG: Detecting and Analyzing Malicious Retweeter Groups23
MalReG, a novel algorithm to detect retweeter groups, and a set of 23 group-based features (entropy-based and temporal-based) to train a supervised model to identify malicious retweeters groups (MRG), are proposed and discovered.
- Detecting and Analyzing Collusive Entities on YouTube22
An in-depth analysis of collusive entities on YouTube fostered by various blackmarket services is provided and CollATe, a novel end-to-end neural architecture that leverages time-series information of posted comments along with static metadata of videos, is proposed.
- Permanence and Community Structure in Complex Networks22
It is argued that the belongingness of nodes in a community is not uniform, and the degree of belongingness is quantified by a new vertex-based metric called permanence, which is an excellent metric for identifying communities.
- Hybrid quantum repeaters with ensemble-based quantum memories and single-spin photon transducers21
This work outlines how ensemble-based quantum memories combined with single-spin photon transducers, which can transfer quantum information between a photon and a single spin, can facilitate massive multiplexing, efficient photon generation, and quantum logic for amplifying communication rates.
- AI-generated content in cross-domain applications: Research trends, challenges and propositions21
This paper brings together 16 scholars from multiple disciplines to provide a cross-domain perspective on the trends and challenges of AIGC, providing insights into its current research trends, ongoing challenges, and future directions.
- Towards privacy-aware mental health AI models21
The authors examine privacy risks in mental health AI, and explore solutions and evaluation frameworks to balance privacy–utility trade-offs, and suggest a pipeline for developing privacy-aware mental health AI systems.
- Deep Exogenous and Endogenous Influence Combination for Social Chatter Intensity Prediction21
This work proposes a novel framework, ChatterNet, which is the first that can model and predict user engagement without considering the underlying user network, and shows considerable improvements beyond recent state-of-the-art models of engagement prediction.
- Citation interactions among computer science fields: a quantitative route to the rise and fall of scientific research21
This work proposes for the first time a suite of metrics that can be used to perform post-hoc analysis of the temporal communities of a large-scale citation network of the computer science domain, and quantifies the impact of a field, the influence imparted by one field on the other, the distribution of the “star” papers and authors, the degree of collaboration and seminal publications to characterize such research trends.
- 20
- 20
- Combining Exogenous and Endogenous Signals with a Semi-supervised Co-attention Network for Early Detection of COVID-19 Fake Tweets20
Experimental results on ECTF, PolitiFact, and GossipCop show that ENDEMIC is highly reliable in detecting early fake tweets, outperforming nine state-of-the-art methods significantly.
- Multitask learning for blackmarket tweet detection20
A multitask learning framework is used to effectively detect tweets that have been posted to these blackmarket services, achieving an F1-score of 0.89 when classifying tweets as blackmarket or genuine.
- Automatic Extraction of Complex Predicates in Bengali20
This paper presents the automatic extraction of Complex Predicates (CPs) in Bengali with a special focus on compound verbs (Verb + Verb) and conjunct verbs (Noun /Adjective + Verb).
- 19
- 19
- Public Wisdom Matters! Discourse-Aware Hyperbolic Fourier Co-Attention for Social-Text Classification19
Extensive experiments on four different social-text classification tasks, namely detecting fake news, hate speech, rumour, and sarcasm, show that Hyphen generalises well, and achieves state-of-the-art results on ten benchmark datasets.
- 19
- ECIR 2017 Workshop on Exploitation of Social Media for Emergency Relief and Preparedness (SMERP 2017)19
An overview of the workshop is presented, including the motivations behind organizing the workshop, and summaries of the research papers and keynote talks at the workshop are presented, to reflect on the future directions as inferred from discussion sessions during the workshop.
- All Fingers are not Equal: Intensity of References in Scientific Articles19
This work collects a rich annotated dataset with references labeled by the intensity, and proposes a novel graph-based semi-supervised model, GraLap to label the intensity of references.
- 18
- Semi-supervised Stance Detection of Tweets Via Distant Network Supervision18
This work presents SANDS, a new semi-supervised stance detector that starts from very few labeled tweets, and achieves a macro-F1 score of 0.55 (0.49) on US (India)-based datasets, outperforming 17 baselines substantially, particularly for minority stance labels and noisy text.
- Nipping in the bud18
This article presents methodological challenges that hinder building automated hate mitigation systems and discusses a series of proposed solutions to limit the spread of hate speech on social media.
- Analyzing and Detecting Collusive Users Involved in Blackmarket Retweeting Activities18
This article investigates these blackmarket customers engaged in collusive retweeting activities and adopts Weighted Generalized Canonical Correlation Analysis (WGCCA) to combine these individual representations to derive user embeddings that allow us to effectively classify users as: genuine users, bots, promotional customers, and normal customers.
- On the Formation of Circles in Co-authorship Networks18
An unsupervised approach to automatically detect circles in an ego network such that each circle represents a densely knit community of researchers using a variety of node features and node similarity measures is proposed.
- 18
- 17
- Emotion Flip Reasoning in Multiparty Conversations17
This article proposes a novel neural architecture called TGIF, which leverages Transformer encoders and stacked GRUs to capture the dialogue context, speaker dynamics, and emotion sequence in a conversation, and establishes the generalizability of TGIF on an unseen dataset in a zero-shot setting.
- Node injection for class-specific network poisoning17
NICKI, a novel attacking strategy that utilizes an optimization-based approach to sabotage the performance of GNN-based node classifiers, is proposed and shows that NICKI is consistently better than four baseline attacking strategies for misclassifying nodes in the target class.
- HIT - A Hierarchically Fused Deep Attention Network for Robust Code-mixed Language Representation17
HIT is a hierarchical transformer-based framework that captures the semantic relationship among words and hierarchically learns the sentence-level semantics using a fused attention mechanism and shows the adaptability of learned representation across tasks in a transfer learning setup.
- Blackmarket-Driven Collusion on Online Media: A Survey17
This survey is the first attempt to provide readers a comprehensive outline of the latest studies dealing with the identification and analysis of blackmarket-driven collusion in online media, and provides an easy-to-access and comprehensive list of methods, tools, and resources proposed so far for detecting and analyzing collusive entities on online media.
- 17
- VIKING: Adversarial Attack on Network Embeddings via Supervised Network Poisoning16
VIKING is proposed, a supervised network poisoning strategy that outperforms the state-of-the-art poisoning methods by upto 18% on the original network structure and is extended to a semi-supervised attack setting and it is shown that it is comparable to its supervised counterpart.
- 16
- Artificial intelligence predicts normal summer monsoon rainfall for India in 202315
The data-driven models trained with historical AISMR data, the Niño3.4 index, and categorical Indian Ocean Dipole values outperform the traditional physical models, and the best-performing model predicts that the 2023 AISMR will be roughly 790 mm, which is typical of a normal monsoon year.
- Multi-modal Fake News Detection15
This chapter presents a thorough survey of the recent approaches to detect multi-modal fake news spreading on various social media platforms and describes the proposed methods by categorizing them through a taxonomy.
- $$C^3$$ C 3 -index: a PageRank based multi-faceted metric for authors’ performance measurement14
C^3-index is proposed that combines the effect of citations and collaborations of an author in a systematic way using a weighted multi-layered network to rank authors and can be used to predict future achievers at the early stage of their career.
- 13
- Fingerprinting Fine-tuned Language Models in the Wild13
Fine-tuning itself is the most effective in attributing the synthetic text generated by fine-tuned LMs, and the limitations of existing fingerprinting approaches are demonstrated.
- 13
- DESYR13
Experimental results indicate that DESYR builds upon the state-of-the-art system across four benchmark claim datasets, most of which were constructed with informal texts.
- NewsBag: A Benchmark Multimodal Dataset for Fake News Detection.13
The experimental results show that data augmentation to increase the quantity of fake news does not hamper the accuracy of fake news detection and the utilization of multimodal data for fake news detection substantially outperforms the unimodal algorithms.
- FairScholar: Balancing Relevance and Diversity for Scientific Paper Recommendation13
A novel scientific paper recommendation system that aims at balancing both relevance and diversity while searching for research papers in response to keyword queries, which performs a vertex reinforced random-walk on the citation graph of papers in order to factor in diversity while serving recommendations.
- Universal trajectories of scientific success13
A large-scale study of the rise and fall of scientific success by analyzing the success of two major scientific entities—papers and authors—in Computer Science and Physics is presented and it is believed that this study will argue in favor of revising the existing metrics used for quantifying scientific success.
- Ensemble-based algorithms to detect disjoint and overlapping communities in networks13
Two ensemble methods EnDisCo and MeDOC that identify disjoint and overlapping communities in G are proposed that outperform state-of-the-art non-ensemble algorithms by a significant margin and are considered to be the first ensemble approach for overlapping community detection.
- 13
- Recent Advances in Hate Speech Moderation: Multimodality and the Role of Large Models12
This comprehensive survey delves into the recent strides in HS moderation, spotlighting the burgeoning role of large language models (LLMs) and large multimodal models (LMMs) and uncover a notable trend towards integrating these modalities.
- Revisiting Hate Speech Benchmarks: From Data Curation to System Deployment12
The proposed solution HEN-mBERT is a modular, multilingual, mixture-of-experts model that enriches the linguistic subspace with latent endogenous signals from history, topology, and exemplars that enhances the hate speech detection task.
- Promoting Topic Coherence and Inter-Document Consorts in Multi-Document Summarization via Simplicial Complex and Sheaf Graph12
This work proposes FABRIC, a novel encoder-decoder model that uses pre-trained BART to comprehensively analyze linguistic nuances, simplicial complex layer to apprehend inherent properties that transcend pairwise associations and sheaf graph attention to effectively capture the het-erophilic properties.
- 12
- 12
- Corpora Evaluation and System Bias Detection in Multi-document Summarization12
An attempt to quantify the quality of summarization corpus and prescribe a list of points to consider while proposing a new MDS corpus and the reason behind the absence of an MDS system which achieves superior performance across all corpora is analyzed.
- InPHYNet: Leveraging attention-based multitask recurrent networks for multi-label physics text classification12
A novel method for multi-label classification of paragraphs, where the paragraphs are chosen from physics subject of 6 t h to 1 2 t h grades from the curriculum of Central Board of Secondary Education (CBSE), India is introduced.
- ${\mathit{MALT^P}}$ : Parallel Prediction of Malicious Tweets12
The results show that the presence of multimodal content is not a major factor and that metapath-based features dominate in separating malicious from benign tweets, significantly outperforming the other methods in isolation.
- 12
- The Art of Scaling Test-Time Compute for Large Language Models11
The first large-scale study of TTS is conducted, spanning over thirty billion tokens generated using eight open-source LLMs, and shows no single TTS strategy universally dominates, and for a given model type, the optimal TTS performance scales monotonically with compute budget.
- 11
- 11
- 11
- Go Wide, Go Deep: Quantifying the Impact of Scientific Papers Through Influence Dispersion Trees11
A novel data structure, called Influence Dispersion Tree (IDT), is proposed to model the organization of follow-up papers and their dependencies through citations and it is shown that an ideal IDT configuration should have equal depth and breadth (and thus minimize the NID value).
- PAGE: Answering Graph Pattern Queries via Knowledge Graph Embedding11
A method called \(\mathtt {PAGE}\) is presented that answers graph pattern queries via knowledge graph embedding methods that computes the energy (or uncertainty) of candidate answers with the learned embeddings and chooses the lower-energy candidates as answers.
- Metadata vs. Ground-truth: A Myth behind the Evolution of Community Detection Methods11
This work shows that there is no single CD method that can outperform others by detecting all types of metadata-based partitions, and suggests a new task-based evaluation framework for CD methods, which shows that a certain type of CD methods is useful for a certaintype of task.
- Semantic Interpretation of Social Network Communities11
Whether individuals in a community possess similar Personalities, Values and Ethical background is attempted and it is shown that Personality and Values models could be used as features to discover more accurate community structure.
- Menstrual Health Education Using a Specialized Large Language Model in India: Development and Evaluation Study of MenstLLaMA10
MenstLLaMA demonstrates exceptional accuracy, empathy, and user satisfaction within the domain of MHE, bridging critical gaps left by general-purpose LLMs.
- 10
- 10
- 10
- 10
- 10
- Zera-Shot Sentiment Analysis for Code-Mixed Data10
This work introduces methods that use multilingual and cross-lingual embeddings to transfer knowledge from monolingual text to code-mixed text for code-Mixed sentiment analysis and beats state-of-the-art English-Spanish code- m mixed sentiment analysis by an absolute 3% F1-score.
- Unsupervised Anomaly Detection in Journal-Level Citation Networks10
A novel approach to detect the anomalies in a journal-level scientific citation network and compare the results with the existing graph anomaly detection algorithms is presented and the results resemble the citations and SCImago Journal Rank rating-change charts, thus indicating the usefulness of the method.
- Spotting Collusive Behaviour of Online Fraud Groups in Customer Reviews.10
DeFrauder is proposed, an unsupervised method to detect online fraud reviewer groups that first detects candidate fraud groups by leveraging the underlying product review graph and incorporating several behavioral signals which model multi-faceted collaboration among reviews.
- DiffQue10
Experiments show that DiffQue outperforms four state-of-the-art methods by a significant margin and is capable of adapting multiple domains (CQA datasets), and can efficiently handle the “cold start” problem that may arise due to the lack of information for newly posted questions or newly arrived users.
- C 3 -index10
This work uses the citations as well as collaboration profile of an author in a novel way using a weighted multi-layered network and proposes a page-rank variant to obtain a new author performance measure, C3-index, which is observed to be consistent over time and to predict future achievers at the early stage of their career.
- Semantic Clustering: an Attempt to Identify Multiword Expressions in Bengali10
This paper presents an approach of identifying bigram noun-noun MWEs from a medium-size Bengali corpus by clustering the semantically related nouns and incorporating a vector space model for similarity measurement.
- Shared task system description: Measuring the Compositionality of Bigrams using Statistical Methodologies10
The experiments deal with various collocation based statistical approaches to compute the relative compositionality of three types of bigram phrases (Adjective-Noun, Verbsubject and Verb-object combinations) and reasonable results have been obtained in terms of average point difference and coarse precision.
- 10
- Semantic Clustering: an Attempt to Identify Multiword Expressions in10
This paper presents an approach of identifying bigram noun-noun MWEs from a medium-size Bengali corpus by clustering the semantically related nouns and incorporating a vector space model for similarity measurement.
- DPHGNN: A Dual Perspective Hypergraph Neural Networks9
DPHGNN is proposed, a novel dual-perspective HGNN that introduces equivariant operator learning to capture lower-order semantics by inducing topology-aware spatial and spectral inductive biases.
- 9
- $\texttt{LM}^\texttt{2}$: A Simple Society of Language Models Solves Complex Reasoning9
Exhaustive experimentation suggests the superiority of LM2 over existing methods on in- and out-domain reasoning problems, outperforming the best baselines by 8.1% on MATH, 7.71% on JEEBench, and 9.7% on MedQA problems.
- 9
- 9
- DiVA: A Scalable, Interactive and Customizable Visual Analytics Platform for Information Diffusion on Large Networks9
DiVA—Diffusion Visualization and Analysis is introduced, a tool that provides a scalable web interface and extendable APIs to analyze various diffusion trends on networks and uniquely offers support for simultaneous comparison of two competing diffusion models.
- Document Retrieval and Claim Verification to Mitigate COVID-19 Misinformation9
This work proposes an automated claim verification solution encompassing two steps – document retrieval and veracity prediction, and presents an open-source, easily deployable, and user-friendly Python API that the community can adopt.
- 9
- 9
- GIRNet: Interleaved Multi-Task Recurrent State Sequence Models9
GITNet, a unified position-sensitive multi-task recurrent neural network (RNN) architecture for such applications, is proposed and the superiority of GIRNet is demonstrated using three applications: sentiment classification of code-switched passages, part-of-speech tagging of code -switched text, and target position- sensitive annotation of sentiment in monolingual passages.
- 9
- 9
- 9
- How to Upscale Neural Networks with Scaling Law? A Survey and Practical Guidelines8
It is suggested that while scaling laws provide a useful guide, they do not always generalize across all architectures and training strategies, advocating for adaptive scaling strategies tailored to real-world applications.
- 8
- Understanding (Dark) Humour with Internet Meme Analysis8
This tutorial delivers an integrated framework for dissecting the complex humor of memes, weaving together disciplines such as natural language processing, computer vision, and multimodal modeling, empowering participants to decode meanings, analyze sentiments, and identify offensive content within memes.
- FLAME: Self-Supervised Low-Resource Taxonomy Expansion Using Large Language Models8
Fine-tuning large language Models for taxonomy Expansion (FLAME), a novel approach for taxonomy expansion in low-resource environments by harnessing the capabilities of large language models (LLMs) that are trained on extensive real-world knowledge, is proposed.
- GraphAnoGAN: Detecting Anomalous Snapshots from Attributed Graphs8
GraphAnoGAN is proposed, an anomalous snapshot ranking framework, which consists of two core components -- generative and discriminative models, which learns to approximate the distribution of anomalous samples from the candidate set of graph snapshots.
- 8
- Aspect-Based Academic Search Using Domain-Specific KB8
An aspect-based retrieval task, which allows the user to specify the aspect along with the query to retrieve a ranked list of relevant documents and outperforms keyword-based expansion of query with aspect with and without relevance feedback.
- Ensemble Detection and Analysis of Communities in Complex Networks8
MeDOF, an ensemble method which performs disjoint, overlapping, and fuzzy community detection and represents one of the first ever ensemble methods for fuzzy and overlapping community detection, is proposed and beats the best-known existing stand-alone community detection algorithms.
- Multi-task Learning for Target-Dependent Sentiment Classification8
MTTDSC is presented, a multi-task target-dependent sentiment classification system that is informed by feature representation learnt for the related auxiliary task of passage-level sentiment classification, and anecdotal evidence that prior systems can make incorrect target-specific predictions because they miss sentiments expressed by words independent of target is presented.
- WWW'18 Workshop on Exploitation of Social Media for Emergency Relief and Preparedness: Chairs' Welcome & Organization8
This 2nd SMERP Workshop with The Web Conference 2018 includes two keynote talks, a peer-reviewed research paper track, and a panel discussion about user-generated content on online social media during emergency events.
- Socio-Ethnic Ingredients of Social Network Communities8
This paper collects Twitter values corpus, extracts the network communities and proposes automatic models to determine personality, values, considered as ethnicity of individuals to understand whether individuals in a community possess similar personalities, values and ethical background.
- MMGAN: Manifold Matching Generative Adversarial Network for Generating Images.8
This paper proposes a novel training method called manifolds-matching, and a new GAN model called manifold- matching GAN (MMGAN), which finds two manifolds representing the vector representations of real and fake images.
- PubIndia: A Framework for Analyzing Indian Research Publications in Computer Science8
PubIndia is described, a framework for analyzing the growth and impact of research activities performed in India in the computer science domain, based on the evidence of scientific publications and the evolutionary landscape of different keywords that indicate how the importance of individual keywords varies over the years.
- An author is known by the context she keeps: significance of network motifs in scientific collaborations8
This empirical study sets the foundation principles of a recommendation system that would forecast how long lasting and productive a given collaboration could be in future.
- Inference of Fine-grained Attributes of Bengali Corpus for Stylometry Detection8
Evaluation results of the attempt for Bengali author' s stylometry detection show reasonably promising accuracy in comparison to the baseline model.
- 7
- Position: Enough of Scaling LLMs! Lets Focus on Downscaling7
This paper outlines practical strategies for transitioning away from traditional scaling paradigms, advocating for a more sustainable, efficient, and accessible approach to LLM development.
- On the Effect of Instruction Tuning Loss on Generalization7
It is found that a low-to-moderate weight for prompt tokens coupled with a moderate-to-high weight for response tokens yields the best-performing models across settings and also serves as a better starting point for the subsequent preference alignment training.
- 7
- PerSEval: Assessing Personalization in Text Summarizers7
This paper challenges the necessity of an accuracy leaderboard, suggesting that relying on accuracy-based aggregated results might lead to misleading conclusions and proposes PerSEval, a novel measure that satisfies the required sufficiency condition.
- Overview of the CLAIMSCAN-2023: Uncovering Truth in Social Media through Claim Detection and Identification of Claim Spans7
CLAIMSCAN was presented at the 2023 Forum for Information Retrieval Evaluation (FIRE’2023), and the primary goals focused on two critical tasks: determining whether a social media post is a claim, and precisely identifying the words or phrases within the post that form the claim.
- Overview of the HASOC Subtracks at FIRE 2023: Detection of Hate Spans and Conversational Hate-Speech7
Two sub-tracks related to Hate Speech detection are reported on, where the task ICHCL provided a dataset in code-mixed Hindi, which includes the conversational context of a tweet, and HateNorm introduced a span annotated dataset and required systems to find a sequence of words relevant to the hatefulness of a post.
- Why Did You Not Compare with That? Identifying Papers for Use as Baselines7
A multi-module attention-based neural classifier is developed for the baseline classification task that outperforms four state-of-the-art citation role classification methods when applied to the baseline Classification task.
- 7
- Attention Beam: An Image Captioning Approach (Student Abstract)7
A heuristic of beam search on top of the encoder-decoder based architecture that gives better quality captions on three benchmark datasets: Flickr8k, Flickr30k and MS COCO is presented.
- Assessing the quality of the datasets by identifying mislabeled samples7
This paper proposes a novel statistic - noise score, as a measure for the quality of each data point to identify mislabeled samples based on the variations in the latent space representation derived by the inference network of data quality supervised variational autoencoder (AQUAVS).
- Automatic Curation of Content Tables for Educational Videos7
This work presents a novel architecture to curate content tables for educational videos that allows users to browse the video smartly by skipping to a particular portion rather than going through the entire video.
- 7
- SENA7
SENA, a structural embedding and network augmentation framework for social network analysis that generates features for both vertices and relations by minimizing a well-designed objective function composed of a loss function and a regularization is proposed.
- Is this conference a top-tier? ConfAssist: An assistive conflict resolution framework for conference categorization7
ConfAssist, a novel assisting framework for conference categorization that aims to address the limitations in the existing systems and portals for venue classification, is proposed and the hypothesis that top-tier conferences are much more stable than other conferences is started.
- Multi-Agent System for Search Engine based Web Server: A Conceptual Framework.7
This paper proposes a system for user specific data mining over World Wide Web (WWW) consisting of several agents which are capable of working separately & intelligently to achieve their individual goal.
- SAFE-MEME: Structured Reasoning Framework for Robust Hate Speech Detection in Memes6
This work curates two novel multimodal hate speech datasets comprising English memes – MHS and MHS-Con, which capture fine-grained hateful abstractions in regular and confounding scenarios, respectively and introduces SAFE-MEME (Structured reAsoning FramEwork), a novel multimodal Chain-of-Thought-based framework employing Q&A-style reasoning and hierarchical categorization to enable robust hate speech detection in memes.
- 6
- 6
- 6
- Literature mining discerns latent disease–gene relationships6
The repurposed BERT predicted disease–gene associations that are not cited in the training data are highlighted, thereby highlighting the feasibility of in silico synthesis of hypotheses linking different biological entities such as genes and conditions.
- 6
- Overview of the HASOC Subtrack at FIRE 2023: Identification of Tokens Contributing to Explicit Hate in English by Span Detection6
HateNorm was organized at HASOC-FIRE 2023, focusing on explicit span detection in English Tweets, with the highest macro-F1 observed at 0.58.
- 6
- QUINT: Node embedding using network hashing6
This paper shows how to extend <inline-formula><tex-math notation="LaTeX") to retain enough structural information which can be used further to approximate many topological properties of networks with high confidence, and proves that <monospace>QUINT</monospace > is a consistent top-performer for both the tasks among the baselines across all the datasets.
- 6
- How Did the Discussion Go: Discourse Act Classification in Social Media Conversations6
A novel attention-based hierarchical LSTM model to classify discourse act sequences in social media conversations, aimed at mining data from online discussion using textual meanings beyond sentence level, and presents a human annotator experiment to unveil important observations about modeling and data annotation.
- Identifying Bengali Multiword Expressions using semantic clustering6
The comparative evaluation shows that the semantic clustering approach outperforms all other competing statistical models and starts developing a standard lexicon in Bengali that serves as a productive Bengali linguistic thesaurus.
- Understanding the Effects of Domain Finetuning on LLMs5
The first systematic study of domain-specific fine-tuning in large medical language models is presented, revealing that fine-tuning modifies only a small subset of the representational subspace, essentially preserving the pre-trained model's representation.
- Multilingual Test-Time Scaling via Initial Thought Transfer5
The first systematic study of test-time scaling in multilingual settings is presented, evaluating DeepSeek-R1-Distill-LLama-8B and DeepSeek-R1-Distill-Qwen-7B across both high- and low-resource Latin-script languages.
- Step-by-Step Unmasking for Parameter-Efficient Fine-tuning of Large Language Models5
ID3 is introduced, a novel selective-PEFT method that calculates parameter importance continually, and dynamically unmasks parameters by balancing exploration and exploitation in parameter selection, and can be integrated with existing additive and reparameterization-based PEFT techniques such as Adapters and LoRA.
- Persona-aware Generative Model for Code-mixed Language5
This work proposes a Persona-aware Generative Model for Code-mixed Generation, PARADOX, a novel Transformer-based encoder-decoder model that encodes an utterance conditioned on a user's persona and generates code-m mixed texts without monolingual reference data.
- 5
- Transcriptomic Profiling Unravels Novel Deregulated Gene Signatures Associated with Acute Myocardial Infarction: A Bioinformatics Approach5
It is found that the deregulated genes such as ADOR-A3, BMP6, VPS8, and GPx3, may be associated with AMI, and these genes and miRNA might play a crucial role the rescue of cardiomyocytes from severe damage.
- 5
- TweetGrep: Weakly Supervised Joint Retrieval and Sentiment Analysis of Topical Tweets5
This work proposes a weakly supervised approach (named, TweetGrep) that lets the data analyst easily define a topic by few keywords and adapt a generic sentiment classifier to the topic – by jointly modeling topics and sentiments using label regularization.
- Dynamics of node influence in network growth models5
This paper studies how a metric for node influence in network growth models behaves over time as the network evolves, which is called node visibility, and observes that a spatial growth model with multiplicative fitness can curtail the global reach of influential nodes, thereby allowing the emergence of a multiplicity of "local leaders" in the network.
- ABOME: A Multi-platform Data Repository of Artificially Boosted Online Media Entities5
ABOME, a novel multi-platform data repository consisting of artificially boosted online media entities (also known as blackmarket-driven collusive entities), which are prevalent but often unnoticed in online media, is introduced.
- 5
- Clark Kent at SemEval-2019 Task 4: Stylometric Insights into Hyperpartisan News Detection5
An XGBoost based system which uses character and word level n-gram features represented using TF-IDF, count vector based correlation matrix, and predicts if an input news article is a hyperpartisan news article.
- MC3: A Multi-class Consensus Classification Framework5
This paper proposes two algorithms, MC3-R andMC3-S that make different trade-offs between quality and runtime, and performs as well or better than the best baseline classifier, achieving on average, a 5.56% performance improvement.
- AI Across Borders: Exploring Perceptions and Interactions in Higher Education4
While perceptions varied across institutions, responses for Computer Sciences students were similar, both in terms of topics and degree of positivity, and after controlling for institution and subject area, there was no effect of gender.
- 4
- QuanTaxo: A Quantum Approach to Self-Supervised Taxonomy Expansion4
QuanTaxo is introduced, a quantum-inspired framework for taxonomy expansion that encodes entities in a Hilbert space and models interference effects between them, yielding richer, context-sensitive representations.
- Can LLMs reason over extended multilingual contexts? Towards long-context evaluation beyond retrieval and haystacks4
This work introduces MLRBench, a new synthetic benchmark for multilingual long-context reasoning that goes beyond surface-level retrieval by including tasks that assess multi-hop inference, aggregation, and epistemic reasoning.
- 4
- 4
- 4
- Temporally Consistent Factuality Probing for Large Language Models4
This study introduces TeCFaP, a novel Temporally Consistent Factuality Probe task to expand the consistent factuality probe in the temporal dimension, and proposes a novel solution CoTSeLF (Consistent-Time-Sensitive Learning Framework) combining multi-task instruction tuning with consistent-time-sensitive reinforcement learning (CTSRL) to improve temporally consistent factuality in LLMs.
- Crowd Intelligence for Early Misinformation Prediction on Social Media4
This work introduces CROWDSHIELD, a crowd intelligence-based method for early misinformation prediction, hypothesize that the crowd's reactions to misinformation reveal its accuracy, and employs Q-learning to capture the two dimensions -- stances and claims.
- 4
- Judging the creative prowess of AI4
The context of the work and its shortand long-term impact on society are some aspects that are factored into the evaluation of creative work, and how does one judge the creative correctness of generative models’ creative correctness?
- 4
- 4
- Learning Multigraph Node Embeddings Using Guided Lévy Flights4
This paper proposes Multigraph2Vec, a random walk based framework for learning multigraph network representation, which outperforms others in the task of link prediction, by beating the best baseline with 5.977% higher AUC score.
- 4
- 4
- ConfAssist4
ConfAssist is proposed, a novel assisting framework for conference categorization that aims to address the limitations in the existing systems and portals for venue classification by increasing the confidence of the experts in their decision.
- 4
- 4
- 3
- 3
- 3
- Clinically Grounded AI-Scribing in Psychotherapy: Benchmarking LLMs Against Expert Documentation in the iCARE Framework3
iCARE (identifiers, Chief Concerns and Clinical History, Assessment and Analysis, Risk and Crisis, Engagement and Next Steps), a comprehensive framework for AI-assisted psychotherapy documentation that addresses limitations of recent advances in Artificial Intelligence and Large Language Models is presented.
- MMA-ASIA: A Multilingual and Multimodal Alignment Framework for Culturally-Grounded Evaluation3
This work proposes MMA-ASIA, a comprehensive framework to evaluate LLMs' cultural awareness with a focus on Asian contexts, and proposes a five-dimensional evaluation protocol that measures cultural-awareness disparities across countries, cross-lingual consistency, cross-modal consistency, cultural knowledge generalization, and grounding validity.
- 3
- 3
- 3
- 3
- Focal inferential infusion coupled with tractable density discrimination for implicit hate detection3
FiADD is introduced, a novel focused inferential adaptive density discrimination framework that enhances the PLM finetuning pipeline by bringing the surface form/meaning of an implicit hate speech closer to its implied form while increasing the intercluster distance among various labels.
- 3
- 3
- 3
- An Embedding-based Joint Sentiment-Topic Model for Short Texts3
ELJST is an embedding enhanced generative joint sentiment-topic model that can discover more coherent and diverse topics from short texts and helps understand users' behaviour at more granular levels which can be explained.
- 3
- Modeling Engagement Dynamics of Online Discussions using Relativistic Gravitational Theory3
This study proposes RGNet (Relativistic Gravitational Nerwork), a novel algorithm that uses Einstein Field Equations of gravity to model online discussions as 'cloud of dust' hovering over a user spacetime manifold, attracting users of different groups at different rates over time.
- Into the Battlefield3
This work presents a novel quantification of conflict in online discussion, which paves the way towards new methods of exploration and modeling of conflict dynamics inside online discussion communities.
- Hybrid adversarial defense: Merging honeypots and traditional security methods3
A Stackelberg-style game situation where the defender models the attacker and uses that model to identify the best placement of traditional defenses and honeypots is considered and a greedy algorithm is proposed that quickly computes near optimal placements.
- VIZ-Wiki: Generating Visual Summaries to Factoid Threads in Community Question Answering Services3
VIZ-Wiki is the first attempt to visualize answers for questions in community question answering services, and deals with answers contradicted by other users, prioritizes highly-recommended ones and avoids sarcasm.
- 3
- Mining Twitter Conversations around E-commerce Promotional Events3
A novel technique for mining conversations in Twitter by weaving together all conversations around an event into one unified graph (Conversation Graph, henceforth) emerges as a variant of the BOWTIE structure as a result of the complex communication patterns amongst these players.
- 2
- 2
- Network based simultaneous embedding of cells and marker genes from scRNA-seq studies2
Stardust is proposed, an iterative, force-directed graph layout algorithm that enables the simultaneous embedding of cells and marker genes and, for the first time, allows a single-stop visualization of cells and marker genes on a single 2D map.
- Task-conditioned probing of instruction-tuned multimodal LLMs: Region-specific brain alignment patterns under naturalistic stimuli2
Evaluation of MLLMs for both video and audio tasks using language-guided instructions shows clear disentanglement in task-specific representations from MLLMs, leading to precise differentiation of multimodal functional processing in the brain.
- 2
- 2
- Co-Designed Mental Health Screening App (Here for You) for University Students: Pilot Feasibility Mixed Methods Study2
Preliminary evidence for the feasibility and user acceptability of the Here for You app is provided, co-designed using a participatory approach with Indian university students, which offers a culturally resonant and scalable model for digital mental health screening in low-resource settings.
- Camellia: Benchmarking Cultural Biases in LLMs for Asian Languages2
Camellia is introduced, a benchmark for evaluating entity-centric cultural biases in nine Asian languages, spanning six Asian cultures, and it is found that LLMs can struggle with context understanding in some Asian languages, creating performance gaps between cultures in entity extraction.
- MenstLLaMA: A Specialized Large Language Model for Menstrual Health Education in India2
MenstLLaMA demonstrates exceptional accuracy, empathy, and user satisfaction in menstrual health education, bridging critical gaps left by general-purpose LLMs.
- 2
- Harmonizing Code-mixed Conversations: Personality-assisted Code-mixed Response Generation in Dialogues2
This study introduces a novel approach centered on harnessing the Big Five personality traits acquired in an unsupervised manner from the conversations to bolster the performance of response generation within code-mixed conversations.
- 2
- Leveraging Social Discourse to Measure Check-worthiness of Claims for Fact-checking2
A new task of fine-grained claim check-worthiness is introduced, which underpins all of these factors and provides probable human grounds for identifying a claim as check-worthy and benchmarked against a unified approach, CheckMate, that jointly determines whether a claim is check- worthy and the factors that led to that conclusion.
- 2
- Half-Day Tutorial on Combating Online Hate Speech: The Role of Content, Networks, Psychology, User Behavior, etc.2
A holistic view of what the research community has explored so far in hate speech, and what it believes are promising future research directions is provided.
- Weakening the Inner Strength: Spotting Core Collusive Users in YouTube Blackmarket Network2
NURSE, a deep fusion framework that only leverages user timelines (without considering the underlying CCN) to detect core blackmarket users is proposed and Experimental results show that NURSE is quite close to KORSE in detecting core users and outperforms nine baselines.
- 2
- 2
- 2
- 2
- 2
- Incremental community discovery via latent network representation and probabilistic inference2
An incremental community detection method that repeats two stages—(i) network scan and (ii) community update, which achieves 10.2% higher accuracy on average over state-of-the-art algorithms for both network scan and community update steps.
- Fair Transfer of Multiple Style Attributes in Text2
This work proposes a neural network architecture for fairly transferring multiple style attributes in a given text and demonstrates that the transfer of multiple styles cannot be achieved by sequentially performing multiple single-style transfers.
- Report on the Second Workshop on Exploitation of Social Media for Emergency Relief and Preparedness (SMERP 2018) at the Web Conference (WWW) 20182
The Second Workshop on Exploitation of Social Media for Emergency Relief and Preparedness (SMERP) was held in conjunction with The Web Conference (WWW) 2018 at Lyon, France to promote multi-modal and multi-view information retrieval from the social media content in disaster situations.
- ComPAS: Community Preserving Sampling for Streaming Graphs2
Compas is proposed, a novel sampling strategy that unlike previous methods, is not only designed for streaming graphs but also preserves the community structure of the original graph in the sample.
- 2
- Visibility of Nodes in Network Growth Models2
The primary finding is that influential nodes in multiplicative growth models can attain and maintain high visibility over time compared to other models; something that might not be apparent by simply looking at global network properties or other local node-centric properties.
- 2
- On the Discovery of Success Trajectories of Authors2
This work examines the career profile of authors in computer science and physics domains and discovers at least six different success trajectories in terms of normalized citation count in longitudinal scale, and builds a two-stage stratification model to predict future success of an author at the early stage of her career.
- Using Semantic Clustering for Detecting Bengali Multiword Expressions2
A novel approach called "semantic clustering" is proposed as an instrument for extracting the MWEs especially for resource constraint languages like Bengali and shows that the semantic clustering outperforms all other competing methods.
- Identifying base clusters and their application to maximizing modularity2
An algorithm for obtaining base clusters is developed and it is shown that identifying base clusters as a preprocessing step can help in improving the modularity values for agglomerative methods.
- Measuring the compositionality of bigrams using statistical methodologies: shared task system description2
The experiments deal with various collocation based statistical approaches to compute the relative compositionality of three types of bigram phrases (Adjective-Noun, Verb-subject and Verb-object combinations) and reasonable results have been obtained in terms of average point difference and coarse precision.
- 1
- Benchmarking Large Language Models Against Psychiatry Residents Using Traditional Institutional Assessments1
While AI demonstrated superior theoretical knowledge, equivalent or variable practical skills performance reveals fundamental limitations in clinical reasoning and contextual understanding, these findings necessitate reconceptualizing psychiatric education to emphasize uniquely human competencies while leveraging AI’s capabilities for knowledge synthesis.
- 1
- 1
- Sentiment-guided Commonsense-aware Response Generation for Mental Health Counseling1
EmpRes, a novel sentiment-guided mechanism incorporating commonsense awareness for generating responses that effectively shape the client's sentiment towards positivity, is proposed and deployed as a chat interface for users seeking mental health support.
- 1
- Temporal Referential Consistency: Do LLMs Favor Sequences Over Absolute Time References?1
A novel benchmark entitled temporal referential consistency is introduced, accompanied by a resource TEMP-ReCon designed to benchmark a wide range of both open-source and closed-source LLMs with various linguistic contexts characterized by differing resource richness.
- 1
- 1
- 1
- 1
- Mechanistic Behavior Editing of Language Models1
Experiments on multiple classification and generation tasks using LLMs of varying sizes reveal the efficacy of TaRot, improving upon both zero- as well as few-shot performance, with average improvements (across models and tasks) of 23.81% and 11.15%, respectively.
- 1
- Large Language Model Predicts Above Normal All India Summer Monsoon Rainfall in 20241
This research focuses on adapting and fine-tuning the latest LLM model, PatchTST, to accurately predict AISMR with a lead time of three months, and predicts an above-normal monsoon for the year 2024, with an accumulated rainfall of 921.6 mm in the month of June-September for the entire country.
- Exploiting Representation Bias for Data Distillation in Abstractive Text Summarization1
This paper aims to discretize the vector space of the abstractive text summarization models to understand the characteristics learned between the input embedding space and the models' encoder space, and shows that deep models fail to capture the diversity of the input space.
- 1
- 1
- Uncertainty in Additive Feature Attribution methods1
A measure to quantify the relative complexity of a blackbox classifier is proposed and incorporated, for example, in LIME-based algorithms' sampling densities, to help different explanation algorithms achieve tighter confidence levels.
- Incomplete Gamma Integrals for Deep Cascade Prediction Using Content, Network, and Exogenous Signals1
A new cascade growth model as a parametric function of time, which combines deep influence signals from content, network features, and exogenous event sources, is proposed that outperforms seven recent and diverse baselines significantly on a large-scale dataset of retweet cascades coupled with time-aligned online news.
- MG2Vec+: A multi-headed graph attention network for multigraph embedding1
MG2Vec+ is proposed, a method that generates node embeddings for a multigraph, a network structure comprising multiple types of edges between pairs of nodes, that uses multi-headed attention layers to aggregate multiple type of edge-relations that can exist among nodes.
- 1
- Detecting Anchors' Opinion in Hinghlish News Delivery1
A novel task of anchors' opinion detection in debates is proposed and an interactive attention-based framework for classifying anchors' utterances is proposed, which shows many interesting patterns in the dataset and predictions.
- 1
- Near-optimal Sparse Neural Trees1
This work aims to build a mathematical formulation of neural trees and gain the complementary benefits of both sparse optimal decision trees and neural trees, and proposes near-optimal sparse neural trees (NSNT) that is shown to be asymptotically consistent and robust in nature.
- Special section on mining knowledge from scientific data1
This special section aims to bring together the latest groundbreaking research on issues related to knowledge extraction and deriving insights from scientific data, and welcomed both theoretical and empirical research, and case studies that lead to the development of novel algorithms, tool, techniques, metrics, decisions and measurements related to scholarly data.
- Modeling Citation Trajectories of Scientific Papers1
Experimental results on the real-world citation networks of Computer Science and Physics domains show that the proposed models can better explain the temporal behavior of citation networks than existing models.
- 1
- 1
- 1
- Towards Identification of Nominal Multiword Expressions in Bengali Language1
The investigation of Noun-Noun bigram collocations from the medium-size untagged Bengali corpus of the articles of Rabindranath Tagore is investigated using simple unsupervised approach with various statistical evidences to show the affinity of the constituents of each bigram candidate as a proof of the Multi-Word Expression (MWE) and build a weighted measurement to get a distinction between MWE or non-MWE.
- An enhanced federated learning-based multistage IDS for detecting zero-day and known threats in smart home ecosystem–
A modified federated learning based novel multi-stage intrusion detection system for real-time identification of attacks in a smart home ecosystem that ensures privacy-preserving model aggregation across distributed IoT devices while minimizing communication overhead and adapting to dynamic network conditions is introduced.
- –
- –
- –
- –
- –
- –
- –
- Innocence in the Crossfire: Roles of Skip Connections in Jailbreaking Visual Language Models–
It is demonstrated that memes, often perceived as humorous or harmless, can be as effective as toxic visuals in eliciting harmful content in eliciting harmful content in Visual Language Models, underscoring the subtle and complex vulnerabilities of VLMs.
- –
- –
- –
- –
- Information Anxiety in Large Language Models–
This work focuses on three critical dimensions - the impact of entity popularity, the models' sensitivity to lexical variations in query formulation, and the progression of hidden state representations across LLM layers.
- –
- –
- –
- Self-similarity of temporal interaction networks arises from hyperbolic geometry with time-varying curvature–
This work proposes a novel technique of scale transformation that dissects temporal interaction networks under spatio-temporal scales, namely, flow scales, and relates the emergence of flow-scale self-similarity to the latent geometry of such networks.
- Energy efficient permanence‐based community detection algorithm–
Amoeba is a task parallel implementation of a permanence‐based community detection algorithm designed for multicore processors that uses dynamic tasking to schedule the inherent irregular computation, and it can dynamically adapt the total number of parallel threads, which results in improved energy efficiency.
- –
- –
- Speaker Profiling in Multiparty Conversations–
This paper has curated a new dataset named SPICE, which comes with specific labels and evaluated various baselines on this dataset and benchmarked it with a new neural model, SPOT, which is introduced in this paper.
- –
- –
- –
- Simultaneous visualization of cells and marker genes from scRNA-seq studies–
Stardust is proposed, an iterative, force-directed graph layouting algorithm that enables the simultaneous embedding of cells and marker genes and, for the first time, allows a single-stop visualization of cells and marker genes on a single 2D map.
- A visual atlas of genes’ tissue-specific pathological roles–
A word embedding technique from the domain of Natural Language Processing (NLP) is employed to learn vector representation of entities such as genes, diseases, tissues, etc., in a way such that their relationship is preserved in a vector space.
- –
- –
- –
- –
- –
- –
- –
- –
- Fake reviewer group detection–
DeFrauder is proposed, an unsupervised method to detect online fraud reviewer groups that first detects candidate fraud groups by leveraging the underlying product review graph and incorporating several behavioral signals which model multi-faceted collaboration among reviewers.
- –
- Detection of content-level collusive activities in online social networks–
The idea of collusion is expanded on by proposing CoReRank, an unsupervised al- algorithm to rank users and tweets based on their participation in the collusive ecosystem, and the presence of sparse labelled data is used to formulate its semi-supervised version called CoReRank+.
- –
- –
- Analyzing user discussion dynamics in social media platforms–
This work studies user behavior and its patterns in primarily three parts: how the sentiment of a user to an entity changes with respect to time, how a user changes the polarity with time, and how many users had the same kind of polarity drift for an entity.
- –
- Study of community aware network expanion–
The motive is to explore the incomplete network such that nodes which are in the community of the known nodes are brought into the network, and it is observed that Machine learning classifier is the best approach to maximize the recall by exploring least number of nodes, whereas global clustering coefficient is thebest approach for maintaining high precision.
- –
- Relative difficulty estimation in community answering services–
This work tried various graph models in order to model the intuition of how difficulty is associated with questions, the answerers, the asker and how over time the difficulty of one’s questions change.
- –
- –
- –
- –
- –
- –
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-12. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.