Conversational prompt-engineering-based large language models (LLMs) have enabled targeted control over the output creation, enhancing versatility, adaptability and adhoc retrieval. From another perspective, digital misinformation has reached alarming levels. The anonymity, availability and reach of social media offer fertile ground for rumours to propagate. This work proposes to leverage the advancement of prompting-dependent LLMs to combat misinformation by extending the research efforts of the RumourEval task on its Twitter dataset. To the end, we employ two prompting-based LLM variants (GPT-3.5-turbo and GPT-4) to extend the two RumourEval subtasks: (1) veracity prediction, and (2) stance classification. For veracity prediction, three classifications schemes are experimented per GPT variant. Each scheme is tested in zero-, one- and few-shot settings. Our best results outperform the precedent ones by a substantial margin. For stance classification, prompting-based-approaches show comparable performance to prior results, with no improvement over finetuning methods. Rumour stance subtask is also extended beyond the original setting to allow multiclass classification. All of the generated predictions for both subtasks are equipped with confidence scores determining their trustworthiness degree according to the LLM, and post-hoc justifications for explainability and interpretability purposes. Our primary aim is AI for social good.
We apply causal mediation analysis to explain the decision-making process of neural models for rumour detection on Twitter. Interventions at the input and network level reveal the causal impacts of tweets and words in the model output. We find that our approach CMA-R -- Causal Mediation Analysis for Rumour detection -- identifies salient tweets that explain model predictions and show strong agreement with human judgements for critical tweets determining the truthfulness of stories. CMA-R can further highlight causally impactful words in the salient tweets, providing another layer of interpretability and transparency into these blackbox rumour detection systems. Code is available at: https://github.com/ltian678/cma-r.
Text classification methods have been widely investigated as a way to detect content of low credibility: fake news, social media bots, propaganda, etc. Quite accurate models (likely based on deep neural networks) help in moderating public electronic platforms and often cause content creators to face rejection of their submissions or removal of already published texts. Having the incentive to evade further detection, content creators try to come up with a slightly modified version of the text (known as an attack with an adversarial example) that exploit the weaknesses of classifiers and result in a different output. Here we systematically test the robustness of common text classifiers against available attacking techniques and discover that, indeed, meaning-preserving changes in input text can mislead the models. The approaches we test focus on finding vulnerable spans in text and replacing individual characters or words, taking into account the similarity between the original and replacement content. We also introduce BODEGA: a benchmark for testing both victim models and attack methods on four misinformation detection tasks in an evaluation framework designed to simulate real use-cases of content moderation. The attacked tasks include (1) fact checking and detection of (2) hyperpartisan news, (3) propaganda and (4) rumours. Our experimental results show that modern large language models are often more vulnerable to attacks than previous, smaller solutions, e.g. attacks on GEMMA being up to 27\% more successful than those on BERT. Finally, we manually analyse a subset adversarial examples and check what kinds of modifications are used in successful attacks.
Social media has become the fulcrum of all forms of communication. Classifying social texts such as fake news, rumour, sarcasm, etc. has gained significant attention. The surface-level signals expressed by a social-text itself may not be adequate for such tasks; therefore, recent methods attempted to incorporate other intrinsic signals such as user behavior and the underlying graph structure. Oftentimes, the `public wisdom' expressed through the comments/replies to a social-text acts as a surrogate of crowd-sourced view and may provide us with complementary signals. State-of-the-art methods on social-text classification tend to ignore such a rich hierarchical signal. Here, we propose Hyphen, a discourse-aware hyperbolic spectral co-attention network. Hyphen is a fusion of hyperbolic graph representation learning with a novel Fourier co-attention mechanism in an attempt to generalise the social-text classification tasks by incorporating public discourse. We parse public discourse as an Abstract Meaning Representation (AMR) graph and use the powerful hyperbolic geometric representation to model graphs with hierarchical structure. Finally, we equip it with a novel Fourier co-attention mechanism to capture the correlation between the source post and public discourse. 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. We also employ a sentence-level fact-checked and annotated dataset to evaluate how Hyphen is capable of producing explanations as analogous evidence to the final prediction.
Occurrences of catastrophes such as natural or man-made disasters trigger the spread of rumours over social media at a rapid pace. Presenting a trustworthy and summarized account of the unfolding event in near real-time to the consumers of such potentially unreliable information thus becomes an important task. In this work, we propose MTLTS, the first end-to-end solution for the task that jointly determines the credibility and summary-worthiness of tweets. Our credibility verifier is designed to recursively learn the structural properties of a Twitter conversation cascade, along with the stances of replies towards the source tweet. We then take a hierarchical multi-task learning approach, where the verifier is trained at a lower layer, and the summarizer is trained at a deeper layer where it utilizes the verifier predictions to determine the salience of a tweet. Different from existing disaster-specific summarizers, we model tweet summarization as a supervised task. Such an approach can automatically learn summary-worthy features, and can therefore generalize well across domains. When trained on the PHEME dataset [29], not only do we outperform the strongest baselines for the auxiliary task of verification/rumour detection, we also achieve 21 - 35% gains in the verified ratio of summary tweets, and 16 - 20% gains in ROUGE1-F1 scores over the existing state-of-the-art solutions for the primary task of trustworthy summarization.
In recent years, visual forgery has reached a level of sophistication that humans cannot identify fraud, which poses a significant threat to information security. A wide range of malicious applications have emerged, such as fake news, defamation or blackmailing of celebrities, impersonation of politicians in political warfare, and the spreading of rumours to attract views. As a result, a rich body of visual forensic techniques has been proposed in an attempt to stop this dangerous trend. In this paper, we present a benchmark that provides in-depth insights into visual forgery and visual forensics, using a comprehensive and empirical approach. More specifically, we develop an independent framework that integrates state-of-the-arts counterfeit generators and detectors, and measure the performance of these techniques using various criteria. We also perform an exhaustive analysis of the benchmarking results, to determine the characteristics of the methods that serve as a comparative reference in this never-ending war between measures and countermeasures.
As social media becomes increasingly prominent in our day to day lives, it is increasingly important to detect informative content and prevent the spread of disinformation and unverified rumours. While many sophisticated and successful models have been proposed in the literature, they are often compared with older NLP baselines such as SVMs, CNNs, and LSTMs. In this paper, we examine the performance of a broad set of modern transformer-based language models and show that with basic fine-tuning, these models are competitive with and can even significantly outperform recently proposed state-of-the-art methods. We present our framework as a baseline for creating and evaluating new methods for misinformation detection. We further study a comprehensive set of benchmark datasets, and discuss potential data leakage and the need for careful design of the experiments and understanding of datasets to account for confounding variables. As an extreme case example, we show that classifying only based on the first three digits of tweet ids, which contain information on the date, gives state-of-the-art performance on a commonly used benchmark dataset for fake news detection --Twitter16. We provide a simple tool to detect this problem and suggest steps to mitigate it in future datasets.
Natural Language Processing algorithms have made incredible progress, but they still struggle when applied to out-of-distribution examples. We address a challenging and underexplored version of this domain adaptation problem, where an algorithm is trained on several source domains, and then applied to examples from unseen domains that are unknown at training time. Particularly, no examples, labeled or unlabeled, or any other knowledge about the target domain are available to the algorithm at training time. We present PADA: An example-based autoregressive Prompt learning algorithm for on-the-fly Any-Domain Adaptation, based on the T5 language model. Given a test example, PADA first generates a unique prompt for it and then, conditioned on this prompt, labels the example with respect to the NLP prediction task. PADA is trained to generate a prompt which is a token sequence of unrestricted length, consisting of Domain Related Features (DRFs) that characterize each of the source domains. Intuitively, the generated prompt is a unique signature that maps the test example to a semantic space spanned by the source domains. In experiments with 3 tasks (text classification and sequence tagging), for a total of 14 multi-source adaptation scenarios, PADA substantially outperforms strong baselines.
The sudden widespread menace created by the present global pandemic COVID-19 has had an unprecedented effect on our lives. Man-kind is going through humongous fear and dependence on social media like never before. Fear inevitably leads to panic, speculations, and the spread of misinformation. Many governments have taken measures to curb the spread of such misinformation for public well being. Besides global measures, to have effective outreach, systems for demographically local languages have an important role to play in this effort. Towards this, we propose an approach to detect fake news about COVID-19 early on from social media, such as tweets, for multiple Indic-Languages besides English. In addition, we also create an annotated dataset of Hindi and Bengali tweet for fake news detection. We propose a BERT based model augmented with additional relevant features extracted from Twitter to identify fake tweets. To expand our approach to multiple Indic languages, we resort to mBERT based model which is fine-tuned over created dataset in Hindi and Bengali. We also propose a zero-shot learning approach to alleviate the data scarcity issue for such low resource languages. Through rigorous experiments, we show that our approach reaches around 89% F-Score in fake tweet detection which supercedes the state-of-the-art (SOTA) results. Moreover, we establish the first benchmark for two Indic-Languages, Hindi and Bengali. Using our annotated data, our model achieves about 79% F-Score in Hindi and 81% F-Score for Bengali Tweets. Our zero-shot model achieves about 81% F-Score in Hindi and 78% F-Score for Bengali Tweets without any annotated data, which clearly indicates the efficacy of our approach.
The inability to correctly resolve rumours circulating online can have harmful real-world consequences. We present a method for incorporating model and data uncertainty estimates into natural language processing models for automatic rumour verification. We show that these estimates can be used to filter out model predictions likely to be erroneous, so that these difficult instances can be prioritised by a human fact-checker. We propose two methods for uncertainty-based instance rejection, supervised and unsupervised. We also show how uncertainty estimates can be used to interpret model performance as a rumour unfolds.
The Internet is rife with flourishing rumours that spread through microblogs and social media. Recent work has shown that analysing the stance of the crowd towards a rumour is a good indicator for its veracity. One state-of-the-art system uses an LSTM neural network to automatically classify stance for posts on Twitter by considering the context of a whole branch, while another, more simple Decision Tree classifier, performs at least as well by performing careful feature engineering. One approach to predict the veracity of a rumour is to use stance as the only feature for a Hidden Markov Model (HMM). This thesis generates a stance-annotated Reddit dataset for the Danish language, and implements various models for stance classification. Out of these, a Linear Support Vector Machine provides the best results with an accuracy of 0.76 and macro F1 score of 0.42. Furthermore, experiments show that stance labels can be used across languages and platforms with a HMM to predict the veracity of rumours, achieving an accuracy of 0.82 and F1 score of 0.67. Even higher scores are achieved by relying only on the Danish dataset. In this case veracity prediction scores an accuracy of 0.83 and an F1 of 0.68. Finally, when using automatic stance labels for the HMM, only a small drop in performance is observed, showing that the implemented system can have practical applications.
This paper describes our submission to SemEval-2019 Task 7: RumourEval: Determining Rumor Veracity and Support for Rumors. We participated in both subtasks. The goal of subtask A is to classify the type of interaction between a rumorous social media post and a reply post as support, query, deny, or comment. The goal of subtask B is to predict the veracity of a given rumor. For subtask A, we implement a CNN-based neural architecture using ELMo embeddings of post text combined with auxiliary features and achieve a F1-score of 44.6%. For subtask B, we employ a MLP neural network leveraging our estimates for subtask A and achieve a F1-score of 30.1% (second place in the competition). We provide results and analysis of our system performance and present ablation experiments.
This paper describes our system submitted to SemEval 2019 Task 7: RumourEval 2019: Determining Rumour Veracity and Support for Rumours, Subtask A (Gorrell et al., 2019). The challenge focused on classifying whether posts from Twitter and Reddit support, deny, query, or comment a hidden rumour, truthfulness of which is the topic of an underlying discussion thread. We formulate the problem as a stance classification, determining the rumour stance of a post with respect to the previous thread post and the source thread post. The recent BERT architecture was employed to build an end-to-end system which has reached the F1 score of 61.67% on the provided test data. It finished at the 2nd place in the competition, without any hand-crafted features, only 0.2% behind the winner.
Analysing how people react to rumours associated with news in social media is an important task to prevent the spreading of misinformation, which is nowadays widely recognized as a dangerous tendency. In social media conversations, users show different stances and attitudes towards rumourous stories. Some users take a definite stance, supporting or denying the rumour at issue, while others just comment it, or ask for additional evidence related to the veracity of the rumour. On this line, a new shared task has been proposed at SemEval-2017 (Task 8, SubTask A), which is focused on rumour stance classification in English tweets. The goal is predicting user stance towards emerging rumours in Twitter, in terms of supporting, denying, querying, or commenting the original rumour, looking at the conversation threads originated by the rumour. This paper describes a new approach to this task, where the use of conversation-based and affective-based features, covering different facets of affect, has been explored. Our classification model outperforms the best-performing systems for stance classification at SemEval-2017 Task 8, showing the effectiveness of the feature set proposed.