Evidence retrieval is a core part of automatic fact-checking. Prior work makes simplifying assumptions in retrieval that depart from real-world use cases: either no access to evidence, access to evidence curated by a human fact-checker, or access to evidence available long after the claim has been made. In this work, we present the first fully automated pipeline to check real-world claims by retrieving raw evidence from the web. We restrict our retriever to only search documents available prior to the claim's making, modeling the realistic scenario where an emerging claim needs to be checked. Our pipeline includes five components: claim decomposition, raw document retrieval, fine-grained evidence retrieval, claim-focused summarization, and veracity judgment. We conduct experiments on complex political claims in the ClaimDecomp dataset and show that the aggregated evidence produced by our pipeline improves veracity judgments. Human evaluation finds the evidence summary produced by our system is reliable (it does not hallucinate information) and relevant to answering key questions about a claim, suggesting that it can assist fact-checkers even when it cannot surface a complete evidence set.
Recent developments in large language models (LLMs) have been impressive. However, these models sometimes show inconsistencies and problematic behavior, such as hallucinating facts, generating flawed code, or creating offensive and toxic content. Unlike these models, humans typically utilize external tools to cross-check and refine their initial content, like using a search engine for fact-checking, or a code interpreter for debugging. Inspired by this observation, we introduce a framework called CRITIC that allows LLMs, which are essentially "black boxes" to validate and progressively amend their own outputs in a manner similar to human interaction with tools. More specifically, starting with an initial output, CRITIC interacts with appropriate tools to evaluate certain aspects of the text, and then revises the output based on the feedback obtained during this validation process. Comprehensive evaluations involving free-form question answering, mathematical program synthesis, and toxicity reduction demonstrate that CRITIC consistently enhances the performance of LLMs. Meanwhile, our research highlights the crucial importance of external feedback in promoting the ongoing self-improvement of LLMs.
In this highly digitised world, fake news is a challenging problem that can cause serious harm to society. Considering how fast fake news can spread, automated methods, tools and services for assisting users to do fact-checking (i.e., fake news detection) become necessary and helpful, for both professionals, such as journalists and researchers, and the general public such as news readers. Experts, especially researchers, play an essential role in informing people about truth and facts, which makes them a good proxy for non-experts to detect fake news by checking relevant expert opinions and comments. Therefore, in this paper, we present aedFaCT, a web browser extension that can help professionals and news readers perform fact-checking via the automatic discovery of expert opinions relevant to the news of concern via shared keywords. Our initial evaluation with three independent testers (who did not participate in the development of the extension) indicated that aedFaCT can provide a faster experience to its users compared with traditional fact-checking practices based on manual online searches, without degrading the quality of retrieved evidence for fact-checking. The source code of aedFaCT is publicly available at https://github.com/altuncu/aedFaCT.
A recent focus of large language model (LLM) development, as exemplified by generative search engines, is to incorporate external references to generate and support its claims. However, evaluating the attribution, i.e., verifying whether the generated statement is fully supported by the cited reference, remains an open problem. Although human evaluation is common practice, it is costly and time-consuming. In this paper, we investigate the automatic evaluation of attribution given by LLMs. We begin by defining different types of attribution errors, and then explore two approaches for automatic evaluation: prompting LLMs and fine-tuning smaller LMs. The fine-tuning data is repurposed from related tasks such as question answering, fact-checking, natural language inference, and summarization. We manually curate a set of test examples covering 12 domains from a generative search engine, New Bing. Our results on this curated test set and simulated examples from existing benchmarks highlight both promising signals and challenges. We hope our problem formulation, testbeds, and findings will help lay the foundation for future studies on this important problem.
To enhance the ability to find credible evidence in news articles, we propose a novel task of expert recommendation, which aims to identify trustworthy experts on a specific news topic. To achieve the aim, we describe the construction of a novel NewsQuote dataset consisting of 24,031 quote-speaker pairs that appeared on a COVID-19 news corpus. We demonstrate an automatic pipeline for speaker and quote extraction via a BERT-based Question Answering model. Then, we formulate expert recommendations as document retrieval task by retrieving relevant quotes first as an intermediate step for expert identification, and expert retrieval by directly retrieving sources based on the probability of a query conditional on a candidate expert. Experimental results on NewsQuote show that document retrieval is more effective in identifying relevant experts for a given news topic compared to expert retrieval
Making the content generated by Large Language Model (LLM), accurate, credible and traceable is crucial, especially in complex knowledge-intensive tasks that require multi-step reasoning and each step needs knowledge to solve. Retrieval-augmented generation is good potential to solve this problem. However, where and how to introduce Information Retrieval (IR) to LLM is a big challenge. Previous work has the problems that wrong knowledge retrieved by IR misleads the LLM and interaction between IR and LLM breaks the reasoning chain of LLM. This paper proposes a novel framework named \textbf{Search-in-the-Chain} (SearChain) for the interaction between LLM and IR to solve the challenges. First, LLM generates the reasoning chain named Chain-of-Query (CoQ) where each node consists of an IR-oriented query-answer pair. Second, IR verifies the answer of each node of CoQ. It corrects the answer that is not consistent with the retrieved information when IR gives high confidence, which improves the credibility. Third, LLM can indicate its missing knowledge in CoQ and rely on IR to provide this knowledge to LLM. These operations improve the accuracy in terms of reasoning and knowledge. Finally, SearChain generates the reasoning process and marks references to supporting documents for each reasoning step, which improves traceability. Interaction with IR in SearChain forms a novel reasoning path based on a tree, which enables LLM to dynamically modify the direction of reasoning. Experiments show that SearChain outperforms state-of-the-art baselines on complex knowledge-intensive tasks including multi-hop Q\&A, slot filling, fact checking, and long-form Q\&A.
Automated fact-checking is often presented as an epistemic tool that fact-checkers, social media consumers, and other stakeholders can use to fight misinformation. Nevertheless, few papers thoroughly discuss how. We document this by analysing 100 highly-cited papers, and annotating epistemic elements related to intended use, i.e., means, ends, and stakeholders. We find that narratives leaving out some of these aspects are common, that many papers propose inconsistent means and ends, and that the feasibility of suggested strategies rarely has empirical backing. We argue that this vagueness actively hinders the technology from reaching its goals, as it encourages overclaiming, limits criticism, and prevents stakeholder feedback. Accordingly, we provide several recommendations for thinking and writing about the use of fact-checking artefacts.
The internet gives the world an open platform to express their views and share their stories. While this is very valuable, it makes fake news one of our society's most pressing problems. Manual fact checking process is time consuming, which makes it challenging to disprove misleading assertions before they cause significant harm. This is he driving interest in automatic fact or claim verification. Some of the existing datasets aim to support development of automating fact-checking techniques, however, most of them are text based. Multi-modal fact verification has received relatively scant attention. In this paper, we provide a multi-modal fact-checking dataset called FACTIFY 2, improving Factify 1 by using new data sources and adding satire articles. Factify 2 has 50,000 new data instances. Similar to FACTIFY 1.0, we have three broad categories - support, no-evidence, and refute, with sub-categories based on the entailment of visual and textual data. We also provide a BERT and Vison Transformer based baseline, which achieves 65% F1 score in the test set. The baseline codes and the dataset will be made available at https://github.com/surya1701/Factify-2.0.
Despite recent concerns about undesirable behaviors generated by large language models (LLMs), including non-factual, biased, and hateful language, we find LLMs are inherent multi-task language checkers based on their latent representations of natural and social knowledge. We present an interpretable, unified, language checking (UniLC) method for both human and machine-generated language that aims to check if language input is factual and fair. While fairness and fact-checking tasks have been handled separately with dedicated models, we find that LLMs can achieve high performance on a combination of fact-checking, stereotype detection, and hate speech detection tasks with a simple, few-shot, unified set of prompts. With the ``1/2-shot'' multi-task language checking method proposed in this work, the GPT3.5-turbo model outperforms fully supervised baselines on several language tasks. The simple approach and results suggest that based on strong latent knowledge representations, an LLM can be an adaptive and explainable tool for detecting misinformation, stereotypes, and hate speech.
Search engines increasingly leverage large language models (LLMs) to generate direct answers, and AI chatbots now access the Internet for fresh data. As information curators for billions of users, LLMs must assess the accuracy and reliability of different sources. This paper audits nine widely used LLMs from three leading providers -- OpenAI, Google, and Meta -- to evaluate their ability to discern credible and high-quality information sources from low-credibility ones. We find that while LLMs can rate most tested news outlets, larger models more frequently refuse to provide ratings due to insufficient information, whereas smaller models are more prone to making errors in their ratings. For sources where ratings are provided, LLMs exhibit a high level of agreement among themselves (average Spearman's $\rho = 0.79$), but their ratings align only moderately with human expert evaluations (average $\rho = 0.50$). Analyzing news sources with different political leanings in the US, we observe a liberal bias in credibility ratings yielded by all LLMs in default configurations. Additionally, assigning partisan roles to LLMs consistently induces strong politically congruent bias in their ratings. These findings have important implications for the use of LLMs in curating news and political information.
Generative Large Language Models (LLMs) such as GPT-3 are capable of generating highly fluent responses to a wide variety of user prompts. However, LLMs are known to hallucinate facts and make non-factual statements which can undermine trust in their output. Existing fact-checking approaches either require access to the output probability distribution (which may not be available for systems such as ChatGPT) or external databases that are interfaced via separate, often complex, modules. In this work, we propose "SelfCheckGPT", a simple sampling-based approach that can be used to fact-check the responses of black-box models in a zero-resource fashion, i.e. without an external database. SelfCheckGPT leverages the simple idea that if an LLM has knowledge of a given concept, sampled responses are likely to be similar and contain consistent facts. However, for hallucinated facts, stochastically sampled responses are likely to diverge and contradict one another. We investigate this approach by using GPT-3 to generate passages about individuals from the WikiBio dataset, and manually annotate the factuality of the generated passages. We demonstrate that SelfCheckGPT can: i) detect non-factual and factual sentences; and ii) rank passages in terms of factuality. We compare our approach to several baselines and show that our approach has considerably higher AUC-PR scores in sentence-level hallucination detection and higher correlation scores in passage-level factuality assessment compared to grey-box methods.
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.
The spread of online misinformation threatens public health, democracy, and the broader society. While professional fact-checkers form the first line of defense by fact-checking popular false claims, they do not engage directly in conversations with misinformation spreaders. On the other hand, non-expert ordinary users act as eyes-on-the-ground who proactively counter misinformation -- recent research has shown that 96% counter-misinformation responses are made by ordinary users. However, research also found that 2/3 times, these responses are rude and lack evidence. This work seeks to create a counter-misinformation response generation model to empower users to effectively correct misinformation. This objective is challenging due to the absence of datasets containing ground-truth of ideal counter-misinformation responses, and the lack of models that can generate responses backed by communication theories. In this work, we create two novel datasets of misinformation and counter-misinformation response pairs from in-the-wild social media and crowdsourcing from college-educated students. We annotate the collected data to distinguish poor from ideal responses that are factual, polite, and refute misinformation. We propose MisinfoCorrect, a reinforcement learning-based framework that learns to generate counter-misinformation responses for an input misinformation post. The model rewards the generator to increase the politeness, factuality, and refutation attitude while retaining text fluency and relevancy. Quantitative and qualitative evaluation shows that our model outperforms several baselines by generating high-quality counter-responses. This work illustrates the promise of generative text models for social good -- here, to help create a safe and reliable information ecosystem. The code and data is accessible on https://github.com/claws-lab/MisinfoCorrect.
Textual entailment models are increasingly applied in settings like fact-checking, presupposition verification in question answering, or summary evaluation. However, these represent a significant domain shift from existing entailment datasets, and models underperform as a result. We propose WiCE, a new fine-grained textual entailment dataset built on natural claim and evidence pairs extracted from Wikipedia. In addition to standard claim-level entailment, WiCE provides entailment judgments over sub-sentence units of the claim, and a minimal subset of evidence sentences that support each subclaim. To support this, we propose an automatic claim decomposition strategy using GPT-3.5 which we show is also effective at improving entailment models' performance on multiple datasets at test time. Finally, we show that real claims in our dataset involve challenging verification and retrieval problems that existing models fail to address.
Leveraging contextual knowledge has become standard practice in automated claim verification, yet the impact of temporal reasoning has been largely overlooked. Our study demonstrates that time positively influences the claim verification process of evidence-based fact-checking. The temporal aspects and relations between claims and evidence are first established through grounding on shared timelines, which are constructed using publication dates and time expressions extracted from their text. Temporal information is then provided to RNN-based and Transformer-based classifiers before or after claim and evidence encoding. Our time-aware fact-checking models surpass base models by up to 9% Micro F1 (64.17%) and 15% Macro F1 (47.43%) on the MultiFC dataset. They also outperform prior methods that explicitly model temporal relations between evidence. Our findings show that the presence of temporal information and the manner in which timelines are constructed greatly influence how fact-checking models determine the relevance and supporting or refuting character of evidence documents.
We introduce a new benchmark, COVID-VTS, for fact-checking multi-modal information involving short-duration videos with COVID19- focused information from both the real world and machine generation. We propose, TwtrDetective, an effective model incorporating cross-media consistency checking to detect token-level malicious tampering in different modalities, and generate explanations. Due to the scarcity of training data, we also develop an efficient and scalable approach to automatically generate misleading video posts by event manipulation or adversarial matching. We investigate several state-of-the-art models and demonstrate the superiority of TwtrDetective.
Automated news credibility and fact-checking at scale require accurately predicting news factuality and media bias. This paper introduces a large sentence-level dataset, titled "FactNews", composed of 6,191 sentences expertly annotated according to factuality and media bias definitions proposed by AllSides. We use FactNews to assess the overall reliability of news sources, by formulating two text classification problems for predicting sentence-level factuality of news reporting and bias of media outlets. Our experiments demonstrate that biased sentences present a higher number of words compared to factual sentences, besides having a predominance of emotions. Hence, the fine-grained analysis of subjectivity and impartiality of news articles provided promising results for predicting the reliability of media outlets. Finally, due to the severity of fake news and political polarization in Brazil, and the lack of research for Portuguese, both dataset and baseline were proposed for Brazilian Portuguese.
We present a human-in-the-loop evaluation framework for fact-checking novel misinformation claims and identifying social media messages that support them. Our approach extracts check-worthy claims, which are aggregated and ranked for review. Stance classifiers are then used to identify tweets supporting novel misinformation claims, which are further reviewed to determine whether they violate relevant policies. To demonstrate the feasibility of our approach, we develop a baseline system based on modern NLP methods for human-in-the-loop fact-checking in the domain of COVID-19 treatments. We make our data and detailed annotation guidelines available to support the evaluation of human-in-the-loop systems that identify novel misinformation directly from raw user-generated content.
Fact verification has attracted a lot of research attention recently, e.g., in journalism, marketing, and policymaking, as misinformation and disinformation online can sway one's opinion and affect one's actions. While fact-checking is a hard task in general, in many cases, false statements can be easily debunked based on analytics over tables with reliable information. Hence, table-based fact verification has recently emerged as an important and growing research area. Yet, progress has been limited due to the lack of datasets that can be used to pre-train language models (LMs) to be aware of common table operations, such as aggregating a column or comparing tuples. To bridge this gap, in this paper we introduce PASTA, a novel state-of-the-art framework for table-based fact verification via pre-training with synthesized sentence-table cloze questions. In particular, we design six types of common sentence-table cloze tasks, including Filter, Aggregation, Superlative, Comparative, Ordinal, and Unique, based on which we synthesize a large corpus consisting of 1.2 million sentence-table pairs from WikiTables. PASTA uses a recent pre-trained LM, DeBERTaV3, and further pretrains it on our corpus. Our experimental results show that PASTA achieves new state-of-the-art performance on two table-based fact verification benchmarks: TabFact and SEM-TAB-FACTS. In particular, on the complex set of TabFact, which contains multiple operations, PASTA largely outperforms the previous state of the art by 4.7 points (85.6% vs. 80.9%), and the gap between PASTA and human performance on the small TabFact test set is narrowed to just 1.5 points (90.6% vs. 92.1%).
Misinformation emerges in times of uncertainty when credible information is limited. This is challenging for NLP-based fact-checking as it relies on counter-evidence, which may not yet be available. Despite increasing interest in automatic fact-checking, it is still unclear if automated approaches can realistically refute harmful real-world misinformation. Here, we contrast and compare NLP fact-checking with how professional fact-checkers combat misinformation in the absence of counter-evidence. In our analysis, we show that, by design, existing NLP task definitions for fact-checking cannot refute misinformation as professional fact-checkers do for the majority of claims. We then define two requirements that the evidence in datasets must fulfill for realistic fact-checking: It must be (1) sufficient to refute the claim and (2) not leaked from existing fact-checking articles. We survey existing fact-checking datasets and find that all of them fail to satisfy both criteria. Finally, we perform experiments to demonstrate that models trained on a large-scale fact-checking dataset rely on leaked evidence, which makes them unsuitable in real-world scenarios. Taken together, we show that current NLP fact-checking cannot realistically combat real-world misinformation because it depends on unrealistic assumptions about counter-evidence in the data.