This study introduces a framework for evaluating consistency in large language model (LLM) binary text classification, addressing the lack of established reliability assessment methods. Adapting psychometric principles, we determine sample size requirements, develop metrics for invalid responses, and evaluate intra- and inter-rater reliability. Our case study examines financial news sentiment classification across 14 LLMs (including claude-3-7-sonnet, gpt-4o, deepseek-r1, gemma3, llama3.2, phi4, and command-r-plus), with five replicates per model on 1,350 articles. Models demonstrated high intra-rater consistency, achieving perfect agreement on 90-98% of examples, with minimal differences between expensive and economical models from the same families. When validated against StockNewsAPI labels, models achieved strong performance (accuracy 0.76-0.88), with smaller models like gemma3:1B, llama3.2:3B, and claude-3-5-haiku outperforming larger counterparts. All models performed at chance when predicting actual market movements, indicating task constraints rather than model limitations. Our framework provides systematic guidance for LLM selection, sample size planning, and reliability assessment, enabling organizations to optimize resources for classification tasks.
With the increasing quality and spread of LLM-based assistants, the amount of LLM-generated content is growing rapidly. In many cases and tasks, such texts are already indistinguishable from those written by humans, and the quality of generation tends to only increase. At the same time, detection methods are developing more slowly, making it challenging to prevent misuse of generative AI technologies. In this work, we investigate the task of generated text detection by proposing the GigaCheck. Our research explores two approaches: (i) distinguishing human-written texts from LLM-generated ones, and (ii) detecting LLM-generated intervals in Human-Machine collaborative texts. For the first task, our approach utilizes a general-purpose LLM, leveraging its extensive language abilities to fine-tune efficiently for the downstream task of LLM-generated text detection, achieving high performance even with limited data. For the second task, we propose a novel approach that combines computer vision and natural language processing techniques. Specifically, we use a fine-tuned general-purpose LLM in conjunction with a DETR-like detection model, adapted from computer vision, to localize AI-generated intervals within text. We evaluate the GigaCheck on five classification datasets with English texts and three datasets designed for Human-Machine collaborative text analysis. Our results demonstrate that GigaCheck outperforms previous methods, even in out-of-distribution settings, establishing a strong baseline across all datasets.
Quantifying the uncertainty of predictions made by large language models (LLMs) in binary text classification tasks remains a challenge. Calibration, in the context of LLMs, refers to the alignment between the model's predicted probabilities and the actual correctness of its predictions. A well-calibrated model should produce probabilities that accurately reflect the likelihood of its predictions being correct. We propose a novel approach that utilizes the inductive Venn--Abers predictor (IVAP) to calibrate the probabilities associated with the output tokens corresponding to the binary labels. Our experiments on the BoolQ dataset using the Llama 2 model demonstrate that IVAP consistently outperforms the commonly used temperature scaling method for various label token choices, achieving well-calibrated probabilities while maintaining high predictive quality. Our findings contribute to the understanding of calibration techniques for LLMs and provide a practical solution for obtaining reliable uncertainty estimates in binary question answering tasks, enhancing the interpretability and trustworthiness of LLM predictions.
Hoaxes are a recognised form of disinformation created deliberately, with potential serious implications in the credibility of reference knowledge resources such as Wikipedia. What makes detecting Wikipedia hoaxes hard is that they often are written according to the official style guidelines. In this work, we first provide a systematic analysis of similarities and discrepancies between legitimate and hoax Wikipedia articles, and introduce Hoaxpedia, a collection of 311 hoax articles (from existing literature and official Wikipedia lists), together with semantically similar legitimate articles, which together form a binary text classification dataset aimed at fostering research in automated hoax detection. In this paper, We report results after analyzing several language models, hoax-to-legit ratios, and the amount of text classifiers are exposed to (full article vs the article's definition alone). Our results suggest that detecting deceitful content in Wikipedia based on content alone is hard but feasible, and complement our analysis with a study on the differences in distributions in edit histories, and find that looking at this feature yields better classification results than context.
With the rapid development and widespread application of Large Language Models (LLMs), the use of Machine-Generated Text (MGT) has become increasingly common, bringing with it potential risks, especially in terms of quality and integrity in fields like news, education, and science. Current research mainly focuses on purely MGT detection without adequately addressing mixed scenarios, including AI-revised Human-Written Text (HWT) or human-revised MGT. To tackle this challenge, we define mixtext, a form of mixed text involving both AI and human-generated content. Then, we introduce MixSet, the first dataset dedicated to studying these mixtext scenarios. Leveraging MixSet, we executed comprehensive experiments to assess the efficacy of prevalent MGT detectors in handling mixtext situations, evaluating their performance in terms of effectiveness, robustness, and generalization. Our findings reveal that existing detectors struggle to identify mixtext, particularly in dealing with subtle modifications and style adaptability. This research underscores the urgent need for more fine-grain detectors tailored for mixtext, offering valuable insights for future research. Code and Models are available at https://github.com/Dongping-Chen/MixSet.
The Forum for Information Retrieval (FIRE) started a shared task this year for classification of comments of different code segments. This is binary text classification task where the objective is to identify whether comments given for certain code segments are relevant or not. The BioNLP-IISERB group at the Indian Institute of Science Education and Research Bhopal (IISERB) participated in this task and submitted five runs for five different models. The paper presents the overview of the models and other significant findings on the training corpus. The methods involve different feature engineering schemes and text classification techniques. The performance of the classical bag of words model and transformer-based models were explored to identify significant features from the given training corpus. We have explored different classifiers viz., random forest, support vector machine and logistic regression using the bag of words model. Furthermore, the pre-trained transformer based models like BERT, RoBERT and ALBERT were also used by fine-tuning them on the given training corpus. The performance of different such models over the training corpus were reported and the best five models were implemented on the given test corpus. The empirical results show that the bag of words model outperforms the transformer based models, however, the performance of our runs are not reasonably well in both training and test corpus. This paper also addresses the limitations of the models and scope for further improvement.
We introduce Ghostbuster, a state-of-the-art system for detecting AI-generated text. Our method works by passing documents through a series of weaker language models, running a structured search over possible combinations of their features, and then training a classifier on the selected features to predict whether documents are AI-generated. Crucially, Ghostbuster does not require access to token probabilities from the target model, making it useful for detecting text generated by black-box models or unknown model versions. In conjunction with our model, we release three new datasets of human- and AI-generated text as detection benchmarks in the domains of student essays, creative writing, and news articles. We compare Ghostbuster to a variety of existing detectors, including DetectGPT and GPTZero, as well as a new RoBERTa baseline. Ghostbuster achieves 99.0 F1 when evaluated across domains, which is 5.9 F1 higher than the best preexisting model. It also outperforms all previous approaches in generalization across writing domains (+7.5 F1), prompting strategies (+2.1 F1), and language models (+4.4 F1). We also analyze the robustness of our system to a variety of perturbations and paraphrasing attacks and evaluate its performance on documents written by non-native English speakers.
Large language models (LLMs) have achieved human-level text generation, emphasizing the need for effective AI-generated text detection to mitigate risks like the spread of fake news and plagiarism. Existing research has been constrained by evaluating detection methods on specific domains or particular language models. In practical scenarios, however, the detector faces texts from various domains or LLMs without knowing their sources. To this end, we build a comprehensive testbed by gathering texts from diverse human writings and texts generated by different LLMs. Empirical results show challenges in distinguishing machine-generated texts from human-authored ones across various scenarios, especially out-of-distribution. These challenges are due to the decreasing linguistic distinctions between the two sources. Despite challenges, the top-performing detector can identify 86.54% out-of-domain texts generated by a new LLM, indicating the feasibility for application scenarios. We release our resources at https://github.com/yafuly/MAGE.
This study evaluates the robustness of two state-of-the-art deep contextual language representations, ELMo and DistilBERT, on supervised learning of binary protest news classification and sentiment analysis of product reviews. A "cross-context" setting is enabled using test sets that are distinct from the training data. Specifically, in the news classification task, the models are developed on local news from India and tested on the local news from China. In the sentiment analysis task, the models are trained on movie reviews and tested on customer reviews. This comparison is aimed at exploring the limits of the representative power of today's Natural Language Processing systems on the path to the systems that are generalizable to real-life scenarios. The models are fine-tuned and fed into a Feed-Forward Neural Network and a Bidirectional Long Short Term Memory network. Multinomial Naive Bayes and Linear Support Vector Machine are used as traditional baselines. The results show that, in binary text classification, DistilBERT is significantly better than ELMo on generalizing to the cross-context setting. ELMo is observed to be significantly more robust to the cross-context test data than both baselines. On the other hand, the baselines performed comparably well to ELMo when the training and test data are subsets of the same corpus (no cross-context). DistilBERT is also found to be 30% smaller and 83% faster than ELMo. The results suggest that DistilBERT can transfer generic semantic knowledge to other domains better than ELMo. DistilBERT is also favorable in incorporating into real-life systems for it requires a smaller computational training budget. When generalization is not the utmost preference and test domain is similar to the training domain, the traditional ML algorithms can still be considered as more economic alternatives to deep language representations.
Recent works in Explainable AI mostly address the transparency issue of black-box models or create explanations for any kind of models (i.e., they are model-agnostic), while leaving explanations of interpretable models largely underexplored. In this paper, we fill this gap by focusing on explanations for a specific interpretable model, namely pattern-based logistic regression (PLR) for binary text classification. We do so because, albeit interpretable, PLR is challenging when it comes to explanations. In particular, we found that a standard way to extract explanations from this model does not consider relations among the features, making the explanations hardly plausible to humans. Hence, we propose AXPLR, a novel explanation method using (forms of) computational argumentation to generate explanations (for outputs computed by PLR) which unearth model agreements and disagreements among the features. Specifically, we use computational argumentation as follows: we see features (patterns) in PLR as arguments in a form of quantified bipolar argumentation frameworks (QBAFs) and extract attacks and supports between arguments based on specificity of the arguments; we understand logistic regression as a gradual semantics for these QBAFs, used to determine the arguments' dialectic strength; and we study standard properties of gradual semantics for QBAFs in the context of our argumentative re-interpretation of PLR, sanctioning its suitability for explanatory purposes. We then show how to extract intuitive explanations (for outputs computed by PLR) from the constructed QBAFs. Finally, we conduct an empirical evaluation and two experiments in the context of human-AI collaboration to demonstrate the advantages of our resulting AXPLR method.
Recent progress in generative language models has enabled machines to generate astonishingly realistic texts. While there are many legitimate applications of such models, there is also a rising need to distinguish machine-generated texts from human-written ones (e.g., fake news detection). However, to our best knowledge, there is currently no benchmark environment with datasets and tasks to systematically study the so-called "Turing Test" problem for neural text generation methods. In this work, we present the TuringBench benchmark environment, which is comprised of (1) a dataset with 200K human- or machine-generated samples across 20 labels {Human, GPT-1, GPT-2_small, GPT-2_medium, GPT-2_large, GPT-2_xl, GPT-2_PyTorch, GPT-3, GROVER_base, GROVER_large, GROVER_mega, CTRL, XLM, XLNET_base, XLNET_large, FAIR_wmt19, FAIR_wmt20, TRANSFORMER_XL, PPLM_distil, PPLM_gpt2}, (2) two benchmark tasks -- i.e., Turing Test (TT) and Authorship Attribution (AA), and (3) a website with leaderboards. Our preliminary experimental results using TuringBench show that FAIR_wmt20 and GPT-3 are the current winners, among all language models tested, in generating the most human-like indistinguishable texts with the lowest F1 score by five state-of-the-art TT detection models. The TuringBench is available at: https://turingbench.ist.psu.edu/
This paper presents our models for WNUT 2020 shared task2. The shared task2 involves identification of COVID-19 related informative tweets. We treat this as binary text classification problem and experiment with pre-trained language models. Our first model which is based on CT-BERT achieves F1-score of 88.7% and second model which is an ensemble of CT-BERT, RoBERTa and SVM achieves F1-score of 88.52%.
The recent advances in language modeling significantly improved the generative capabilities of deep neural models: in 2019 OpenAI released GPT-2, a pre-trained language model that can autonomously generate coherent, non-trivial and human-like text samples. Since then, ever more powerful text generative models have been developed. Adversaries can exploit these tremendous generative capabilities to enhance social bots that will have the ability to write plausible deepfake messages, hoping to contaminate public debate. To prevent this, it is crucial to develop deepfake social media messages detection systems. However, to the best of our knowledge no one has ever addressed the detection of machine-generated texts on social networks like Twitter or Facebook. With the aim of helping the research in this detection field, we collected the first dataset of \real deepfake tweets, TweepFake. It is real in the sense that each deepfake tweet was actually posted on Twitter. We collected tweets from a total of 23 bots, imitating 17 human accounts. The bots are based on various generation techniques, i.e., Markov Chains, RNN, RNN+Markov, LSTM, GPT-2. We also randomly selected tweets from the humans imitated by the bots to have an overall balanced dataset of 25,572 tweets (half human and half bots generated). The dataset is publicly available on Kaggle. Lastly, we evaluated 13 deepfake text detection methods (based on various state-of-the-art approaches) to both demonstrate the challenges that Tweepfake poses and create a solid baseline of detection techniques. We hope that TweepFake can offer the opportunity to tackle the deepfake detection on social media messages as well.
Legal judgment prediction is the task of automatically predicting the outcome of a court case, given a text describing the case's facts. Previous work on using neural models for this task has focused on Chinese; only feature-based models (e.g., using bags of words and topics) have been considered in English. We release a new English legal judgment prediction dataset, containing cases from the European Court of Human Rights. We evaluate a broad variety of neural models on the new dataset, establishing strong baselines that surpass previous feature-based models in three tasks: (1) binary violation classification; (2) multi-label classification; (3) case importance prediction. We also explore if models are biased towards demographic information via data anonymization. As a side-product, we propose a hierarchical version of BERT, which bypasses BERT's length limitation.
Employers actively look for talents having not only specific hard skills but also various soft skills. To analyze the soft skill demands on the job market, it is important to be able to detect soft skill phrases from job advertisements automatically. However, a naive matching of soft skill phrases can lead to false positive matches when a soft skill phrase, such as friendly, is used to describe a company, a team, or another entity, rather than a desired candidate. In this paper, we propose a phrase-matching-based approach which differentiates between soft skill phrases referring to a candidate vs. something else. The disambiguation is formulated as a binary text classification problem where the prediction is made for the potential soft skill based on the context where it occurs. To inform the model about the soft skill for which the prediction is made, we develop several approaches, including soft skill masking and soft skill tagging. We compare several neural network based approaches, including CNN, LSTM and Hierarchical Attention Model. The proposed tagging-based input representation using LSTM achieved the highest recall of 83.92% on the job dataset when fixing a precision to 95%.