Large language models readily memorize arbitrary training instances, such as label noise, yet they perform strikingly well on reasoning tasks. In this work, we investigate how language models memorize label noise, and why such memorization in many cases does not heavily affect generalizable reasoning capabilities. Using two controllable synthetic reasoning tasks with noisy labels, four-digit addition (FDA) and two-hop relational reasoning (THR), we discover a reliance of memorization on generalizable reasoning mechanisms: models continue to compute intermediate reasoning outputs even when retrieving memorized noisy labels, and intervening reasoning adversely affects memorization. We further show that memorization operates through distributed encoding, i.e., aggregating various inputs and intermediate results, rather than building a look-up mechanism from inputs to noisy labels. Moreover, our FDA case study reveals memorization occurs via outlier heuristics, where existing neuron activation patterns are slightly shifted to fit noisy labels. Together, our findings suggest that memorization of label noise in language models builds on, rather than overrides, the underlying reasoning mechanisms, shedding lights on the intriguing phenomenon of benign memorization.

Data perspectivism goes beyond majority vote label aggregation by recognizing various perspectives as legitimate ground truths.However, current evaluation practices remain fragmented, making it difficult to compare perspectivist approaches and analyze their impact on different users and demographic subgroups. To address this gap, we introduce PersEval, the first unified framework for evaluating perspectivist models in NLP. A key innovation is its evaluation at the individual annotator level and its treatment of annotators and users as distinct entities, consistently with real-world scenarios. We demonstrate PersEval’s capabilities through experiments with both Encoder-based and Decoder-based approaches, as well as an analysis of the effect of sociodemographic prompting. By considering global, text-, trait- and user-level evaluation metrics, we show that PersEval is a powerful tool for examining how models are influenced by user-specific information and identifying the biases this information may introduce.

Human language production exhibits remarkable richness and variation, reflecting diverse communication styles and intents. However, this variation is often overlooked in summarization evaluation. While having multiple reference summaries is known to improve correlation with human judgments, the impact of the reference set on reference-based metrics has not been systematically investigated. This work examines the sensitivity of widely used reference-based metrics in relation to the choice of reference sets, analyzing three diverse multi-reference summarization datasets: SummEval, GUMSum, and DUC2004. We demonstrate that many popular metrics exhibit significant instability. This instability is particularly concerning for n-gram-based metrics like ROUGE, where model rankings vary depending on the reference sets, undermining the reliability of model comparisons. We also collect human judgments on LLM outputs for genre-diverse data and examine their correlation with metrics to supplement existing findings beyond newswire summaries, finding weak-to-no correlation. Taken together, we recommend incorporating reference set variation into summarization evaluation to enhance consistency alongside correlation with human judgments, especially when evaluating LLMs.

Many researchers have reached the conclusion that ai models should be trained to be aware of the possibility of variation and disagreement in human judgments, and evaluated as per their ability to recognize such variation. The LeWiDi series of shared tasks on Learning With Disagreements was established to promote this approach to training and evaluating ai models, by making suitable datasets more accessible and by developing evaluation methods. The third edition of the task builds on this goal by extending the LeWiDi benchmark to four datasets spanning paraphrase identification, irony detection, sarcasm detection, and natural language inference, with labeling schemes that include not only categorical judgments as in previous editions, but ordinal judgments as well. Another novelty is that we adopt two complementary paradigms to evaluate disagreement-aware systems: the soft-label approach, in which models predict population-level distributions of judgments, and the perspectivist approach, in which models predict the interpretations of individual annotators. Crucially, we moved beyond standard metrics such as cross-entropy, and tested new evaluation metrics for the two paradigms. The task attracted diverse participation, and the results provide insights into the strengths and limitations of methods to modeling variation. Together, these contributions strengthen LeWiDi as a framework and provide new resources, benchmarks, and findings to support the development of disagreement-aware technologies.

Recently, several scholars have contributed to the growth of a new theoretical framework in NLP called perspectivism. This approach aimsto leverage data annotated by different individuals to model diverse perspectives that affect their opinions on subjective phenomena such as irony. In this context, we propose MultiPICo, a multilingual perspectivist corpus of ironic short conversations in different languages andlinguistic varieties extracted from Twitter and Reddit. The corpus includes sociodemographic information about its annotators. Our analysis of the annotated corpus shows how different demographic cohorts may significantly disagree on their annotation of irony and how certain cultural factors influence the perception of the phenomenon and the agreement on the annotation. Moreover, we show how disaggregated annotations and rich annotator metadata can be exploited to benchmark the ability of large language models to recognize irony, their positionality with respect to sociodemographic groups, and the efficacy of perspective-taking prompting for irony detection in multiple languages.

Complex linguistic phenomena such as stereotypes or irony are still challenging to detect, particularly due to the lower availability of annotated data. In this paper, we explore Back-Translation (BT) as a data augmentation method to enhance such datasets by artificially introducing semantics-preserving variations. We investigate French and Italian as source languages on two multilingual datasets annotated for the presence of stereotypes or irony and evaluate French/Italian, English, andArabic as pivot languages for the BT process. We also investigate cross-translation, i.e., augmenting one language subset of a multilingual dataset with translated instances from the other languages. We conduct an intrinsic evaluation of the quality of back-translated instances, identifying linguistic or translation model-specific errors that may occur with BT. We also perform an extrinsic evaluation of different data augmentation configurations to train a multilingual Transformer-based classifier forstereotype or irony detection on mono-lingual data.

Works in perspectivism and human label variation have emphasized the need to collect and leverage various voices and points of view in the whole Natural Language Processing pipeline.PERSEID places itself in this line of work. We consider the task of irony detection from short social media conversations in Italian collected from Twitter (X) and Reddit. To do so, we leverage data from MultiPICO, a recent multilingual dataset with disaggregated annotations and annotators’ metadata, containing 1000 Post, Reply pairs with five annotations each on average.We aim to evaluate whether prompting LLMs with additional annotators’ demographic information (namely gender only, age only, and the combination of the two) results in improved performance compared to a baseline in which only the input text is provided.The evaluation is zero-shot; and we evaluate the results on the disaggregated annotations using f1.

Gender-fair language aims at promoting gender equality by using terms and expressions that include all identities and avoid reinforcing gender stereotypes. Implementing gender-fair strategies is particularly challenging in heavily gender-marked languages, such as Italian. To address this, the Gender-Fair Generation challenge intends to help shift toward gender-fair language in written communication. The challenge, designed to assess and monitor the recognition and generation of gender-fair language in both mono- and cross-lingual scenarios, includes three tasks: (1) the detection of gendered expressions in Italian sentences, (2) the reformulation of gendered expressions into gender-fair alternatives, and (3) the generation of gender-fair language in automatic translation from English to Italian. The challenge relies on three different annotated datasets: the GFL-it corpus, which contains Italian texts extracted from administrative documents provided by the University of Brescia; GeNTE, a bilingual test set for gender-neutral rewriting and translation built upon a subset of the Europarl dataset; and Neo-GATE, a bilingual test set designed to assess the use of non-binary neomorphemes in Italian for both fairformulation and translation tasks. Finally, each task is evaluated with specific metrics: average of F1-score obtained by means of BERTScore computed on each entry of the datasets for task 1, an accuracy measured with a gender-neutral classifier, and a coverage-weighted accuracy for tasks 2 and 3.

Generating ironic content is challenging: it requires a nuanced understanding of context and implicit references and balancing seriousness and playfulness. Moreover, irony is highly subjective and can depend on various factors, such as social, cultural, or generational aspects. This paper explores whether Large Language Models (LLMs) can learn to generate ironic responses to social media posts. To do so, we fine-tune two models to generate ironic and non-ironic content and deeply analyze their outputs’ linguistic characteristics, their connection to the original post, and their similarity to the human-written replies. We also conduct a large-scale human evaluation of the outputs. Additionally, we investigate whether LLMs can learn a form of irony tied to a generational perspective, with mixed results.

Research in the field of NLP has recently focused on the variability that people show in selecting labels when performing an annotation task. Exploiting disagreements in annotations has been shown to offer advantages for accurate modelling and fair evaluation. In this paper, we propose a strongly perspectivist model for supervised classification of natural language utterances. Our approach combines the predictions of several perspective-aware models using key information of their individual confidence to capture the subjectivity encoded in the annotation of linguistic phenomena. We validate our method through experiments on two case studies, irony and hate speech detection, in in-domain and cross-domain settings. The results show that confidence-based ensembling of perspective-aware models seems beneficial for classification performance in all scenarios. In addition, we demonstrate the effectiveness of our method with automatically extracted perspectives from annotations when the annotators’ metadata are not available.

Sharing datasets and benchmarks has been crucial for rapidly improving Natural Language Processing models and systems. Documenting datasets’ characteristics (and any modification introduced over time) is equally important to avoid confusion and make comparisons reliable. Here, we describe the case of BigPatent, a dataset for patent summarization that exists in at least two rather different versions under the same name. While previous literature has not clearly distinguished among versions, their differences do not only lay on a surface level but also modify the dataset’s core nature and, thus, the complexity of the summarization task. While this paper describes a specific case, we aim to shed light on new challenges that might emerge in resource sharing and advocate for comprehensive documentation of datasets and models.

This paper is a description of our participation in the Multi-document Summarization for Literature Review (MSLR) Shared Task, in which we explore summarization models to create an automatic review of scientific results. Rather than maximizing the metrics using expensive computational models, we placed ourselves in a situation of scarce computational resources and explore the limits of a base sequence to sequence models (thus with a limited input length) to the task. Although we explore methods to feed the abstractive model with salient sentences only (using a first extractive step), we find the results still need some improvements.

This paper describes a classifier for tweets that mention medications or supplements, based on a pretrained transformer. We developed such a system for our participation in Subtask 1 of the Social Media Mining for Health Application workshop, which featured an extremely unbalanced dataset. The model showed promising results, with an F1 of 0.8 (task mean: 0.66).