Andrés Carvallo

Vladimir Araujo, Andrés Carvallo, Souvik Kundu et al.

Due to the success of pre-trained language models, versions of languages other than English have been released in recent years. This fact implies the need for resources to evaluate these models. In the case of Spanish, there are few ways to systematically assess the models' quality. In this paper, we narrow the gap by building two evaluation benchmarks. Inspired by previous work (Conneau and Kiela, 2018; Chen et al., 2019), we introduce Spanish SentEval and Spanish DiscoEval, aiming to assess the capabilities of stand-alone and discourse-aware sentence representations, respectively. Our benchmarks include considerable pre-existing and newly constructed datasets that address different tasks from various domains. In addition, we evaluate and analyze the most recent pre-trained Spanish language models to exhibit their capabilities and limitations. As an example, we discover that for the case of discourse evaluation tasks, mBERT, a language model trained on multiple languages, usually provides a richer latent representation than models trained only with documents in Spanish. We hope our contribution will motivate a fairer, more comparable, and less cumbersome way to evaluate future Spanish language models.

José Cañete, Sebastián Donoso, Felipe Bravo-Marquez et al.

In recent years there have been considerable advances in pre-trained language models, where non-English language versions have also been made available. Due to their increasing use, many lightweight versions of these models (with reduced parameters) have also been released to speed up training and inference times. However, versions of these lighter models (e.g., ALBERT, DistilBERT) for languages other than English are still scarce. In this paper we present ALBETO and DistilBETO, which are versions of ALBERT and DistilBERT pre-trained exclusively on Spanish corpora. We train several versions of ALBETO ranging from 5M to 223M parameters and one of DistilBETO with 67M parameters. We evaluate our models in the GLUES benchmark that includes various natural language understanding tasks in Spanish. The results show that our lightweight models achieve competitive results to those of BETO (Spanish-BERT) despite having fewer parameters. More specifically, our larger ALBETO model outperforms all other models on the MLDoc, PAWS-X, XNLI, MLQA, SQAC and XQuAD datasets. However, BETO remains unbeaten for POS and NER. As a further contribution, all models are publicly available to the community for future research.

Natalia Abarca, Andrés Carvallo, Claudia López Moncada et al.

The increasing use of Machine Learning (ML) in sensitive domains such as healthcare, finance, and public policy has raised concerns about the transparency of automated decisions. Explainable AI (XAI) addresses this by clarifying how models generate predictions, yet most methods demand technical expertise, limiting their value for novices. This gap is especially critical in no-code ML platforms, which seek to democratize AI but rarely include explainability. We present a human-centered XAI module in DashAI, an open-source no-code ML platform. The module integrates three complementary techniques, which are Partial Dependence Plots (PDP), Permutation Feature Importance (PFI), and KernelSHAP, into DashAI's workflow for tabular classification. A user study (N = 20; ML novices and experts) evaluated usability and the impact of explanations. Results show: (i) high task success ($\geq80\%$) across all explainability tasks; (ii) novices rated explanations as useful, accurate, and trustworthy on the Explanation Satisfaction Scale (ESS, Cronbach's $α$ = 0.74, a measure of internal consistency), while experts were more critical of sufficiency and completeness; and (iii) explanations improved perceived predictability and confidence on the Trust in Automation scale (TiA, $α$ = 0.60), with novices showing higher trust than experts. These findings highlight a central challenge for XAI in no-code ML, making explanations both accessible to novices and sufficiently detailed for experts.

Andrés Carvallo, Tamara Quiroga, Carlos Aspillaga et al.

While civilized users employ social media to stay informed and discuss daily occurrences, haters perceive these platforms as fertile ground for attacking groups and individuals. The prevailing approach to counter this phenomenon involves detecting such attacks by identifying toxic language. Effective platform measures aim to report haters and block their network access. In this context, employing hate speech detection methods aids in identifying these attacks amidst vast volumes of text, which are impossible for humans to analyze manually. In our study, we expand upon the usual hate speech detection methods, typically based on text classifiers, to develop a Named Entity Recognition (NER) System for Identity Groups. To achieve this, we created a dataset that allows extending a conventional NER to recognize identity groups. Consequently, our tool not only detects whether a sentence contains an attack but also tags the sentence tokens corresponding to the mentioned group. Results indicate that the model performs competitively in identifying groups with an average f1-score of 0.75, outperforming in identifying ethnicity attack spans with an f1-score of 0.80 compared to other identity groups. Moreover, the tool shows an outstanding generalization capability to minority classes concerning sexual orientation and gender, achieving an f1-score of 0.77 and 0.72, respectively. We tested the utility of our tool in a case study on social media, annotating and comparing comments from Facebook related to news mentioning identity groups. The case study reveals differences in the types of attacks recorded, effectively detecting named entities related to the categories of the analyzed news articles. Entities are accurately tagged within their categories, with a negligible error rate for inter-category tagging.

Vladimir Araujo, Andrés Carvallo, Carlos Aspillaga et al.

The success of pretrained word embeddings has motivated their use in the biomedical domain, with contextualized embeddings yielding remarkable results in several biomedical NLP tasks. However, there is a lack of research on quantifying their behavior under severe "stress" scenarios. In this work, we systematically evaluate three language models with adversarial examples -- automatically constructed tests that allow us to examine how robust the models are. We propose two types of stress scenarios focused on the biomedical named entity recognition (NER) task, one inspired by spelling errors and another based on the use of synonyms for medical terms. Our experiments with three benchmarks show that the performance of the original models decreases considerably, in addition to revealing their weaknesses and strengths. Finally, we show that adversarial training causes the models to improve their robustness and even to exceed the original performance in some cases.

Iván Cantador, Andrés Carvallo, Fernando Diez

The success of neural network embeddings has entailed a renewed interest in using knowledge graphs for a wide variety of machine learning and information retrieval tasks. In particular, recent recommendation methods based on graph embeddings have shown state-of-the-art performance. In general, these methods encode latent rating patterns and content features. Differently from previous work, in this paper, we propose to exploit embeddings extracted from graphs that combine information from ratings and aspect-based opinions expressed in textual reviews. We then adapt and evaluate state-of-the-art graph embedding techniques over graphs generated from Amazon and Yelp reviews on six domains, outperforming baseline recommenders. Additionally, our method has the advantage of providing explanations that involve the coverage of aspect-based opinions given by users about recommended items.

Iván Cantador, Andrés Carvallo, Fernando Diez et al.

The success of neural network embeddings has entailed a renewed interest in using knowledge graphs for a wide variety of machine learning and information retrieval tasks. In particular, current recommendation methods based on graph embeddings have shown state-of-the-art performance. These methods commonly encode latent rating patterns and content features. Different from previous work, in this paper, we propose to exploit embeddings extracted from graphs that combine information from ratings and aspect-based opinions expressed in textual reviews. We then adapt and evaluate state-of-the-art graph embedding techniques over graphs generated from Amazon and Yelp reviews on six domains, outperforming baseline recommenders. Our approach has the advantage of providing explanations which leverage aspect-based opinions given by users about recommended items. Furthermore, we also provide examples of the applicability of recommendations utilizing aspect opinions as explanations in a visualization dashboard, which allows obtaining information about the most and least liked aspects of similar users obtained from the embeddings of an input graph.

Andres Carvallo, Denis Parra, Gabriel Rada et al.

The COVID-19 has brought about a significant challenge to the whole of humanity, but with a special burden upon the medical community. Clinicians must keep updated continuously about symptoms, diagnoses, and effectiveness of emergent treatments under a never-ending flood of scientific literature. In this context, the role of evidence-based medicine (EBM) for curating the most substantial evidence to support public health and clinical practice turns essential but is being challenged as never before due to the high volume of research articles published and pre-prints posted daily. Artificial Intelligence can have a crucial role in this situation. In this article, we report the results of an applied research project to classify scientific articles to support Epistemonikos, one of the most active foundations worldwide conducting EBM. We test several methods, and the best one, based on the XLNet neural language model, improves the current approach by 93\% on average F1-score, saving valuable time from physicians who volunteer to curate COVID-19 research articles manually.

Vladimir Araujo, Andres Carvallo, Carlos Aspillaga et al.

The success of pre-trained word embeddings has motivated its use in tasks in the biomedical domain. The BERT language model has shown remarkable results on standard performance metrics in tasks such as Named Entity Recognition (NER) and Semantic Textual Similarity (STS), which has brought significant progress in the field of NLP. However, it is unclear whether these systems work seemingly well in critical domains, such as legal or medical. For that reason, in this work, we propose an adversarial evaluation scheme on two well-known datasets for medical NER and STS. We propose two types of attacks inspired by natural spelling errors and typos made by humans. We also propose another type of attack that uses synonyms of medical terms. Under these adversarial settings, the accuracy of the models drops significantly, and we quantify the extent of this performance loss. We also show that we can significantly improve the robustness of the models by training them with adversarial examples. We hope our work will motivate the use of adversarial examples to evaluate and develop models with increased robustness for medical tasks.

Carlos Aspillaga, Andrés Carvallo, Vladimir Araujo

There has been significant progress in recent years in the field of Natural Language Processing thanks to the introduction of the Transformer architecture. Current state-of-the-art models, via a large number of parameters and pre-training on massive text corpus, have shown impressive results on several downstream tasks. Many researchers have studied previous (non-Transformer) models to understand their actual behavior under different scenarios, showing that these models are taking advantage of clues or failures of datasets and that slight perturbations on the input data can severely reduce their performance. In contrast, recent models have not been systematically tested with adversarial-examples in order to show their robustness under severe stress conditions. For that reason, this work evaluates three Transformer-based models (RoBERTa, XLNet, and BERT) in Natural Language Inference (NLI) and Question Answering (QA) tasks to know if they are more robust or if they have the same flaws as their predecessors. As a result, our experiments reveal that RoBERTa, XLNet and BERT are more robust than recurrent neural network models to stress tests for both NLI and QA tasks. Nevertheless, they are still very fragile and demonstrate various unexpected behaviors, thus revealing that there is still room for future improvement in this field.