Evaluating language models inherit biases through both their biases across multiple languages is critical as training and alignment processes (Feng et al., 2023; LLMs trained in one linguistic and cultural context Scherrer et al., 2024; Motoki et al., 2024). Identifying may not generalize fairly or accurately to others, the opinions and values that LLMs possess has leading to culturally inappropriate or biased outputs been a particularly intriguing area of research, as it when used globally. Our multilingual experiments carries significant sociological and quantitative implications further reveal that models exhibit different for real-world applications (Naous et al., biases in their secondary languages, such as Arabic 2023). Understanding the biases embedded in these and Chinese, which underscores the importance of powerful tools is crucial, given their widespread cross-linguistic evaluations in understanding bias use and the potential influence they may exert on resilience. Furthermore, we introduce a comprehensive users, often in unintended ways (Hartmann et al., human evaluation to compare how humans 2023) or in downstream tasks, such as content moderation.

Large language models have evolved to process multiple modalities beyond text, such as images and audio, which motivates us to explore how to effectively leverage them for graph machine learning tasks. The key question, therefore, is how to transform graphs into linear sequences of tokens, a process we term graph linearization, so that LLMs can handle graphs naturally. We consider that graphs should be linearized meaningfully to reflect certain properties of natural language text, such as local dependency and global alignment, in order to ease contemporary LLMs, trained on trillions of textual tokens, better understand graphs. To achieve this, we developed several graph linearization methods based on graph centrality, degeneracy, and node relabeling schemes. We then investigated their effect on LLM performance in graph reasoning tasks. Experimental results on synthetic graphs demonstrate the effectiveness of our methods compared to random linearization baselines. Our work introduces novel graph representations suitable for LLMs, contributing to the potential integration of graph machine learning with the trend of multi-modal processing using a unified transformer model.

The era of transfer learning has revolutionized the fields of Computer Vision and Natural Language Processing, bringing powerful pretrained models with exceptional performance across a variety of tasks. Specifically, Natural Language Processing tasks have been dominated by transformer-based language models. In Natural Language Inference and Natural Language Generation tasks, the BERT model and its variants, as well as the GPT model and its successors, demonstrated exemplary performance. However, the majority of these models are pretrained and assessed primarily for the English language or on a multilingual corpus. In this paper, we introduce GreekBART, the first Seq2Seq model based on BART-base architecture and pretrained on a large-scale Greek corpus. We evaluate and compare GreekBART against BART-random, Greek-BERT, and XLM-R on a variety of discriminative tasks. In addition, we examine its performance on two NLG tasks from GreekSUM, a newly introduced summarization dataset for the Greek language. The model, the code, and the new summarization dataset will be publicly available.

The number of senses of a given word, or polysemy, is a very subjective notion, which varies widely across annotators and resources. We propose a novel method to estimate polysemy, based on simple geometry in the contextual embedding space. Our approach is fully unsupervised and purely data-driven. We show through rigorous experiments that our rankings are well correlated (with strong statistical significance) with 6 different rankings derived from famous human-constructed resources such as WordNet, OntoNotes, Oxford, Wikipedia etc., for 6 different standard metrics. We also visualize and analyze the correlation between the human rankings. A valuable by-product of our method is the ability to sample, at no extra cost, sentences containing different senses of a given word. Finally, the fully unsupervised nature of our method makes it applicable to any language. Code and data are publicly available at https://github.com/ksipos/polysemy-assessment . The paper was accepted as a long paper at EACL 2021.

Both [1, 11] suggests "the ease of in the legal domain. We extract legal indicators from judicial access to information" is a solution to address the gap in accessing judgments to decrease the asymmetry of information of the legal justice. Access to free basic legal information could help the user system and the access-to-justice gap. We use NLP methods to extract to navigate the justice system easily, understand better the legal interesting entities/data from judgments to construct networks area his problem falls into, and choose a lawyer with experience of lawyers and judgments. We propose metrics to rank lawyers on the subject matter of the dispute. In our work, we extract and based on their experience, wins/loss ratio and their importance in represent information from past judgments to increase the transparency the network of lawyers. We also perform community detection in of judicial procedures and make them more accessible to the network of judgments and propose metrics to represent the laypersons.