A well-known issue with Retrieval Augmented Generation (RAG) is that retrieved passages that are irrelevant to the query sometimes distract the answer-generating LLM, causing it to provide an incorrect response. In this paper, we shed light on this core issue and formulate the distracting effect of a passage w.r.t. a query (and an LLM). We provide a quantifiable measure of the distracting effect of a passage and demonstrate its robustness across LLMs. Our research introduces novel methods for identifying and using hard distracting passages to improve RAG systems. By fine-tuning LLMs with these carefully selected distracting passages, we achieve up to a 7.5% increase in answering accuracy compared to counterparts fine-tuned on conventional RAG datasets. Our contribution is two-fold: first, we move beyond the simple binary classification of irrelevant passages as either completely unrelated vs. distracting, and second, we develop and analyze multiple methods for finding hard distracting passages. To our knowledge, no other research has provided such a comprehensive framework for identifying and utilizing hard distracting passages.

We introduce DataMorgana, a tool for generating synthetic Q&A benchmarks tailored to RAG applications in enterprise settings. DataMorgana enables customization of the generated benchmark according to the expected diverse traffic of the RAG application. It allows for specifying question types and their associated distribution via a lightweight configuration mechanism. We demonstrate via a series of quantitative and qualitative experiments that DataMorgana surpasses existing tools in terms of lexical, syntactic, and semantic diversity of the generated benchmark while maintaining high quality. We run our experiments over domain-specific and general-knowledge public datasets, as well as two private datasets from governmental RAG applications: one for citizens and the other for government employees. The private datasets have been shared with us by AI71, an AI company, which has integrated DataMorgana into its offerings. In addition, DataMorgana has been offered to about 150 researchers worldwide as part of the SIGIR’2025 LiveRAG Challenge held in Spring 2025.

Retrieval Augmented Generation enhances LLM accuracy by adding passages retrieved from an external corpus to the LLM prompt. This paper investigates how positional bias - the tendency of LLMs to weight information differently based on its position in the prompt - affects not only the LLM’s capability to capitalize on relevant passages, but also its susceptibility to distracting passages. Through extensive experiments on three benchmarks, we show how state-of-the-art retrieval pipelines, while attempting to retrieve relevant passages, systematically bring highly distracting ones to the top ranks, with over 60% of queries containing at least one highly distracting passage among the top-10 retrieved passages. As a result, the impact of the LLM positional bias, which in controlled settings is often reported as very prominent by related works, is actually marginal in real scenarios since both relevant and distracting passages are, in turn, penalized. Indeed, our findings reveal that sophisticated strategies that attempt to rearrange the passages based on LLM positional preferences do not perform better than random shuffling.

Large Language Models (LLMs) have demonstrated excellent capabilities in Question Answering (QA) tasks, yet their ability to identify and address ambiguous questions remains underdeveloped. Ambiguities in user queries often lead to inaccurate or misleading answers, undermining user trust in these systems. Despite prior attempts using prompt-based methods, performance has largely been equivalent to random guessing, leaving a significant gap in effective ambiguity detection. To address this, we propose a novel framework for detecting ambiguous questions within LLM-based QA systems. We first prompt an LLM to generate multiple answers to a question, and then analyze them to infer the ambiguity. We propose to use a lightweight Random Forest model, trained on a bootstrapped and shuffled 6-shot examples dataset. Experimental results on ASQA, PACIFIC, and ABG-COQA datasets demonstrate the effectiveness of our approach, with accuracy up to 70.8%. Furthermore, our framework enhances the confidence calibration of LLM outputs, leading to more trustworthy QA systems able to handle complex questions.

Large Language Models (LLMs) operating in 0-shot or few-shot settings achieve competitive results in Text Classification tasks. In-Context Learning (ICL) typically achieves better accuracy than the 0-shot setting, but it pays in terms of efficiency, due to the longer input prompt. In this paper, we propose a strategy to make LLMs as efficient as 0-shot text classifiers, while getting comparable or better accuracy than ICL. Our solution targets the low resource setting, i.e., when only 4 examples per class are available. Using a single LLM and few-shot real data we perform a sequence of generation, filtering and Parameter-Efficient Fine-Tuning steps to create a robust and efficient classifier. Experimental results show that our approach leads to competitive results on multiple text classification datasets.

Methods to generate text from structured data have advanced significantly in recent years, primarily due to fine-tuning of pre-trained language models on large datasets. However, such models can fail to produce output faithful to the input data, particularly on out-of-domain data. Sufficient annotated data is often not available for specific domains, leading us to seek an unsupervised approach to improve the faithfulness of output text. Since the problem is fundamentally one of consistency between the representations of the structured data and text, we evaluate the effectiveness of cycle training in this work. Cycle training uses two models which are inverses of each other: one that generates text from structured data, and one which generates the structured data from natural language text. We show that cycle training, when initialized with a small amount of supervised data (100 samples in our case), achieves nearly the same performance as fully supervised approaches for the data-to-text generation task on the WebNLG, E2E, WTQ, and WSQL datasets. We perform extensive empirical analysis with automated evaluation metrics and a newly designed human evaluation schema to reveal different cycle training strategies’ effectiveness of reducing various types of generation errors. Our code is publicly available at https://github.com/Edillower/CycleNLG.

Even if recent Transformer-based architectures, such as BERT, achieved impressive results in semantic processing tasks, their fine-tuning stage still requires large scale training resources. Usually, Data Augmentation (DA) techniques can help to deal with low resource settings. In Text Classification tasks, the objective of DA is the generation of well-formed sentences that i) represent the desired task category and ii) are novel with respect to existing sentences. In this paper, we propose a neural approach to automatically learn to generate new examples using a pre-trained sequence-to-sequence model. We first learn a task-oriented similarity function that we use to pair similar examples. Then, we use these example pairs to train a model to generate examples. Experiments in low resource settings show that augmenting the training material with the proposed strategy systematically improves the results on text classification and natural language inference tasks by up to 10% accuracy, outperforming existing DA approaches.

In real scenarios, a multilingual model trained to solve NLP tasks on a set of languages can be required to support new languages over time. Unfortunately, the straightforward retraining on a dataset containing annotated examples for all the languages is both expensive and time-consuming, especially when the number of target languages grows. Moreover, the original annotated material may no longer be available due to storage or business constraints. Re-training only with the new language data will inevitably result in Catastrophic Forgetting of previously acquired knowledge. We propose a Continual Learning strategy that updates a model to support new languages over time, while maintaining consistent results on previously learned languages. We define a Teacher-Student framework where the existing model “teaches” to a student model its knowledge about the languages it supports, while the student is also trained on a new language. We report an experimental evaluation in several tasks including Sentence Classification, Relational Learning and Sequence Labeling.

Voice assistants, e.g., Alexa or Google Assistant, have dramatically improved in recent years. Supporting voice-based search, exploration, and refinement are fundamental tasks for voice assistants, and remain an open challenge. For example, when using voice to search an online shopping site, a user often needs to refine their search by some aspect or facet. This common user intent is usually available through a “filter-by” interface on online shopping websites, but is challenging to support naturally via voice, as the intent of refinements must be interpreted in the context of the original search, the initial results, and the available product catalogue facets. To our knowledge, no benchmark dataset exists for training or validating such contextual search understanding models. To bridge this gap, we introduce the first large-scale dataset of voice-based search refinements, VoiSeR, consisting of about 10,000 search refinement utterances, collected using a novel crowdsourcing task. These utterances are intended to refine a previous search, with respect to a search facet or attribute (e.g., brand, color, review rating, etc.), and are manually annotated with the specific intent. This paper reports qualitative and empirical insights into the most common and challenging types of refinements that a voice-based conversational search system must support. As we show, VoiSeR can support research in conversational query understanding, contextual user intent prediction, and other conversational search topics to facilitate the development of conversational search systems.

Kernel methods enable the direct usage of structured representations of textual data during language learning and inference tasks. Expressive kernels, such as Tree Kernels, achieve excellent performance in NLP. On the other side, deep neural networks have been demonstrated effective in automatically learning feature representations during training. However, their input is tensor data, i.e., they can not manage rich structured information. In this paper, we show that expressive kernels and deep neural networks can be combined in a common framework in order to (i) explicitly model structured information and (ii) learn non-linear decision functions. We show that the input layer of a deep architecture can be pre-trained through the application of the Nystrom low-rank approximation of kernel spaces. The resulting “kernelized” neural network achieves state-of-the-art accuracy in three different tasks.

This paper describes the KeLP system participating in the SemEval-2017 community Question Answering (cQA) task. The system is a refinement of the kernel-based sentence pair modeling we proposed for the previous year challenge. It is implemented within the Kernel-based Learning Platform called KeLP, from which we inherit the team’s name. Our primary submission ranked first in subtask A, and third in subtasks B and C, being the only systems appearing in the top-3 ranking for all the English subtasks. This shows that the proposed framework, which has minor variations among the three subtasks, is extremely flexible and effective in tackling learning tasks defined on sentence pairs.