Transformer-based document cross-encoder rerankers are a central component of modern information retrieval systems. Despite their success, these models suffer from high computational costs due to processing long query-document sequences at inference time. A known approach to improve efficiency is token compression, which consists of aggregating groups of tokens together in the initial embedding layer, reducing the effective number of tokens and making the computation faster. While token compression has proven to be successful for bi-encoder retrievers, we empirically observed that this approach may be ineffective for cross-encoder rerankers. In this paper, we propose Layer-wise Token Compression (LTC), which applies adaptive token pooling at intermediate transformer layers. Through extensive ablation studies on MS MARCO passage and document ranking tasks, we demonstrate that compression at middle layers preserves ranking quality while increasing inference QPS by up to 25% for passage ranking and up to 116% for document ranking. We also extend LTC to listwise LLM rerankers and show that the same approach can be easily applied to long-context listwise reranking, where the QPS improvements are even greater. More surprisingly, when applying rerankers trained on short passages to long-document ranking tasks, models trained with compression outperform their uncompressed counterparts, suggesting that compression may act as a beneficial regularizer that encourages length-invariant representations.

Ivano Lauriola, Stefano Campese, Alessandro Moschitti. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.

Question Answering (QA) systems aim to return correct and concise answers in response to user questions. QA research generally assumes all questions are intelligible and unambiguous, which is unrealistic in practice as questions frequently encountered by virtual assistants are ambiguous or noisy. In this work, we propose to make QA systems more robust via the following two-step process: (1) classify if the input question is intelligible and (2) for such questions with contextual ambiguity, return a clarification question. We describe a new open-domain clarification corpus containing user questions sampled from Quora, which is useful for building machine learning approaches to solving these tasks.

Combining internal representations of a pre-trained Transformer model, such as the popular BERT, is an interesting and challenging task nowadays. Usually, internal representations are combined by simple heuristics, e.g. concatenation or average of a subset of layers, with a consequent need for calibrating multiple hyper-parameters during the fine-tuning phase. Inspired by the recent literature, we propose a principled approach to optimally combine internal representations of a Transformer model via Multiple Kernel Learning strategies. Broadly speaking, the proposed system consists of two elements. The former is a canonical Transformer model fine-tuned on the target task. The latter is a Multiple Kernel Learning algorithm that extracts and combines representations developed in the internal layers of the Transformer and performs predictions. Most important, we use the system as a powerful tool to inspect the information encoded into the Transformer network, emphasizing the limits of state-of-the-art models.

In recent years, the increasing interest in the development of automatic approaches for unmasking deception in online sources led to promising results. Nonetheless, among the others, two major issues remain still unsolved: the stability of classifiers performances across different domains and languages. Tackling these issues is challenging since labelled corpora involving multiple domains and compiled in more than one language are few in the scientific literature. For filling this gap, in this paper we introduce DecOp (Deceptive Opinions), a new language resource developed for automatic deception detection in cross-domain and cross-language scenarios. DecOp is composed of 5000 examples of both truthful and deceitful first-person opinions balanced both across five different domains and two languages and, to the best of our knowledge, is the largest corpus allowing cross-domain and cross-language comparisons in deceit detection tasks. In this paper, we describe the collection procedure of the DecOp corpus and his main characteristics. Moreover, the human performance on the DecOp test-set and preliminary experiments by means of machine learning models based on Transformer architecture are shown.