Papers by Alessandro Moschitti

Adversarial Domain Adaptation for Duplicate Question Detection (D18-1)

Challenge:	Recent studies show that Question Answering (QA) based on Answer Sentence Selection (AS2) can be improved by generating an improved answer from the top-k ranked answer sentences.
Approach:	They propose to train a GenQA model by transferring knowledge from a trained AS2 model . they use top ranked candidate as the generation target and next k top rated candidates as context .
Outcome:	The proposed model outperforms existing models on public and industrial datasets.

Copied to clipboard

Darsh Shah, Tao Lei, Alessandro Moschitti, Salvatore Romeo, Preslav Nakov

Responsible & Trustworthy NLP

Challenge:	Recent years have seen the rise of community question answering forums . duplicate questions easily become ubiquitous as users often ask the same question, possibly in a slightly different formulation, making it difficult to find the best (or one correct) answer.
Approach:	They propose to use domain adaptation to detect duplicate questions in forums . they find that domain adaptation improves performance over multiple pairs of domains .
Outcome:	The proposed approach improves 5.6% over the best baseline across multiple pairs of domains.

Pre-training Transformer Models with Sentence-Level Objectives for Answer Sentence Selection (2022.emnlp-main)

Copied to clipboard

Luca Di Liello, Siddhant Garg, Luca Soldaini, Alessandro Moschitti

Efficient and Accurate Contextual Re-Ranking for Knowledge Graph Question Answering (2024.lrec-main)

Challenge:	Existing models for answer sentence selection (AS2) are not yet available for AS2 .
Approach:	They propose to incorporate paragraph-level semantics within and across documents to improve transformers for AS2 . they propose to use a dataset to predict whether two sentences are extracted from the same paragraph .
Outcome:	The proposed model outperforms baseline models on public and industrial datasets on three public and one industrial dataset.

Copied to clipboard

Kexuan Sun, Nicolaas Paul Jedema, Karishma Sharma, Ruben Janssen, Jay Pujara, Pedro Szekely, Alessandro Moschitti

Question Answering

Challenge:	Existing approaches to QA over textual data are based on a "retrieve-then-generate" pipeline.
Approach:	They propose a "triple-level" labeling strategy that infers fine-grained labels and trains a re-ranker to improve relevance of retrieved triples.
Outcome:	The proposed pipeline improves on prior KGQA systems by 5.56% Exact Match.

Cross-Lingual Open-Domain Question Answering with Answer Sentence Generation (2022.aacl-main)

Copied to clipboard

Benjamin Muller, Luca Soldaini, Rik Koncel-Kedziorski, Eric Lind, Alessandro Moschitti

Measuring Retrieval Complexity in Question Answering Systems (2024.findings-acl)

Challenge:	Open-Domain Generative Question Answering has achieved impressive performance in English . combining document-level retrieval with answer generation can generate complete sentences .
Approach:	They propose an open-domain approach that combines document retrieval with answer generation to generate complete sentences in English . they propose a cross-lingual generative model that exploits passages written in multiple languages .
Outcome:	The proposed model outperforms answer sentence selection baselines for all 5 languages and monolingual pipelines for three out of five languages.

Copied to clipboard

Matteo Gabburo, Nicolaas Jedema, Siddhant Garg, Leonardo Ribeiro, Alessandro Moschitti

Analyzing and Improving Coherence of Large Language Models in Question Answering (2025.naacl-long)

Challenge:	a new metric, retrieval complexity (RC), measures the difficulty of answering questions.
Approach:	They propose a retrieval complexity metric conditioned on the completeness of retrieved documents . they propose an unsupervised pipeline to measure RC given an arbitrary retrieval system .
Outcome:	The proposed pipeline measures RC more accurately than alternative estimators on six challenging QA benchmarks.

Copied to clipboard

Ivano Lauriola, Stefano Campese, Alessandro Moschitti

WikiSum: Coherent Summarization Dataset for Efficient Human-Evaluation (2021.acl-short)

Challenge:	Large language models suffer from instability or lack of coherence when receiving diverse input variations.
Approach:	They analyze the behavior of large language models when dealing with multiple lexical variations of the same info-seeking questions.
Outcome:	The proposed model generates equivalent outputs when receiving diverse input variations.

Copied to clipboard

Nachshon Cohen, Oren Kalinsky, Yftah Ziser, Alessandro Moschitti

Text Generation

Challenge:	Existing summarization datasets are limited in their ability to evaluate output . a human evaluation is necessary to understand and improve summarizing systems .
Approach:	They propose a dataset based on how-to articles and coherent paragraph summaries written in plain language.
Outcome:	The proposed dataset makes human evaluation easier and more effective . the authors compare the proposed dataset to existing ones on PubMed and the literature.

Effective Pretraining Objectives for Transformer-based Autoencoders (2022.findings-emnlp)

Copied to clipboard

Luca Di Liello, Matteo Gabburo, Alessandro Moschitti

Integrating Stance Detection and Fact Checking in a Unified Corpus (N18-2)

Challenge:	ELECTRA is more accurate than BERT, but it is not clear if this is due to its innovative architecture or to the long and extensive training, which highly increases the computation cost for obtaining the final language model.
Approach:	They propose to replace BERT’s Masked Language Modeling objective (MLM) with Token Detection (TD) by using a statistical approach to generate light tokens.
Outcome:	The proposed method can replace ELECTRA's computationally heavy generators without a significant drop in performance.

Copied to clipboard

Ramy Baly, Mitra Mohtarami, James Glass, Lluís Màrquez, Alessandro Moschitti, Preslav Nakov

Responsible & Trustworthy NLP

Challenge:	Existing methods for fact checking are not supported by existing datasets, which treat fact checking, document retrieval, source credibility, stance detection and rationale extraction as independent tasks.
Approach:	They propose to implement automatic fact checking on an Arabic fact checking corpus, which is the first of its kind.
Outcome:	The proposed approach is based on an Arabic fact checking corpus, the first of its kind.

Context-Aware Transformer Pre-Training for Answer Sentence Selection (2023.acl-short)

Copied to clipboard

Luca Di Liello, Siddhant Garg, Alessandro Moschitti

Joint Models for Answer Verification in Question Answering Systems (2021.acl-long)

Challenge:	Existing approaches to perform Answer Sentence Selection (AS2) using only the candidate sentence are sub-optimal.
Approach:	They propose to use pre-trained transformers to perform contextual AS2 fine-tuning . they propose to apply pre-training objectives to local contextual AS2.
Outcome:	The proposed methods improve baseline AS2 accuracy by up to 8% on some datasets.

Copied to clipboard

Zeyu Zhang, Thuy Vu, Alessandro Moschitti

Cross-Pair Text Representations for Answer Sentence Selection (D18-1)

Challenge:	Using a joint approach, we found that the model is more efficient than those developed in machine reading (MR) work.
Approach:	They propose a joint model for selecting correct answer sentences among the top k provided by answer sentence selection modules.
Outcome:	The proposed model improves on WikiQA, TREC-QA, and a real-world dataset.

Copied to clipboard

Kateryna Tymoshenko, Alessandro Moschitti

Representation Learning

Challenge:	Existing approaches to textual entailment and question answering focus on intra-pair similarity . a simple lexical matching (marked with italics) is not enough to learn a model based on intrapair Qto-A similarities.
Approach:	They propose to compute scalar products representing similarity between members of different pairs instead of using a single vector for each pair.
Outcome:	The proposed approach outperforms more complex models based on neural networks.

Ensemble Transformer for Efficient and Accurate Ranking Tasks: an Application to Question Answering Systems (2022.findings-emnlp)

Copied to clipboard

Yoshitomo Matsubara, Luca Soldaini, Eric Lind, Alessandro Moschitti

Learning Answer Generation using Supervision from Automatic Question Answering Evaluators (2023.acl-long)

Challenge:	Large transformer models are expensive and slow to use in many applications.
Approach:	They propose an efficient neural network to distill large transformers into a single smaller model.
Outcome:	The proposed model outperforms existing models on English datasets . it outperformed existing models with 2.7 more parameters and 2.5 slower .

Copied to clipboard

Matteo Gabburo, Siddhant Garg, Rik Koncel-Kedziorski, Alessandro Moschitti

Supervised Neural Clustering via Latent Structured Output Learning: Application to Question Intents (2021.naacl-main)

Challenge:	Recent studies show sentence-level extractive QA is outperformed by Generation-based QA (GenQA) models.
Approach:	They propose a training paradigm for GenQA using automatic QA evaluation models . they augment training data with answers generated by the GenQA model and labelled by GAVA .
Outcome:	The proposed training paradigm improves answering accuracy over existing models.

Copied to clipboard

Iryna Haponchyk, Alessandro Moschitti

Text Clustering

Challenge:	Recent work on structured prediction has produced very effective supervised clustering algorithms using linear classifiers.
Approach:	They propose to use latent structured prediction loss and Transformer models to approach supervised clustering.
Outcome:	The proposed approach outperforms the state-of-the-art in recreating intents from public question corpora.

Strong and Light Baseline Models for Fact-Checking Joint Inference (2021.findings-acl)

Copied to clipboard

Kateryna Tymoshenko, Alessandro Moschitti

Double Retrieval and Ranking for Accurate Question Answering (2023.findings-eacl)

Challenge:	Automated fact checking is rapidly gaining attention of the NLP and AI communities.
Approach:	They propose lightweight strong baselines for automated fact-checking systems . they propose to combine multiple pieces of evidence to verify a claim .
Outcome:	The proposed methods outperform heavier models on the leaderboard with blind TEST set.

Copied to clipboard

Zeyu Zhang, Thuy Vu, Alessandro Moschitti

AVA: an Automatic eValuation Approach for Question Answering Systems (2021.naacl-main)

Challenge:	Recent work shows that answer verification models can improve the state of the art in Question Answering . despite the fact that the supporting candidates are ranked only according to the relevancy with the question, the model still lacks the support needed for other answer candidates.
Approach:	They propose a double reranking model that selects the best support for each target answer . they propose 'second neural retrieval stage' to encode question and answer pair as query .
Outcome:	The proposed approach improves the state of the art in Question Answering . the proposed model ranked candidates according to relevancy and not the answer . but the proposed approach fails to provide the best support .

Copied to clipboard

Thuy Vu, Alessandro Moschitti

A Study on Efficiency, Accuracy and Document Structure for Answer Sentence Selection (2020.coling-main)

Challenge:	AVA is an automatic evaluation approach for question answering . it uses transformer-based language models to encode question, answer, and reference texts .
Approach:	They propose an automatic evaluation approach for Question Answering that uses Transformer-based language models to encode question, answer, and reference texts.
Outcome:	AVA can estimate system Accuracy with an error lower than 7% at 95% confidence level . the proposed approach achieves 74.7% F1 score in predicting human judgment for single answers .

Copied to clipboard

Daniele Bonadiman, Alessandro Moschitti

FocusQA: Open-Domain Question Answering with a Context in Focus (2022.findings-emnlp)

Challenge:	Existing approaches to QA re-rank sentences use huge neural models or complex attentive architectures.
Approach:	They propose to exploit the intrinsic structure of the original rank with an effective word-relatedness encoder to achieve the highest accuracy among the cost-efficient models.
Outcome:	The proposed model takes 9.5 seconds to train on the WikiQA dataset, compared with 18 minutes required by a standard BERT-base fine-tuning.

Copied to clipboard

Gianni Barlacchi, Ivano Lauriola, Alessandro Moschitti, Marco Del Tredici, Xiaoyu Shen, Thuy Vu, Bill Byrne, Adrià de Gispert

Will this Question be Answered? Question Filtering via Answer Model Distillation for Efficient Question Answering (2021.emnlp-main)

Challenge:	a new method for question answering with a context in focus simulates a free interaction with QA systems.
Approach:	They introduce question answering with a cotext in focus task that simulates a free interaction with QA systems.
Outcome:	The proposed model outperforms state-of-the-art models for question answering with a context in focus up to 21.3% absolute points.

Copied to clipboard

Siddhant Garg, Alessandro Moschitti

A Study of Latent Structured Prediction Approaches to Passage Reranking (N19-1)

Challenge:	Existing methods to improve QA efficiency do not take specific answers into account.
Approach:	They propose a transformer-based approach to improve QA efficiency by filtering out questions that will not be answered by the system.
Outcome:	The proposed model can approximate the Precision/Recall curves of the target QA system.

Copied to clipboard

Iryna Haponchyk, Alessandro Moschitti

Retrieving Support to Rank Answers in Open-Domain Question Answering (2025.emnlp-main)

Challenge:	a structured output framework is useful for learning to rank problems . current approaches for answer sentence reranking are mostly based on pairwise ranking signals or simple binary classification.
Approach:	They propose a structured output approach which regards rankings as latent variables . they propose an inference procedure to find the max-violating ranking based on decomposition of the corresponding loss.
Outcome:	The proposed approach solves the optimization problem on WikiQA and TREC13 datasets.

Copied to clipboard

Zeyu Zhang, Alessandro Moschitti, Thuy Vu

Answer Generation for Retrieval-based Question Answering Systems (2021.findings-acl)

Challenge:	a novel question answering architecture retrieves content relevant to the combined pair . previous work on automatic claim verification has shown hallucinations .
Approach:	They propose a question-answer architecture that prioritizes supporting evidence . it retrieves paragraphs that directly substantiate the correctness of a with respect to q .
Outcome:	The proposed approach can be used by large language models to retrieve explanatory paragraphs that ground their reasoning.

Copied to clipboard

Chao-Chun Hsu, Eric Lind, Luca Soldaini, Alessandro Moschitti

A Flexible, Efficient and Accurate Framework for Community Question Answering Pipelines (P18-4)

Challenge:	Question Answering systems are a core component of many commercial applications . answer sentence selection (AS2) models are trained to select the best answer sentence .
Approach:	They propose to train a sequence to sequence transformer model to generate an answer from a set of candidates.
Outcome:	The proposed model improves accuracy by 32 points over the state-of-the-art model on English AS2 datasets.

Copied to clipboard

Salvatore Romeo, Giovanni Da San Martino, Alberto Barrón-Cedeño, Alessandro Moschitti

Question Answering

Challenge:	Community Question Answering is a research area that benefits from deep linguistic analysis . previous cQA challenges have shown that neural approaches are not enough to deliver state-of-the-art results .
Approach:	They propose a framework to distribute computation of cQA tasks over computer clusters . community question answering is a research area that benefits from deep linguistic analysis .
Outcome:	The proposed framework scales to large datasets and delivers fast processing.

Speechworthy Instruction-tuned Language Models (2024.emnlp-main)

Copied to clipboard

Hyundong Cho, Nicolaas Jedema, Leonardo Ribeiro, Karishma Sharma, Pedro Szekely, Alessandro Moschitti, Ruben Janssen, Jonathan May

Accurate Training of Web-based Question Answering Systems with Feedback from Ranked Users (2023.acl-industry)

Challenge:	Current instruction tuned language models are trained on textual preference data and therefore not aligned to speech domain.
Approach:	They propose to use radio-industry best practices to prompt and learn speech-based preference data to improve speech-suitability of popular instruction tuned language models.
Outcome:	The proposed methods achieve the best win rates in head-to-head comparisons, resulting in preferred or tied to the base model in 76.2% of comparisons on average.

Copied to clipboard

Liang Wang, Ivano Lauriola, Alessandro Moschitti

The Cascade Transformer: an Application for Efficient Answer Sentence Selection (2020.acl-main)

Challenge:	Recent work shows that large-scale annotated datasets are essential for training state-of-the-art Question Answering (QA) models.
Approach:	They use large-scale annotated datasets to train question answering models . they use feedback data collected from deployed QA systems to provide cheaper supervision .
Outcome:	The proposed model improves on the large scale annotated datasets from QA systems . the proposed model can be easily supervised on large-scale unlabeled web data .

Copied to clipboard

Luca Soldaini, Alessandro Moschitti

Learning to Progressively Recognize New Named Entities with Sequence to Sequence Models (C18-1)

Challenge:	Recent research shows that transformer-based neural networks can greatly advance the state of the art over many natural language processing tasks.
Approach:	They propose a technique to adapt transformer-based models into a cascade of rankers.
Outcome:	The proposed technique reduces computation by 37% with almost no impact on accuracy on two English question answering datasets.

Copied to clipboard

Lingzhen Chen, Alessandro Moschitti

Supervised Clustering of Questions into Intents for Dialog System Applications (D18-1)

Challenge:	Existing models for Named Entity Recognition (NER) are trained on data with the same NE label set, but they are not able to recognize previously unseen NE categories.
Approach:	They propose to use a sequence to sequence model for Named Entity Recognition (NER) and propose to reshape and re-parametrize the output layer of the first learned model to enable the recognition of new NEs.
Outcome:	The proposed model can recognize previously unseen NE categories while keeping the knowledge of previously seen categories.

Copied to clipboard

Iryna Haponchyk, Antonio Uva, Seunghak Yu, Olga Uryupina, Alessandro Moschitti

Pre-Training Methods for Question Reranking (2024.eacl-short)

Challenge:	Existing methods for detecting intents in text are task-specific and costly . current methods focus on manually analyzing user questions and creating a taxonomy of intents to be attached to the appropriate actions.
Approach:	They propose a model for automatically clustering questions into user intents to help design tasks . they use powerful semantic classifiers and supervised clustering methods based on structured output .
Outcome:	The proposed model improves on two intent clustering corpora on two languages/domains.

Copied to clipboard

Stefano Campese, Ivano Lauriola, Alessandro Moschitti

Semantic Linking in Convolutional Neural Networks for Answer Sentence Selection (D18-1)

Challenge:	Existing methods for Question Answering to search for semantically similar questions are not suitable for new questions.
Approach:	They propose an unsupervised method for retrieving and ranking questions . they use a question retrieval model and a selection model to rerank questions based on their relevance .
Outcome:	The proposed method achieves state-of-the-art performance on QRC and Quora-match datasets . it provides better and cheaper access to answers than the system generated them .

Copied to clipboard

Massimo Nicosia, Alessandro Moschitti

Challenge:	Recent NLP approaches that model relations between text use complex architectures and attention.
Approach:	They propose to use labelled data to model semantic relations between two pieces of text . they use word representations to encode matching features directly in the word representation .
Outcome:	The proposed approach beats tree kernel models and neural models with similar input encodings while keeping the model simple and fast to train.

Reference-based Weak Supervision for Answer Sentence Selection using Web Data (2021.findings-emnlp)

Copied to clipboard

Vivek Krishnamurthy, Thuy Vu, Alessandro Moschitti

Low-Resource NLP

Challenge:	Existing solutions for QA use weakly-supervised data from Web.
Approach:	They propose a data pipeline that harvests weakly-supervised answer sentences from Web data . they train TANDA models, which are the state of the art for AS2 .
Outcome:	The proposed pipeline improves on three different datasets and sets the state-of-the-art models to P@1=90.1% and MAP=92.9%.

Cross-Lingual Knowledge Distillation for Answer Sentence Selection in Low-Resource Languages (2023.findings-acl)

Copied to clipboard

Shivanshu Gupta, Yoshitomo Matsubara, Ankit Chadha, Alessandro Moschitti

Multilinguality

Challenge:	Cross-Lingual Knowledge Distillation (CLKD) is a method to train AS2 models for low-resource languages without labeled data.
Approach:	They propose a method to train AS2 models for low-resource languages without labeled data . they use a translation-based WikiQA dataset and a multilingual AS2 dataset .
Outcome:	The proposed method outperforms or rivals fine-tuning with labeled data and machine translation and the teacher model.

QUADRo: Dataset and Models for QUestion-Answer Database Retrieval (2023.findings-emnlp)

Copied to clipboard

Stefano Campese, Ivano Lauriola, Alessandro Moschitti

Injecting Relational Structural Representation in Neural Networks for Question Similarity (P18-2)

Challenge:	Existing tools do not consider answers (question-question similarity only) or their quality in the annotation process.
Approach:	They propose an open-domain annotated resource that retrieves questions from pre-computed databases . they use a binary label associated with each question/answer pair to determine relevance .
Outcome:	The proposed model trains and evaluates models on 443,000 examples with respect to key aspects of QA systems.

Copied to clipboard

Antonio Uva, Daniele Bonadiman, Alessandro Moschitti

Representation Learning

Challenge:	Recent years have seen exponential growth and use of web forums, where users can exchange and find information just asking questions in natural language.
Approach:	They propose to use Tree Kernels to learn a model on relatively few pairs of questions as gold standard (GS) predicting labels on a very large corpus of question pairs is also a useful approach, they propose .
Outcome:	The proposed model can learn more accurate models after fine tuning on GS.

Modeling Context in Answer Sentence Selection Systems on a Latency Budget (2021.eacl-main)

Copied to clipboard

Rujun Han, Luca Soldaini, Alessandro Moschitti

Green & Sustainable NLP

Challenge:	Current AS2 models score question-answer pairs individually, ignoring any information from the document each potential answer was extracted from.
Approach:	They propose an approach to efficiently incorporate contextual information into AS2 models . they use unsupervised similarity techniques to extract relevant sentences from source document .
Outcome:	The proposed approach improves 6% to 11% over state-of-the-art in AS2 with minimal latency.

Datasets for Multilingual Answer Sentence Selection (2024.findings-emnlp)

Copied to clipboard

Matteo Gabburo, Stefano Campese, Federico Agostini, Alessandro Moschitti

Multilinguality

Challenge:	Existing datasets for Answer Sentence Selection (AS2) focus on English due to the scarcity of annotated datasets.
Approach:	They propose to use a large language model to train answer sentences in English . they use annotated datasets from English and other languages to train AS2 models .
Outcome:	The proposed datasets are highly performant and close the performance gap between English and other languages.

Building a Dataset for Automatically Learning to Detect Questions Requiring Clarification (2022.lrec-1)

Copied to clipboard

Ivano Lauriola, Kevin Small, Alessandro Moschitti

Paragraph-based Transformer Pre-training for Multi-Sentence Inference (2022.naacl-main)

Challenge:	Existing work on question answering systems assumes all questions are intelligible and unambiguous . however, available datasets do not meet requirements for building commercial virtual assistants .
Approach:	They propose to make question answering systems more robust by classifying if question is intelligible and returning a clarification question for contextual ambiguity.
Outcome:	The proposed system can classify if the input question is intelligible and return a clarification question for ambiguous questions.

Copied to clipboard

Luca Di Liello, Siddhant Garg, Luca Soldaini, Alessandro Moschitti