A5066077296- Natural Language Processing Techniques
- Topic Modeling
- Semantic Web and Ontologies
- Speech and dialogue systems
- Biomedical Text Mining and Ontologies
- Advanced Text Analysis Techniques
- Linguistic Studies and Language Acquisition
- Service-Oriented Architecture and Web Services
- Text Readability and Simplification
- Advanced Database Systems and Queries
- Data Quality and Management
- Web Data Mining and Analysis
- Multi-Agent Systems and Negotiation
- Text and Document Classification Technologies
- AI in Service Interactions
- Information Retrieval and Search Behavior
- Authorship Attribution and Profiling
- linguistics and terminology studies
- Language and cultural evolution
- Image Retrieval and Classification Techniques
- Intelligent Tutoring Systems and Adaptive Learning
- Sentiment Analysis and Opinion Mining
- Algorithms and Data Compression
- Open Education and E-Learning
- Machine Learning and Algorithms
Fondazione Bruno Kessler
2015-2024
Free University of Bozen-Bolzano
2021-2023
University of Pisa
2020-2021
Ca' Foscari University of Venice
2019
University of Bergamo
2017-2018
University of Pavia
2017-2018
Delft University of Technology
2018
Bridge University
2017
Ferioli & Gianotti (Italy)
2011-2015
Bar-Ilan University
2007-2014
This paper presents the Third PASCAL Recognising Textual Entailment Challenge (RTE-3), providing an overview of dataset creating methodology and submitted systems.In this year's dataset, a number longer texts were introduced to make challenge more oriented realistic scenarios.Additionally, pool resources was offered so that participants could share common tools.A pilot task also set up, aimed at differentiating unknown entailments from identified contradictions justifications for overall...
Abstract The goal of identifying textual entailment – whether one piece text can be plausibly inferred from another has emerged in recent years as a generic core problem natural language understanding. Work this area been largely driven by the PASCAL Recognizing Textual Entailment (RTE) challenges, which are series annual competitive meetings. current work exhibits strong ties to some earlier lines research, particularly automatic acquisition paraphrases and lexical semantic relationships...
In recent years, fostered by deep learning technologies and the high demand for conversational AI, various approaches have been proposed that address capacity to elicit understand user's needs in task-oriented dialogue systems. We focus on two core tasks, slot filling (SF) intent classification (IC), survey how neural based models rapidly evolved natural language understanding introduce three architectures: independent models, which model SF IC separately, joint exploit mutual benefit of...
Answer Validation is an emerging topic in Question Answering, where open domain systems are often required to rank huge amounts of candidate answers. We present a novel approach answer validation based on the intuition that amount implicit knowledge which connects question can be quantitatively estimated by exploiting redundancy Web information. Experiments carried out TREC-2001 judged-answer collection show achieves high level performance (i.e. 81% success rate). The simplicity and...
The continuous expansion of the multilingual information society has led in recent years to a pressing demand for linguistic resources suitable be used different applications.
This paper explores the role of domain information in word sense disambiguation. The underlying hypothesis is that labels, such as M EDICINE , A RCHITECTURE and S PORT provide a useful way to establish semantic relations among senses, which can be profitably used during disambiguation process. Results obtained at ENSEVAL -2 initiative confirm for significant subset words disambiguate with very high level precision.
Anne-Lyse Minard, Manuela Speranza, Eneko Agirre, Itziar Aldabe, Marieke van Erp, Bernardo Magnini, German Rigau, Rubén Urizar. Proceedings of the 9th International Workshop on Semantic Evaluation (SemEval 2015). 2015.
Bernardo Magnini, Roberto Zanoli, Ido Dagan, Kathrin Eichler, Guenter Neumann, Tae-Gil Noh, Sebastian Pado, Asher Stern, Omer Levy. Proceedings of 52nd Annual Meeting the Association for Computational Linguistics: System Demonstrations. 2014.
This paper aims at providing a comprehensive overview of recent developments in dialogue state tracking (DST) for task-oriented conversational systems. We introduce the task, main datasets that have been exploited as well their evaluation metrics, and we analyze several proposed approaches. distinguish between static ontology DST models, which predict fixed set states, dynamic can states even when changes. also discuss model’s ability to track either single or multiple domains scale new...
We describe Evalita-LLM, a new benchmark designed to evaluate Large Language Models (LLMs) on Italian tasks. The distinguishing and innovative features of Evalita-LLM are the following: (i) all tasks native Italian, avoiding issues translating from potential cultural biases; (ii) in addition well established multiple-choice tasks, includes generative enabling more natural interaction with LLMs; (iii) evaluated against multiple prompts, this way mitigating model sensitivity specific prompts...
This paper investigates proactivity, a characteristic phenomenon of collaborative human-human interaction, where participant in the dialogue offers addressee some useful and not explicitly requested information. More precisely, proactive behaviour is: (i) self-prompted simply reactive, that is, speaker does act merely response to requests other has made; (ii) somehow effective for achievement goal, since long-term, goal-directed predicts future states needs. Proactivity been poorly...
We present E3C-3.0, a multilingual dataset in the medical domain, comprising clinical cases annotated with diseases and test-result relations. The includes both native texts five languages (English, French, Italian, Spanish Basque) translated projected from English source into target (Greek, Polish, Slovak, Slovenian). A semi-automatic approach has been implemented, including automatic annotation projection based on Large Language Models (LLMs) human revision. several experiments showing...