Table des matières

Johanna Monti, Felice Dell’Orletta et Fabio Tamburini

• Conference and Programme Chairs
• Area Chairs
• Website chair
• Local Organisation
• Reviewers

Keynote Talks and Tutorial

Veronique Hoste

Stefan Kopp

Alessandro Lenci

Contributed Papers

Chiara Alzetta, Felice Dell’Orletta, Simonetta Montemagni et al.

• 1. Introduction and Motivation
• Contribution
• 2. Method
• 3. Data
• 4. Results
• 4.1 Ranking of Dependencies
• 4.2 Distribution of Leaves
• 5. Conclusion

Luca Bacco, Andrea Cimino, Luca Paulon et al.

• 1. Introduction
• 2. Dataset
• 3 Methods
• 3.1 System 1 based on Support Vector Machine (SVM)
• 3.1.1 Feature Extraction
• 3.2 System 2 based on BERT
• 4. Experiments
• System 1
• System 2
• 5. Results and Discussion
• 6. Conclusion

Vevake Balaraman et Bernardo Magnini

• 1. Introduction
• 2. Methodology
• 2.1 Defining Proactivity
• 2.2 Detecting Proactivity
• 3. Experimental Data
• 3.1 Data Acquisition Approaches
• Wizard of Oz (WoZ)
• Bootstrapping
• Role-Taking
• 3.2 Datasets
• WOZ2.0
• MultiWOZ2.1
• Schema-Guided Dataset (SGD)
• Microsoft Dialogue
• Maluuba Frames
• Nespole
• 4. Results and Discussion
• Does our failure-based methodology provide reasonable coverage about proactivity in our datasets?
• Does proactivity correlate with the method of collection of the dataset?
• 5. Conclusion

Pierpaolo Basile, Annalina Caputo, Tommaso Caselli et al.

• 1. Motivation and Background
• 2. Italian diachronic corpora
• 3. Corpus Creation
• Downloading
• Text extraction
• Cleaning
• Processing
• 4. Corpus Statistics
• 5. Conclusions

Valerio Basile

• 1. Introduction
• 2. Related Work
• 3. Weirdness-based Embedding Adaptation
• 3.1 Polarized Weirdness
• 3.2 Word Embedding Adaptation
• 4. Experimental Evaluation
• 4.1 Experiment 1: Multilingual Hate Speech Detection
• 4.1 Experiment 2: Gender Prediction
• 5. Conclusion and Future Work

Elisa Bassignana, Malvina Nissim et Viviana Patti

• 1. Introduction
• 2. Related Work
• 3. Personal-ITY
• Data
• Trait Model
• Ethics Statement
• 3.1 Corpus Creation
• 3.2 Final Corpus Statistics
• 4. Preliminary Experiments
• 5. Conclusions

Nicola Benvenuti, Andrea Bolioli, Alessio Bosca et al.

• Introduction
• 1. The Alzheimer’s language
• 2. Related works
• 3. The Corpus Anchise 320
• 4. Computational linguistic analysis of Corpus Anchise 320
• 4.1 Lexical analysis
• 4.2 Morphosyntactic analysis
• 5. Conclusion and further research

Davide Biasion, Alessandro Fabris, Gianmaria Silvello et al.

• 1. Introduction
• 2. Related work
• 3. Gender in Italian WEs
• 3.1 Identifying gender information
• 3.1.1 Gender score
• 3.1.2 WEAT
• Handling grammatical gender
• 4. Datasets and embeddings
• 5. Experiments
• 5.1 Occupations
• 5.2 Science and Arts
• 5.3 Career and Family
• 5.4 Mitigating the effect of grammatical gender
• 6. Discussion
• 7. Conlusion

Silvia Brambilla, Danilo Croce, Fabio Tamburini et al.

• 1. Introduction
• 2. Related Work
• 3. IFrameNet status
• 4. Automatic Frame Induction
• 5. Experimental Evaluation
• 6. IFrameNet Navigator
• 7. Conclusions and Research Perspectives

Ana-Maria Bucur et Liviu P. Dinu

• 1. Introduction
• 2. Related Work
• 3. Dataset
• 4. Method
• 5. Results
• 6. Conclusion

Gaia Caligiore, Cristina Bosco et Alessandro Mazzei

• 1. Introduction and research goals
• 2. Data Annotation
• 2.1 Analysis on ELAN
• 2.2 Annotation in UD:
• 2.2.1 Morphological Annotation
• 2.2.2 Syntactic Annotation
• 3. Annotation of Language-specific Morpho-syntactic Phenomena
• 3.1 Pointing signs
• 3.2 Repetition of non-pointing signs
• 4. Conclusion

Pierluigi Cassotti, Pierpaolo Basile, Marco De Gemmis et al.

• 1. Motivation and Background
• 2. Diachronic Engine
• 2.1 DE tools
• Word frequency
• Collocations
• Concordances
• Time-series
• 3. Use cases
• 3.1 Event detection through time-series
• 3.2 Annotation of semantic shifts
• 4. Evaluation
• 5. Conclusions

Camilla Casula et Sara Tonelli

• 1. Introduction
• 2. Related Work
• 3. Data
• 4. Experimental Setup
• 5. Results and Discussion
• 5.1 Setting 1: Hateful + Abusive Tweets
• 5.2 Setting 2: Hateful Only Tweets
• 5.3 Setting 3: Hateful + Abusive Tweets (Random Undersampling)
• 5.4 Qualitative Analysis
• 6. Conclusions

Flavio M. Cecchini, Rachele Sprugnoli, Giovanni Moretti et al.

• 1. Introduction
• 2. Treebank Development
• 2.1 From tei-xml to CoNLL-U
• 2.2 Gold Standard Creation
• 2.3 Tripartite Training Process and Control
• 2.4 Evaluation and Analysis
• 3. Conclusion and Future Work

Francesca Chiusaroli, Johanna Monti, Maria Laura Pierucci et al.

• 1. Introduzione
• 2. Il corpus Spotted-Ita
• 3. La lingua della pandemia: metodi per l’analisi del corpus
• 3.1 Spotto la quarantena
• 4. Conclusioni

Yi-Ling Chung, Serra Sinem Tekiroğlu et Marco Guerini

• 1. Introduction
• 2. Related Work
• 3. Italian Counter Narrative Generation
• 4. Dataset
• 5. Models
• 5.1 Training Details
• 5.2 Evaluation
• 6. Results and Discussion
• 7. Conclusion and Future Work

Riccardo Coltrinari, Alessandro Antinori et Fabio Celli

• 1. Motivation and Background
• 1.1 Related Work
• 2. Data
• 3. Experiments and Discussion
• 3.1 Multinomial Classification
• 3.2 Multi-Label Classification
• 4. Conclusion and Future

Ilaria Colucci, Elisabetta Jezek et Vít Baisa

• 1. Introduction
• 1.1 Clustering verbal Objects fillers
• 1.2 T-PAS System of Semantic Types
• 2. Data and method
• 3. Clustering evaluation
• 3.1 SkE clustering evaluation
• 3.2 Linguistic analysis of the clusters
• 4. Conclusion

Lorenzo De Mattei, Michele Cafagna, Felice Dell’Orletta et al.

• 1. Introduction
• 2. GePpeTto
• 2.1 Data
• 2.2 Model
• 2.3 Examples
• 3. Automatic Evaluation
• 3.1 Perplexity
• 3.2 Linguistic Profiling
• 4. Human evaluation
• 4.1 Tasks
• 4.2 Results
• 5. Conclusion

Andrea Gregor de Varda et Carlo Strapparava

• 1. Introduction
• 2. Methods
• 2.1 Semantic vectors
• 2.2 Phonetic vectors
• 2.3 Neural architecture
• 2.4 Experimental conditions
• 3. Results
• 4. Qualitative analysis
• 5. Limitations and further directions

Mirko Di Lascio, Manuela Sanguinetti, Luca Anselma et al.

• 1. Introduction
• 2. A Dialogue Corpus for Customer-care Domain
• 3. Balancing information sources in NLG for DS
• 4. Conclusion and Future Work

Luca Di Liello, Daniele Bonadiman, Cristina Giannone et al.

• 1. Introduction
• 2. Related Works
• 3. Datasets
• 3.1 Quora Question Pairs
• 3.2 FAQ: RDC and LCN
• 3.3 FAQ: ItaFAQ
• 4. ATTandA Approach
• 4.1 Machine Translation of Quora
• 4.2 Transformer architectures
• 4.3 Cross-Domain training
• 5. Results
• 6. Conclusion

Elisa Di Nuovo, Elisa Corino et Cristina Bosco

• 1. Introduction
• 2. Related work
• 3. Test Description
• 3.1 Test Results and Discussion
• 3.2 Error Analysis
• 4. Conclusion

Martina Ducret, Lauren Kruse, Carlos Martinez et al.

• 1. Introduction
• 2. Previous Work
• 3. Our Approach
• 4. Data Set
• 5. Feature Extraction
• 5.1 LIWC
• 5.2 Valence, Arousal, Dominance (VAD)
• 5.3 VADER
• 5.4 Emotional Embeddings
• 5.5 Tweet-Context Similarity Scores
• 5.6 Feature Analysis
• 6. Experimental Evaluation
• 6.1 Data Preprocessing
• 7. Results
• 7. Conclusion

Manuel Favaro, Marco Biffi et Simonetta Montemagni

• 1. Introduzione
• 2. Corpus
• 3. Metodo
• 4. Primi risultati
• 4.1 Analisi quantitativa degli errori
• 4.2 Analisi qualitativa degli errori
• 5. Baseline di confronto
• 6. Conclusioni e sviluppi in corso

Francesco Fernicola, Shibingfeng Zhang, Federico Garcea et al.

• 1. Introduction
• 2. The AriEmozione 1.0 Corpus
• 3. Models and Representations
• 4. Experiments
• 5. Related Work
• 6. Conclusions and Future Work

Marcello Ferro, Sara Giulivi et Claudia Cappa

• 1. Introduction
• 2. ReadLet
• 3. The AEREST protocol
• 3.1 Reading comprehension
• 3.2 Listening comprehension
• 3.3 Reading aloud
• 4. Data structure
• 5. Collected Data
• 6. Results and discussion
• 7. Conclusions and future work

Greta Franzini, Marco Passarotti, Francesco Mambrini et al.

• 1. Introduction
• 2. Data and Methodology
• 2.1 Data Preparation
• 2.2 Data Model
• 3. Conclusion

Gloria Gagliardi, Lorenzo Gregori et Alice Suozzi

• 1. Introduzione
• 2. Materiali
• 2.1 Il corpus ItalC-Covid19
• 2.2 Il corpus Twitter Covid-19
• 2.3 Il dataset
• 3. Analisi della sentiment polarity
• 4. Risultati e discussione
• 5. Conclusioni

Marco Gaido, Mattia Antonino Di Gangi, Matteo Negri et al.

• 1. Introduction
• 2. Sequence-level Knowledge Distillation
• 3. Experimental Settings
• 3.1 Data
• 3.2 Models
• 3.3 Training
• 4. Results
• 5. Analysis
• 6. Conclusions
• Acknowledgments

Greta Gandolfi et Carlo Strapparava

• 1. Introduction
• 2. Background
• 3. Methods and Tools
• 3.1 Dataset
• 3.2 Moderator selection
• 3.1.2 Sampling
• 3.2 Models
• 4. Feature selection
• 4.1 Linguistic Features
• 4.2 Extra-linguistic features
• 5. Experiment
• 6. Results
• Linguistic and Extra-linguistic Features
• 6.2 Linguistic + Extralinguistic Features
• 7. Conclusion

Eleonora Gualdoni, Raffaella Bernardi, Raquel Fernández et al.

• 1. Introduction
• 2. Dataset
• 3. Linguistics Features
• Length
• Word frequency
• Concreteness
• Parts of Speech distributions
• 4. Statistical Analysis
• Length
• Frequency
• Concreteness
• Parts of speech
• 5. Automatic Classification
• 6. Related Work
• 7. Conclusion

Benedetta Iavarone et Felice Dell’Orletta

• 1. Introduction
• 2. Data
• 2.1 Data creation
• 2.2 Data statistics and data selection
• 3 Emotions prediction
• 3.1 Prediction with linguistic and lexical features
• 3.2 Prediction with language model
• 4 Results and discussion
• 5. Conclusion

Alina Karakanta, Matteo Negri et Marco Turchi

• 1. Introduction
• 2. Related work
• 3. Methodology
• 3.1 Data
• 3.2 Model
• 3.3 Evaluation
• 4. Results
• 5. Analysis and Discussion
• 6. Conclusion

Alfred Lim, Beth A. O’Brien et Luca Onnis

• 1. Introduction
• 1.1 Is there an optimal reading experience?
• 2. Method
• 2.1 Model Architecture
• 2.2 Training Procedure
• 2.2.1 Adaptive training
• 2.3 Testing Procedure
• 2.4 Corpus
• 3. Results
• 4. Discussion

Samuel Louvan et Bernardo Magnini

• 1. Introduction
• 3.1 Slot Filling and Intent Classification
• Data Augmentation (DA) Methods
• 3.1 Slot Substitution (Slot-Sub)
• 3.2 Crop and Rotate
• 4. Experiments
• Baseline and Data Augmentation (DA) Methods
• Settings
• Datasets
• 5. Results
• Performance on varying data size
• Performance on different numbers of augmentation per utterance (N).
• 6. Related Work
• 7. Conclusion

Bernardo Magnini, Begoña Altuna, Alberto Lavelli et al.

• 1. Introduction
• 2. Clinical Cases
• 3. Data Collection
• 3.1 Data Protection in the E3C Corpus
• 4. Annotation Scheme
• 4.1 THYME-driven Annotation
• 4.2 Taxonomy-driven Annotation
• 4.3 Language-dependent Decisions
• 4.4 Discussion
• 5. Conclusions and Future Work

Raffaele Manna, Antonio Pascucci, Wanda Punzi Zarino et al.

• 1. Introduction
• 2. Related Work
• 3. The UNIOR Eye Corpus
• 4. Case Study
• 4.1 Dataset
• 4.2 Inter-annotator Agreement
• 4.3 Preprocessing
• 4.4 Machine Learning Approaches
• 5. Results
• 5.1 Confusion Matrices
• 6. Conclusions and Future Work

Claudia Marzi, Anna Rodella, Andrea Nadalini et al.

• 1. Introduction
• 2. ReadLet
• 3. The Data
• 4. Typical and atypical reading development
• 5. General discussion
• 6. Concluding remarks

Francesca Masini, M. Silvia Micheli, Andrea Zaninello et al.

• 1. Introduction
• 2. Methodology
• 3. The Resource
• 3.1 Caveat
• 3.2 POS-patterns
• 3.3 Lemmas used to form MWEs
• 3.4 Distribution in corpora
• 4. Discussion

Andrea Mattei, Dominique Brunato et Felice Dell’Orletta

• 1. Introduction
• 2. Dataset collection
• 3. The success of a fanfiction story: an exploratory study
• 3.1 Linguistic Features
• 4. Data Analysis
• 5. Conclusion

Stefano Menini, Alessio Palmero Aprosio et Sara Tonelli

• 1. Introduction
• 2. Related Work
• 3. Annotation Tool and Process
• 4. Annotated Corpus
• 5. Corpus Analysis
• 6. Dataset comparison
• 7. Conclusions

Enrico Mensa, Gian Manuel Marino, Davide Colla et al.

• 1. Introduction
• 2. Related Work
• 3. Data Analysis
• Dataset
• Analysis result
• 4. Dictionaries Creation and Evaluation
• 4.1 Source Dictionaries
• 5. Evaluation
• Building the dataset
• Evaluating the dictionary
• 6. Conclusions and Future Work

Alessio Miaschi, Chiara Alzetta, Dominique Brunato et al.

• 1. Introduction and Motivation
• Contributions
• 2. Experimental Design
• 2.1 Models
• 2.2 Corpora
• 3. Sentence Perplexity and Readability
• 4. In-Depth Linguistic Investigation
• 5. Conclusion

Alessio Miaschi, Gabriele Sarti, Dominique Brunato et al.

• 1. Introduction and Background
• Contributions
• 2. Approach
• 2.1 Models and Data
• 2.2 Probing features
• 3. Results
• 4. Conclusion

Matteo Muffo et Enrico Bertino

• 1. Introduction
• 2. Related work
• 2.1 Knowledge distillation
• 3. BERTino
• 3.1 Corpus
• 3.2 Pre-training
• 4. Results
• 5. Conclusions

Gennaro Nolano, Carola Carlino, Maria Pia di Buono et al.

• 1. Introduction
• 2. Related Work
• 3. Methodology
• 3.1 Dataset
• 3.2 Annotation Process
• 3.3 Intent classification
• 4. Evaluation and Result Analysis
• 5. Conclusion and Future Work
• 6. Acknowledgments

Isabeau Oliveri, Luca Ardito, Giuseppe Rizzo et al.

• 1. Introduction
• 2. Approach
• 2.1 Self-assessment creativity metrics
• 2.2 Creativity Embedding
• 3. Model Architecture
• 3.1 BERT
• 3.2 Creativity Neural Network and Creativity CLS Embedding
• 4. Experiment
• 5. Result and Conclusion

Alessio Palmero Aprosio, Stefano Menini et Sara Tonelli

• 1. Introduction
• 2. Related Work
• 3. Annotation Tool
• 4. Use Case: Creation of Offensive Posts
• 5. Release
• 6. Conclusions

Matteo Pellegrini et Alessandra Teresa Cignarella

• 1. Introduction
• 2. The Resource
• 3. The Method
• 4. Results
• 5. Conclusions
• 6. Availability of Data and Tools

Marco Polignano, Pierpaolo Basile, Marco De Gemmis et al.

• 1. Introduction
• 2. Related Work
• 3. Model and dataset
• 4. Experimental evaluation
• 5. Conclusion

Andrea Amelio Ravelli, Antonio Origlia et Felice Dell’Orletta

• 1. Introduction
• 2. The CHROME Project
• 2.1 The CHROME multimodal corpus
• 3. Engagement annotation
• 4. Evaluation of the corpus
• 4.1 Linguistic features correlation
• 5. Conclusions and Future Works

Argentina Anna Rescigno, Eva Vanmassenhove, Johanna Monti et al.

• 1. Introduction
• 2. Related Work
• 3. Experimental setup
• 3.1 Compilation of Datasets
• 3.2 Description of the MT Systems
• GT
• BMT
• DL
• 4. Results
• 4.1 Analysis
• GT
• BMT
• DL
• 5. Conclusions and Future Work

Gabriel Roccabruna, Alessandra Cervone et Giuseppe Riccardi

• 1. Introduction
• 2. Related work
• Dialogue act corpora
• Dialogue Act tagging
• 3. iLISTEN2ISO: Mapping iLISTEN to ISO standard
• 4. Analysis of iLISTEN2ISO
• 5. Models
• 6. Experiments
• Experimental setup
• Results
• Error analysis
• 7. Conclusions

Emma Romani et Elisabetta Jezek

• 1. Introduction
• 2. Related works
• 3. Resource and methodology
• 3.1 The resource: T-PAS
• 3.2 Methodology
• 4. Results
• 5. Conclusions and future works

Gaetana Ruggiero, Albert Gatt et Malvina Nissim

• 1. Introduction and Background
• Contributions
• 2. Data
• 2. Preprocessing
• 2.2 Reformatting
• 3. Method
• Topic
• Gender
• Evidence
• Genre
• Splits and Evaluation
• Model
• 4. Results and Discussion
• Discussion
• 5. Conclusion

Giulia Speranza, Raffaele Manna, Maria Pia di Buono et al.

• 1. Introduction
• 2. Related Work
• 3. Archaeo-Term project
• 3.1 Data and Methodology
• 3.2 Results and Evaluation
• 4. Conclusions and future works

Giuseppe Spillo, Cataldo Musto, Marco de Gemmis et al.

• 1. Introduction
• 2. Related Work
• 3. Methodology
• 4. Experimental Evaluation
• 5. Conclusions and Future Work

Rachele Sprugnoli

• 1. Introduction
• 2. Related Works
• 3. Dataset Development
• 3.1 Data Collection
• 3.2 Data Annotation
• Annotation Guidelines
• Inter-Annotator Agreement
• Creation of the Ground Truth
• 4. Dataset Analysis
• 5. Baseline System
• 6. Conclusion and Future Work

Irene Sucameli, Alessandro Lenci, Bernardo Magnini et al.

• 1. Introduction
• 2. Background
• The WoZ approach
• The template-based approach
• The Map Task approach
• 3. Experimental setup
• 4. Results and Discussion
• 5. Conclusion

Fabio Tamburini

• 1. Introduction
• 2. Italian “BERTology”
• 3. Evaluation Tasks
• 3.1 Part-of-Speech Tagging
• 3.1.1 PoS-tagging on Speech Data
• 3.2 Named Entity Recognition
• 3.3 Parsing Universal Dependencies
• 3.4 Sentiment Analysis
• 3.5 Hate Speech Detection
• 4. Discussion and Conclusions

Sirio Papa

• 1. Introduzione
• 2. Il sistema valutato
• 3. Composizione del corpus
• 4. Formazione del campione
• 5. Criteri di valutazione
• 6. Esito generale della valutazione
• 7. Valutazioni particolari
• 8. Confronto con il testo giornalistico
• 9. Conclusioni e sviluppi futuri

Alberto Testoni et Raffaella Bernardi

• 1. Introduction
• 2. Related Work
• 3. Task and Models
• 3.1 Task
• 3.2 Models
• 4. Metrics
• 5. Datasets
• 5.1 Human data
• 5.2 Mixed Batches
• 6. Experiment and Results
• 6.1 Experiment
• 6.2 Results
• Conclusion

Rocco Tripodi

• 1. Introduction
• 2. Related Work
• 3. Topic Modelling Games
• Data Preparation
• Payoff Function and System Dynamics
• 4. Experimental Results
• 4.1 Data and Setting
• 4.2 Evaluation
• 4.3 Topic Coherence and Interpretability
• Internal PMI
• Sparsity
• 4.4 Qualitative Evaluation
• 5. Conclusion and Future Work

Antonio Uva, Pierluigi Roberti et Alessandro Moschitti

• 1. Introduction
• 2. Related Work
• 3. System Description
• 4. Answer Sentence Selection (AS2)
• 4.1 Model
• 5. Intent Classification
• 5.1 Models
• 6. Experiments
• 6.1 Data Description
• 6.2 Model results
• 6.2.1 Answer Sentence Reranking
• 6.2.2 Intent Classification
• 6.3 End-to-End system evaluation
• 7. Conclusions

Pierluigi Vitale, Serena Pelosi et Mariacristina Falco

• Introduction
• 1. State of the Art
• 2. Methodology
• 3. Visualization and Results
• Conclusion

Marco Vassallo, Giuliano Gabrieli, Valerio Basile et al.

• 1. Introduction and Motivation
• 2. Affective Lexicons
• 2.1 MAL
• 3. Polarity Imbalance in Lexicon-based Sentiment Analysis
• 4. The Effect of Lexicon on SA
• 5. The Effect of Topic on Sentiment Analysis
• 6. Conclusion and Future Work

Linda Wiechetek, Chiara Argese, Tommi A. Pirinen et al.

• 1. Introduction
• 2. Background
• 3. Compounds in three NLP applications
• 3.1 An e-dictionary (NDS)
• 3.2 A corpus tool
• 3.3 A grammar checker (GramDivvun)
• 4. Evaluation
• 4.1 An e-dictionary (NDS)
• 4.2 A grammar checker (GramDivvun)
• 5. Conclusion

Mehmet Can Yavuz

• 1. Introduction
• 1.1 Related Works
• 2. Methodology
• 2.1 EmoLex Unigrams
• 2.2 SVD (Golub, 1970)
• 3. Discussion
• 3.1 Analyses of Character Emotions
• 3.2Temporal-Emotional Evaluation of the Tragedies
• 4. Conclusion