My research interests

Text categorization

Logical Models of Information Retrieval

Non-Monotonic Reasoning

Knowledge Representation Formalisms for Taxonomic Reasoning

Natural Language Processing

My students

My research interests

Automated text categorization by machine learning techniques

This is the more recent side of my research. Issues I am working on are:

• "boosting" methods for automated text categorization, on which I am collaborating with Pio Nardiello and Alessandro Sperduti;
• the application of maximum-margin reweighting techniques to automated text categorization, on which I am collaborating with Fabio Aiolli, Guido Ricci, and Alessandro Sperduti;
• indexing techniques specifically designed for text categorization applications, on which I am collaborating with Franca Debole;
• the application of text categorization techniques to automated survey coding, on which I am collaborating with Daniela Giorgetti and Irina Prodanof;
• the application of term categorization techniques to the automatic generation of thematic lexical resources, on which I am collaborating with Henri Avancini, Alberto Lavelli and Bernardo Magnini, and Roberto Zanoli;
• the use of clustering in text categorization under hierarchically organized sets of categories, on which I am collaborating with Andreas Rauber and Henri Avancini.

• the CYCLADES project, funded by the European union and aimed at the development of a layer of personalized and flexible community-based access services to digital archives of scientific literature compliant with the Open Archives Initiative standard;
• the WEBFAQ project, funded by the Province of Trento and dealing with the access to information contained in very large, unstructured, heterogeneous repositories;
• the EUROSEARCH research project, funded by the European Union, and aimed at the development of a European federation of search engines, each covering a national space;
• the IAHA project ("Integrated Access to Heterogeneous Data"), funded by ASI (the Italian Space Agency), aimed at the development of an integrated set of tools for accessing heterogeneous data in the space sciences. I was the IAHA Principal Investigator for the IEI-CNR unit.

Logical Models of Information Retrieval

Speaking of IR, I originally put together for my personal use the complete bibliography of the Information Retrieval journal (Kluwer Academic Publishers); you can either view it or download it as a single file or access it in fully-searchable mode.

Non-Monotonic Reasoning

• model-preference default systems, a family of formalisms belonging to the class of preferential logics. I have investigated isses concerning their denotational semantics, proof theory, and integration with monotonic reasoning (joint work with Carlo Meghini);
• multiple inheritance networks with exceptions. I have worked on the problem of incremental acquisition of knowledge expressed by means of them, and on the problem of their integration with description logics (joint work with Umberto Straccia).

Knowledge Representation Formalisms for Taxonomic Reasoning

The work I have done with Alessandro Saffiotti is the research work I am most proud of in my entire research career, maybe because of the intellectual excitement with which I and Alessandro, at that time young and very unexperienced researchers, carried it out.

Natural Language Processing

Since 1986 to 1988 I worked in NLP within the ESPRIT CFID Project, dealing with techniques for detection and repair of communication failures in dialogue. In particular, I worked on the development of knowledge representation tools for supporting NLP-based detection and repair of miscommunications due to undetected discrepancies in the beliefs of the participants in the communication.

Since 1988 I had to sideline my NLP interests, but I am currently picking them up again, due to the importance of NLP in text retrieval and classification.

My students

• Henri Avancini is working towards his PhD degree in Computer Science at the University of Buenos Aires, Argentina. His interest is in machine learning for hierarchical text categorization.

• Franca Debole obtained her "Laurea" degree in Computer Science at the University of Pisa. Her thesis concerned novel indexing techniques for text categorization.
• Guido Ricci obtained his "Laurea" degree in Computer Science at the University of Pisa. His thesis concerned the application of novel reweighting techniques to learning text classifiers. His work was co-supervised by Fabio Aiolli and Alessandro Sperduti
• Pio Nardiello obtained his "Laurea" degree in Computer Science at the University of Pisa. His thesis concerned attribute discretization for boosting algorithms and their application to text categorization. His work was co-supervised by Alessandro Sperduti
• Maria Fernanda Caropreso obtained her "Licentiatura" degree in Computer Science at the University of Buenos Aires, Argentina. Her thesis concerned  statistical noun phrases for the automatic categorization of text documents. Her work was co-supervised by Stan Matwin.
• Samanta Nelli is working towards her "Laurea" degree in Computer Science at the University of Pisa. Her thesis concerned novel techniques for document similarity computation and document indexing in automatic text classification.
• Attilia Pezzica obtained a "Laurea" degree in Computer Science at the University of Pisa; her thesis concerned the use of self-organizing maps for automatic document classification. Her work was co-supervised by Alessandro Sperduti .
• Mauro Scarpa obtained a "Laurea" degree in Computer Science at the University of Pisa; his work concerned the design and implementation of an information retrieval system for the legal domain.
• Fabio Ferraguto obtained a "Laurea" degree in Computer Science at the University of Pisa; his work concerned the application of theories of citation indexing to automated Web page classification. His work was co-supervised by Giuseppe Attardi.
• Nicola Valdambrini obtained a "Laurea" degree in Computer Science at the University of Pisa;  his work concerned the design and implementation of a new boosting algorithm, and its application to text categorization. His work was co-supervised by Alessandro Sperduti .
• Luigi Galavotti obtained a "Laurea" degree in Computer Science at the University of Pisa; his work concerned the design and implementation of a rich environment for performing text categorization. His work was co-supervised by Maria Simi.
• Lucio Del Testa obtained a "Laurea" degree in Computer Science at the University of Pisa; his work concerned the design and implementation of a system for the semi-automatic classification of documents in the aerospace discipline. His work was co-supervised by Maria Simi.
• Luigi Salafia obtained a "Laurea" degree in Computer Science at the University of Catania;  his work concerned the design and implementation of a distributed digital library for grey literature. His work was co-supervised by Santo Motta.
• Benedetto Condorelli obtained a "Laurea" degree in Computer Science at the University of Pisa;  his work concerned the design and implementation of a mechanism for performing relevance feedback in a logical model of information retrieval.
• Paolo Buongarzoni obtained a "Laurea" degree in Computer Science at the University of Pisa; his work concerned the design and implementation of a theorem prover for the description logic MIRTL. His work was co-supervised by Carlo Meghini.
• Rossella Salis obtained a "Laurea" degree in Computer Science at the University of Pisa; her work concerned the design and implementation of a theorem prover for the description logic MIRTL. His work was co-supervised by Carlo Meghini.
• Giovanna Senette obtained a "Laurea" degree in Computer Science at the University of Pisa; her work concerned the design of a description logic for reasoning about structural descriptions.
• Stefano Abbiani obtained a "Laurea" degree in Computer Science at the University of Pisa; his work concerned circumscription for non-monotonic reasoning.
• Umberto Straccia obtained a "Laurea" degree in Computer Science at the University of Pisa; his work concerned the design and implementation of a system for default reasoning in description logics.
• Piergiorgio Petrilli obtained a "Laurea" degree in Computer Science at the University of Pisa; his work concerned the design and implementation of a theorem prover for the multi-agent epistemic logic KD45. His work was co-supervised by Alessandro Saffiotti.
• Carmelo Ruggiero obtained a "Laurea" degree in Computer Science at the University of Pisa; his work concerned the design and implementation of a theorem prover for the multi-agent epistemic logic KD45. His work was co-supervised by Alessandro Saffiotti.