The following is a list of articles, conference papers and reports authored or co-authored by researchers of the Cognitive and Communication Technologies division at ITC-irst (Istituto per la Ricerca Scientifica e Tecnologica). Entries refer to papers published internally or which have appeared in refereed media - in particular, journals, conference proceedings, and books - in 1999.
The publications in the list are organized alphabetically by the first author.

Articles, Conference Papers, Books

• Bagnasco C., Cappelli A., Magnini B., Zamatteo D., 'Natural Language Access to Public Adiministration Data: the TAMIC-P System', in Lamma E., Mello P. (eds.) AI*IA 99: Advances in Artificial Intelligence , Lecture Notes on Artificial Intelligence 1792, Springer-Verlag, 2000, pp. 249-260.

  The main goal of the TAMIC-P project is to demonstrate the opportunities offered by the use of Natural Language Processing technologies in the human-machine interaction, in particular related to data access in complex environments. The Natural Language interface has been proposed as a modality of access complementary to other techniques, such as graphical interfaces. It shows its power when used in scenarios characterised by a relevant number of distributed information sources, in which current interfaces do not offer appropriate solutions to the complexity of data handling, and often generate difficulties in navigation. These problems turns out to become even more critical in the presence of not skilled users, having low technical knowledge. Using Natural Language, as normally used between persons for communicating, reduces the skill requirements and enhances the system usability. The evaluation tests performed on the TAMIC-P system confirm this point, showing that upon just a short training, a non-skilled operator is able to operate with effectiveness.

  Ref. No. 9912-86
  

• Bagnasco C., Cappelli A., Magnini B., Zamatteo D., 'Accesso in linguaggio naturale ai dati della Pubblica Amministrazione: il sistema TAMIC-P', in Proceedings of the Sesto Congresso della Associazione Italiana per l'Intelligenza Artificiale, [AIIA-99], Pitagora, Bologna, 1999.

  L'obiettivo principale del progetto TAMIC-P è dimostrare le opportunità offerte dall'utilizzo di tecniche di Elaborazione del Linguaggio Naturale nell'interazione uomo-macchina, in particolare nell'accedere a dati in ambienti complessi. L'interfaccia in Linguaggio Naturale è proposta come modalità di accesso complementare ad altre tecniche, quali le interfacce basate su grafica, e dimostra tutta la sua potenzialità in scenari caratterizzati dalla presenza di un numero rilevante di informazioni distribuite. In questi scenari le attuali interfacce non forniscono soluzioni soddisfacenti ai problemi di complessità e di disorientamento nella navigazione. Problemi che risultano ancor più critici in presenza di utenti poco esperti, in possesso di minori conoscenze tecniche. L'impiego di una modalità come il linguaggio naturale, usata normalmente nella comunicazione tra persone, e pertanto più naturale, riduce i requisiti di competenza e rende il sistema facilmente utilizzabile. I test di valutazione realizzati sul sistema TAMIC-P confortano questa ipotesi, confermando che già dopo un breve training anche un operatore poco esperto è in grado di operare in modo efficace e completo.

  Ref. No. 9912-90

• Bentivogli L., 'Relazioni lessicali e semantiche nella costruzione di un lessico computazionale multilingue: Problematiche Tecniche e Filosofiche', Tesi di Laurea in Filosofia del Linguaggio, Università degli Studi di Bologna, Facoltà di Lettere e Filosofia, Anno Accademico 1997-1998. Bologna, Italy, March, 1999.

  Ref. No.: 9912-112
  

• Bentivogli L., Pianta E., 'Looking for Lexical Gaps', December 1999, in Proceedings of the Ninth EURALEX International Congress, Stuttgart, Germany, August 8 - 12, 2000.

In this paper we present the results of a quantitative evaluation of the discrepancies between the Italian and English lexica in terms of lexical gaps. This evaluation has been carreid out in the context of MultiWordNet, an ongoing project that aims at building a multilingual lexical database. Unlike EuroWordNet, MultiWordNet preserves most of the semantic relations available in the Princeton WordNet. The quantitative evaluation of the English-to-Italian lexical gaps shows that the English and Italian lexica are highly comparable and give empirical support to the MultiWordNet model.

Ref. No. 9912-110

• Bentivogli L., Pianta E., Pianesi F., 'Coping with Lexical Gaps when Building Aligned Multilingual Wordnets', December 1999, 5 pp., in Proceedings of the Second International Conference on Language Resources and Evaluation [LREC 2000], Athens, Greece, May 31 - June 2, 2000.

  In this paper we present a methodology that automatically classifies the translation equivalents of a machine readable bilingual dictionary in three sets: lexical units, lexical gaps and translation equivalents that need to be manually classified as lexical units or lexical gap. This preventive classification minimizes the manual work necessary to cope with lexical gaps in the construction of aligned multilingual wordnets.

  Ref. No. 9912-91
  

• Bianchi A., Zancanaro M., 'Tracking users'movements in an artistic physical space', in Proceedings of the i³ Annual Conference, European Network for Intelligent Information Interfaces, Siena, Italy, October 20-22, 1999, pp. 103-106.

  HIPS studies new technologies and interaction modalities that allow people to navigate both a physical space and a related information space at the same time. In this paper we briefly describe the steps from detecting a set of infrared sensors to the building of an internal representation of the visitor behaviour and discuss how this knowledge is employed to build a personalised presentation. In HIPS the Physical Organisation Knowledge Base generalises the notion of space relying on a notion of naive geometry rather the usual Cartesian geometry. A Visiting Style Module estimates the degree of compatibility between the user's movement patterns and four categories. Hence the Presentation Composer builds a new presentation selecting from an existing repository of audio files using knowledge about the visitor's area/orientation in the Physical Organisation Knowledge Base, while information about the Visiting Style is exploited to collect specific information to be delivered to the user.

  Ref. No. 9912-92

• Cavaglià G., 'Estrazione di Informazione da testi e WordNet: aspetti pratici e applicativi in un dominio specifico', Thesis, December 1999, Tesi di Laurea in Filosofia, Università degli Studi di Torino, Facoltà di Lettere e Filosofia, Anno Accademico 1997-1998

  Ref. No. 9912-113

• Cavaglià G., 'The Development of Lexical Resources for Information Extraction from Text Combining WordNet and Dewey Decimal Classification', in Proceedings of the Ninth Conference of the European Chapter of the Association for Computational Linguistics [EACL'99], Bergen, Norway, June 8-12, 1999.

  Lexicon definition is one of the main bottlenecks in the development of new applications in the field of Information Extraction from text. Generic resources (e.g. lexical database) are promising for reducing the cost of specific lexica definition, but they introduce lexical ambiguity. This paper proposes a methodology for building application-specific lexica by using WordNet. Lexical ambiguity is kept under control by marking synsets in WordNet with field labels taken from the Dewey Decimal Classification.

  Ref. No. 9905-04
  

• Ciravegna F., Lavelli A., 'Full Text Parsing using Cascades of Rules: an Information Extraction Perspective', in Proceedings of 9th Conference of the European Chapter of the Association for Computational Linguistics [EACL'99], Bergen, Norway, June 8-12, 1999.

  This paper proposes an approach to full parsing suitable for Information Extraction from texts. Sequences of cascades of rules deterministically analyze the text, building unambiguous structures. Initially basic chunks are analyzed; then argumental relations are recognized; finally modifier attachment is performed and the global parse tree is built. The approach was proven to work for three languages and different domains. It was implemented in the IE module of FACILE, a EU project for multilingual text classification and IE.

  Ref. No.9905-06
  

• Ciravegna F., Lavelli A., 'Grammar Organization for Cascade-Based Parsing in Information Extraction', December 1999, in Proceedings of the Sixth International Workshop on Parsing Technologies [IWPT 2000], Trento, Italy, February 23-25, 2000.

  Ref. No.9912-93
  

• Ciravegna F., Lavelli A., Mana N., Matiasek J., Gilardoni L., Mazza S., Ferraro M., Balck W.J., Rinaldi F., Mowatt D., 'FACILE: Classifying Text Integrating Pattern Matching and Information Extraction', May 1999, 6 pp., in Proceedings of 16th International Joint Conference on Artificial Intelligence [IJCAI'99], Stockholm, Sweden, July 31-August 6, 1999.

  Successfully managing information means being able to find relevant new information and to correctly integrate it with pre-existing knowledge. Much information in nowadays stored as multilingual textual data; therefore advanced classification systems are currently considered as strategic components for effective knowledge management. We describe an experience integrating different innovative AI technologies such as hierarchical pattern matching and information extraction to provide flexible multilingual classification adaptable to user needs. Pattern matching produces fairly accurate and fast categorisation over a large number of classes, while information extraction provides fine-grained classification for a reduced number of classes. The resulting system was adopted by the main Italian financial news agency providing a pay-to-view service.

  Ref. No.9905-05
  

• Ciravegna F., Lavelli A., Satta G., 'Bringing Information Extraction out of the Labs: the Pinocchio Environment', November 1999, 5pp., in Horn W. (ed.) Proceedings of the 14th European Conference on Artificial Intelligence [ECAI-2000], Berlin, Germany, August 20-25, 2000. IOS Press, Amsterdam, 2000.

  Pinocchio is an environment for developing information extraction applications. New applications and languages can be covered by just writing declarative resources. Information is represented uniformly through the architecture: all the modules use the same input structure and the same type of declarative resources. Modules are implemented via the same basic processors and share a common environment for resource development and debugging. The result is an environment easy to use with limited training and skills.

  Ref. No.9911-09
  

• Corazza A., Cettolo M., Lazzari G., Pianta E., Pianesi F., L.M. Tovena, 'The ITC-irst Speech Translation System', in Proceedings of the Annual Workshop of the Associazione Italiana per l'Intelligenza Artificiale (Gruppo di lavoro sul linguaggio naturale), on Elaborazione del linguaggio e riconoscimento del parlato, Trento, Italy, December 16-17, 1999, pp. 30-32.

  The C-STAR consortium recently demonstrated the results achieved in speech-to-speech machine translation by means of a public demo. ITC-irst participated hosting its own demo, playing the role of a tourist who aims at organizing a travel to New York and Hiedelberg. For this purpose, he contacts, via a video-conference, two tourism agencies, the first in the United States (played by the CMU lab) and the other in Germany (Karlsruhe University). In other occasions, ITC-irst also played the role of the agent. Moreover, another demo was played with ETRI, in Korea. The system performs speech-to-speech translation via a kind of (simple) inter-lingua, the Interchange Format (IF): this allowed us to use the same systems while considering different partner laguages (such as English, German, Korean). In fact, each partner in the Consortium provides the translation chains from its own language to the IF and from the IF to its language. In the following, the two chains for Italian are described.

  Ref. No.9912-04

• De Angeli A., Gerbino W., Nodari E., Petrelli D., 'From Tools to Friends: Where Is the Borderline?', September 1999, to be published in Proceedings of the Workshop on 'Attitude, Personality and Emotions in User-Adapted Interaction', held in conjunction with the Seventh International Conference on User Modeling [UM'99], Banff, Canada, June 20-24, 1999.

  Ref. No. 9909-09

• Giorgi A., Pianesi F., 'The Double Access Reading and Complementiser Deletion in Italian', in Proceedings of the XVII West Coast Conference on Formal Linguistics [WCCFL XVII], Vancouver, Canada, February 20-22, 1998. CSLI Publications, 1999.

  Ref. No. 9912-94
  

• Giorgi A., Pianesi F. 'Ways of Terminating', September 1999, 51 pp., in Cecchetto C., Guasti M.T., Chierchia G. (eds.), At the Interface with Semantics, CSLI Publications, Stanford University, 2000.

  Ref. No. 9909-10
  

• Giorgi A., Pianesi F., 'Imperfect Dreams: the Temporal Dependencies of Fictional Predicates', to be published Probus, International Journal of Latin and Romance Linguistics, Muton De Gruyter, Amsterdam, The Netherlands.

  Ref. No. 9912-116
  

• Not E., Tovena L., Zancanaro M., 'Bridging Anaphora in Deverbal NPs', in Proceedings of the Sixth Congress of the Italian Association for Artificial Intelligence [AI*IA99], Bologna, Italy, September 15-17, 1999.

  Ref. No. 9912-100
  

• Not E., Tovena L., Zancanaro M., 'Positing and Resolving Bridging Anaphora in Deverbal NPs', in Proceedings of the 37th Annual Meeting of the Association for Computational Linguistics, [ACL'99],Workshop on The Relationship Between Discourse/Dialogue Structure and Reference , University of Maryland, USA, June 20-26, 1999.

  Ref. No. 9912-99
  

• Not E., Zancanaro M., 'Reusing Information Repositories for Flexibly Generating Adaptive Presentations', in Proceedings of IEEE International Conference of Information, Intelligence and Systems, [ICIIS'99], Washington, USA, November 1-3, 1999.

  Ref. No. 9912-101

• Petrelli D., 'Museums as Theatre: Suggestions for an Emotional Visit', in Proceedings of the International Workshop on Affect in Interactions held in conjunction with the Annual Conference of the EC I3, [AC'99], Siena, Italy, October 21-22, 1999.

  HIPS1 is a hand-held electronic guide that adapts its behaviour to museum visitors. It uses physical movements as a main interaction modality and responds to this implicit interaction with suitable commentaries created on the fly. To achieve more involving presentations two further steps are needed: inference of visitor's emotional status and composition of emotional presentation. Investigations on how visitor behaviour is influenced by external (environment) or internal (attitude) factors and on the effect of emotional cues are currently being carried on at the University of Siena. Nevertheless, when the picture will be clear, a computational model to support emotional presentations will be needed. This paper discusses a first proposal on this more technological aspect.

  Ref. No. 9912-102
  

• Petrelli D., Not E., Zancanaro M., 'Getting Engaged and Getting Tired: What Is in a Museum Experience', in Proceedings of the Workshop on 'Attitude, Personality and Emotions in User-Adapted Interaction', held in conjunction with the Seventh International Conference on User Modeling [UM'99], Banff, Canada, June 20-24, 1999.

  The physical experience of visiting a museum includes emotion and intellect. The way a person feels during an experience becomes an integral part of that memory. Evoking the feeling or mood at a later time may trigger details of a memory associated with it. Thus in a museum environment it is essential that personal discovering is appropriately supported providing visitors with the suitable amount of information they need, at the right time and place, and in the form that makes it the most acceptable and enjoyable. The quality of the received presentation is liable to subjective judgement. Personal taste typically overcomes objective effectiveness and can prejudice the pleasure of the visit. When designing new personal electronic guide to support museum visit, designers have to bear in mind all the known effects and be aware of others unexpected side effects. Moreover the system should monitor visitor's reactions in order to infer the effectiveness of the choices done and, if necessary, to rearrange the presentation style. This paper presents the work that is going to be done in the HIPS project to cope with the feedback-rearrangement process when emotion and involvement are directed to museum visitors.

  Ref. No. 9909-08

• Roventini A., Alonge A., Bertagna F., Magnini B., Calzolari N., 'ItalWordNet: a large semantic database for Italian', December 1999, in Gravilidou M., Crayannis G., markantonatu S., Piperidis S., Stainhaouer G. (eds.), Proceedings of the Second International Conference on Language Resources and Evaluation [LREC-2000], Athens, Greece, May 31 - June 2, 2000, pp. 783-790.

  The focus of this paper is on the work we are carrying out to develop a large semantic database within an Italian national project, SI-TAL, aiming at realizing a set of integrated (compatible) resources and tools for the automatic processing of the Italian language. Within SI-TAL, ItalWordNet is the reference lexical resource which will contain information related to about 130,000 word senses grouped into synsets. This lexical database is not being created ex novo, but extending and revising the Italian lexical wordnet built in the framework of the EuroWordNet project. In this paper we firstly describe how the lexical coverage of our wordnet is being extended by adding adjectives, adverbs and proper nouns, plus a terminological subset belonging to the economic and financial domain. The relevant changes involved by these extensions both in the linguistic model and in the data structure are then illustrated. In particular we discuss: i) the new semantic relations identified to encode information on adjectives and adverbs; ii) the new architecture including the terminological subset .

  Ref. No. 9912-117
  

• Huber M., Magnini B., 'Processing of Object Oriented Dialogues', in Proceedings of EUROLAN'99, Workshop on Procedures in Discourse Interpretation, Iasi, Romania, July 23-27, 1999.

In this paper we introduce object oriented dialogues (OODs), a subclass of dialogues where the communicative goal is just the retrieval of information about objects in a particular domain. OODs allow a number of computational simplifications which make them attractive for many information access applications. Starting point for this approach was the TAMIC-P prototype system, which will be briefly presented. Finally, OODs are well suited to be represented as graphic object (e.g. folders and documents), easily guiding the user through the discourse structure and at the same time through parts of the underlying knowledge base: in this way typical problems of discourse ambiguity and domain opacity can be overcome.

Ref. No. 9912-95

• Huber M., Magnini B., 'Toward a Robust Processing of Information-Seeking Dialogues', in Proceedings of AI*IA'99, Workshop on 'Elaborazione del linguaggio e riconoscimento del parlato', Trento, Italy, December 13-14, 1999.

  The main assumption of this paper is that the interpretation of information-seeking dialogues is mainly a domain driven process, where much of the knowledge necessary for their processing can be derived from the Domain Model represantation. Following this line, some discourse structure and the recovering from error situations, are all addressed building a semantic dialogue tree which mirrors the Domain Model structure. The algorithm we propose relies on three steps: (i) first a logical form is built for a single turn, whose clauses are grouped according to their functional role in the user request; (ii) then a dialogue chain for a single turn exploiting the semantic relations among the nodes is derived from the logical form; (iii) finally, the overall dialogue tree is searched for the positions where the chain can be best appended or merged.

  Ref. No. 9912-96

• Magnini B., Artale A., Huber M., Strapparava C., Zancanaro M., 'Efficient Natural Language Access to Databases: The TAMIC-P System', Presented at the 37th Annual Meeting of the Association for Computational Linguistics, [ACL'99], University of Maryland, USA, June 20-26, 1999.

  TAMIC-P proposes an interaction style in which direct manipulation and natural language are strictly integrated, each of them providing a communication style complementary to the other. The system has been designed to be used by a Public Administration operator during an information session with a citizen. On field evaluations demonstrated the high usability of the system even with a short training phase.

  Ref. No. 9912-97
  

• Magnini B., Cavaglià G., 'Integrating Subject Field Codes into WordNet', in Proceedings of the Second International Conference on Language Resources and Evaluation, [LREC-2000], Athens, Greece, May 31 - June 2, 2000, pp. 1413-1418.

  In this paper we present a lexical resource which provides an annotation of WordNet synsets with Subject Field Codes. We discuss both the methodological issues we dealt with and the annotation techniques we implemented. A quantitative analysis of the resource coverage is reported, as well as a qualitative evaluation of the proposed annotations.

  Ref. No. 9912-98
  

• Pianesi F., Pianta E., Tovena L.M., 'Comparing Methodologies for Evaluating the Generator in a Speech-to-Speech Translation System' , in Proceedings of the Seventh European Workshop on Natural Language Generation [EWNLG'99], Tolouse, France, May 13-15, 1999, pp. 145-154.

  This paper compares two formal methods for evaluating the output of generation systems. In one case humans are used for hand coding and assessment, while in the other case they also perform the task of producing the reference text. It is argued that the methodology that tries to limit the human intervention gives less accurate results.

  Ref. No. 9912-103
  

• Pianesi F., Varzi A., 'Events and Event Talks: an Introduction', in Higginbotham J, Pianesi F., Varzi A. (eds.) Speaking of Events, Oxford University Press, New York, 2000.

  Ref. No. 9912-87

• Pianta E., Tovena L.M., 'XIG: Generating from Interchange Format using Mixed Representations', in Proceedings Sixth Congress of the Italian Association for Artificial Intelligence, (AI*IA'99) , Bologna, Italy, September 14-17, 1999.

  We present the C-STAR Italian Generator (XIG), a system for generating Italian text from the interlingua content representation (Interchange Format) adopted within the C-STAR II speech to speech translation project. The constraints of the application scenario led us to develop a generator based on Mixed Representations, providing a good trade-off between flexibility and efficiency.

  Ref. No. 9912-104

• Pianta E., Tovena L.M., 'Mixing Representation levels: the Hybrid Approach to Automatic Text Generation', in Proceedings of the AISB'99 Workshop on Reference Architectures and Data Standards for NLP, Edinburgh, Scotland, UK, April 6-9, 1999, pp. 8-13.

  Natural language generation systems (NLG) map non-linguistic representations into strings of words through a number of steps using intermediate representations of various levels of abstraction. Template based systems, by contrast, tend to use only one representation level, i.e. fixed strings, which are combined, possibly in a sophisticated way, to generate the final text. In some circumstances, it may be profitable to combine NLG and template based techniques. The issue of combining generation techniques can be seen in more abstract terms as the issue of mixing levels of representation of different degrees of linguistic abstraction. This paper aims at defining a reference architecture for systems using mixed representations. We argue that mixed representations can be used without abandoning a linguistically grounded approach to language generation.

  Ref. No. 9912-105

• Sarini A., Strapparava C., 'User Modelling in a Museum Exploration Adaptive System', in Poster Proceedings of the Seventh International Conference on User Modeling '99 [UM'99], Banff, Canada, June 20-24, 1999.

  Ref. No. 9912-106
  

• Stefani A., Strapparava C., 'One to One Marketing on the Web: How to Exploit User Modelling in the SiteIF Project', in Posters Proceedings of the 8th International World Wide Web Conference [WWW8], Toronto, Canada, May 11-14, 1999.

  Ref. No. 9912-108

• Stefani A., Strapparava C., 'Exploiting NLP Techniques to Build User Model for WEB Sites: the Use of WordNet in SiteIF Project', in Proceedings of the 8th International World Wide Web Conference [WWW8], the Second Workshop on Adaptive Systems and User Modeling on the World Wide Web, Toronto, Canada, May 11-14, 1999.

  Ref. No. 9912-107

• Tovena L.M., Pianta E., 'Generating felicitous sentences from underspecified semantic representations', in Proceedings 3rd International Workshop on Computational Semantics, Tilburg, The Netherlands, January 1999, pp. 410-412.

  Interchange Format is a semantic formalism used as interlingua in a multilingual speech-to-speech translation project. Underspecification, in the sense of nonspecification of aspects of the sentence semantics, plays a crucial role in easying the analysis of spontaneous speech. However, it causes difficulties when generating. We put forth solutions which rely mainly on domain knowledge.

  Ref. No. 9912-109

Technical Reports

• Ciravegna F., Lavelli A., Satta G., 'Full Parsing Approximation, Finite-State Cascades and Grammar Organization for Information Extraction', Technical Report, November 1999, 7 pp.

  This paper proposes an approach to full parsing approximation suitable for Information Extraction from texts. Sequences of cascades of finite-state rules deterministically analyze the text, building unambiguous structures. Initially basic chunks are analyzed; then clauses are recognized and nested; finally modifier attachment is performed and the global parse tree is built. The approach has been extensively proven to work mainly for Italian, but it was also tested for English and Russian. A parser based on such approach has been implemented as part of Pinocchio, an environment for developing and running IE applications.

  Ref. No. 9911-02