My research area is the cognitive science of language, a field at the intersection of linguistics, psychology, and computer science / artificial intelligence. More specifically, I am interested in modelling the semantic side of language processing, developing theories that are grounded in actual empirical evidence - as provided by linguistic studies, the results of psychological experiments, and computational modelling using corpora. Some of the issues I'm interested in are the role of `partial' semantic interpretations in a theory of language processing, the interpretation and generation of anaphoric expressions, and semantic interpretation in spoken dialogues. I'm also working towards making 'Cognitively Plausible Semantic Processing' a more active research area, both through the organization of workshops and by teaching courses on the subject.
Empirically grounded work on language processing shares a number of concerns (e.g., experimental evaluation) and methodologies (e.g., statistical methods) with Natural Language Engineering, in which I'm also interested both from the point of view of research and from a teaching perspective. A lot of my early research was driven by issues arising out of work on task-oriented spoken dialogue systems, but I recently started applying my work on computational models of anaphora to tasks such as multimedia text generation, information extraction / text mining, and segmentation.
The papers mentioned in this document are available from my home page at http://cswww.essex.ac.uk/poesio/papers.html.
A question I've long been interested in is whether humans fully recover the meaning of an expression: i.e., is everything that we hear or read fully interpreted, or are we recovering only part of its meaning? My work in this area was initially motivated by the problem of scope disambiguation: whether and how humans decide the respective scope of scope-taking operators (negation, modality, quantifiers) in sentences such as In most democratic countries, most politicians can fool most of the people on almost every issue most of the time. What is interesting about this question is that the number of potential alternative interpretations of such sentences is enormous; also, humans do not seem to be very good at this (Poesio, 1991; Poesio, 1993a; Poesio, 1994a ). My initial concern was to explain how humans could assign a semantic interpretation to such sentences without generating all of the interpretations predicted by current semantic theory, and what kind of interpretations they would obtain if they were not disambiguating such sentences completely. The starting point was the idea of `underspecified logical form' introduced in NLP research (e.g., (Schubert and Pelletier, 1982)) as a way of characterizing the space of possible semantic interpretations of a sentence without actually computing them. I hypothesized that, in some cases at least, the final result of semantic interpretation might be an underspecified logical form, and concentrated on trying to identify what semantics these representations might have - i.e., what inferences we would expect humans to be able to draw from them ( Poesio, 1991 ). However, I quickly came to realize that a cognitively plausible theory of disambiguation had to take into account the results of psychological work on ambiguity processing, which emphasized the role of incrementality ( Poesio, 1994a; Poesio, 1996c ). Unfortunately, very little work had been done to study the respective roles of incrementality and underspecification in scope disambiguation. So, for my dissertation, I ended up relying quite a lot on corpus data - my own analysis of the preferred scopal assignments in the TRAINS corpus (see below); and in subsequent work, I concentrated on the role of underspecified interpretations in other aspects of semantic processing for which more psychological evidence was available, such as lexical disambiguation and anaphoric processing ( Poesio, 1996b , Poesio, 2003 ). I also arrived at the conclusion that the `disjunctive' semantics proposed in my earlier work (and equivalent to the semantics later proposed by Reyle) was not a good characterization of such partial interpretations.
In later work ( Poesio, 1996b; Poesio, 1996a; ) and in particular in my forthcoming CSLI monograph (Poesio, 2003 ) I propose a theory of utterance processing and semantic underspecification that, from the semantic point of view, is based on the idea - closely related to ideas developed in work on syntactic underspecification - that the interpretations produced by the language processor are descriptions of the properties of utterances and of how utterances fit together with other utterances in hierarchical structures. The idea that disambiguation is a form of defeasible reasoning also plays a central role in the theory; defeasible logics are used to define the notion of precisification (disambiguated interpretation) that in other theories of underspecification is incorporated in the semantics of the language.
I've also been concerned with finding further evidence concerning the role of underspecification. In work with Uwe Reyle, Frank Keller, Rosemary Stevenson, and Patrick Sturt I used both corpus analysis and psychological experimentation in the attempt to identify cases in which interpretations are truly left underspecified - in the case of anaphoric expression. This work was motivated by the results on investigating semantic judgments using corpora, and in particular the question whether the semantic disagreements on anaphoric expressions (observed, e.g., in the work with Vieira, and in the GNOME project) could be explained as cases in which the text doesn't allow to completely determine the semantic interpretation of an expression ( Poesio and Reyle, 1999 ; Poesio and Reyle, 2001 ). We then used magnitude estimation experiments to test whether in fact one could argue that such anaphoric expressions were left uninterpreted Poesio, Keller, Reyle, Stevenson and Sturt, 2001 ). This work led to the ARRAU project, funded by EPSRC - a three year project investigating the extent to which certain types of anaphoric expressions are actually resolved.
Poesio, M. To appear 2005. Incrementality and Underspecification in Semantic Processing. CSLI Lecture Notes.
Poesio, M. and U. Reyle. 2001. Underspecification in anaphoric reference. In Proc. of the Fourth International Workshop on Computational Semantics, Tilburg, January 2001.
Poesio, M. 1996b. Disambiguation as (defeasible) reasoning about underspecified representations. In P. Dekker and M.Stokhof, editors, Proc. of the Tenth Amsterdam Colloquium, pages 607-625. ILLC, December.
Poesio, M. 1996c. Semantic ambiguity and perceived ambiguity. In K. van Deemter and S. Peters, editors, Semantic Ambiguity and Underspecification . CSLI, Stanford, CA, chapter 8, pages 159-201.
Poesio, M. 1994a. Discourse Interpretation and the Scope of Operators. Ph.D. thesis, University of Rochester, Department of Computer Science, Rochester, NY.
Poesio, M. 1993a. Assigning a semantic scope to operators. In Proc. ACL-93 , Columbus, OH, June.
Poesio, M. 1991. Relational semantics and scope ambiguity. In J. Barwise,
J. M. Gawron, G. Plotkin, and S. Tutiya, editors, Situation Semantics
and its Applications, vol.2. CSLI, Stanford, CA, chapter 20, pages
469-497.
Using corpora to study the interpretation of anaphora
The type of linguistic expression I have been mainly concerned with are anaphoric expressions such as pronouns (like he or she) and definite descriptions (like the cat). I am interested in their semantic properties , in how they are assigned an interpretation in context, and in how they are generated . The most distinctive aspect of my own work in the area has been the emphasis on the use of corpus data both to investigate how humans process such expressions and to evaluate computational models of anaphora interpretation.
My interest in using corpus data as a way of supplementing and verifying linguistic judgments and psycholinguistic evidence dates back to my involvement in the TRAINS project at the University of Rochester (see below) , but this topic became an especially important part of my research work as a result of my work with my former student, Renata Vieira, now at Unisinos (Brazil). My work with Renata focuses on the study of definite descriptions , i.e., expressions such as the car. The reason for this interest is that interpreting definite descriptions involve access to encyclopedic and commonsense knowledge to a much greater extent that it is the case with pronouns; studying this process may therefore lead to insights in the interface between grammar and commonsense knowledge.
One of the first questions Renata and I looked at was whether humans agree on the interpretation of definite descriptions ( Poesio and Vieira, 1998 ), since a positive answer to this question is a prerequisite to using corpora annotated with such information. We found, however, that even on this simple task, there wasn't complete agreement among our subjects. In particular, our subjects had real problems agreeing on the antecedents for bridging references (Clark, 1978). This study led us to an interest in exploring the extent to which humans agree on various other semantic judgments, which we have been exploring as a part of the GNOME project .
Annotated corpora also offer the opportunity to evaluate processing models in a quantitative fashion. While giving a quantitative evaluation of the performance of natural language processing techniques is problematic in many respects, the attempts at doing so are proving the only way of assessing the actual impact of differences between processing methods. Renata Vieira and myself have been developing a series of prototypes capable of performing definite description interpretation on unrestricted text ( Vieira and Poesio, 1996; Poesio, Vieira and Teufel, 1997 ;Vieira and Poesio, 1999; Vieira and Poesio, 2000a ; Vieira and Poesio, 2000b ). This work has allowed us, first of all, to study the role of syntactic knowledge in identifying discourse-new definite descriptions; and then to begin isolating the types of commonsense knowledge that are needed to resolve definite descriptions, and especially the bridging uses.
My collaboration with Renata continued with my PhD student at Essex Mijail Alexandrov-Kabadjov. One of the concerns of my research with Mijail has been to develop a tool for anaphora resolution that would take advantage of recent NL technology and of XML, as well as being very modular. Our system, GUITAR ( Poesio and Alexandrov-Kabadjov, 2004) can be downloaded from the Web. The main focus of our work on GUITAR was to develop better discourse-new detection algorithms ( Poesio et al, 2005; Poesio et al, 2004), and to use the system in actual applications, such as summarization (Steinberger et al, 2007).
Josef Steinberger, Massimo Poesio, Mijail Kabadjov and Karel Jezek 2007. Two uses of anaphora resolution in summarization. Information Processing and Management, v. 43, n. 6, 1663-1680. Special issue on Summarization (Donna Harman, ed.). (pdf of preliminary version)
Poesio, M., M. Alexandrov-Kabadjov, R. Vieira, R. Goulart, and O. Uryupina, 2005. Does discourse-new detection help definite description resolution? Proc. of the Sixth IWCS, Tilburg, January. (pdf)
Poesio, M. and Alexandrov-Kabadjov, M. 2004. " A general-purpose, off-the-shelf system for anaphora resolution ", Proc. of LREC, Lisbon, May.
Poesio, M. and R. Vieira. 1998. A corpus-based investigation of definite description use. Computational Linguistics. v. 24, n. 2, 183-216. Also available as Research Paper CCS-RP-71, Centre for Cognitive Science, University of Edinburgh.
Poesio, M., Vieira, R., Alexandrov-Kabadjov, M., Uryupina, O. and Goulart, R. 2004. "Discourse-new detectors for definite description resolution: A survey and a preliminary proposal", Proc. of the ACL Workshop on Reference Resolution, Barcelona, July.
Vieira, R. and M. Poesio. 1999. Processing definite descriptions in corpora. In S. Botley and T. McEnery, editors, Corpus-based and Computational Approaches to Discourse Anaphora. John Benjamins. Also available as HCRC Research Paper HCRC/RP-86, University of Edinburgh.
Vieira, R. and M. Poesio. 2000a. Corpus-based development and evaluation of a system for processing definite descriptions. In Proc. of the 18th COLING, Saarbruecken, August.
Vieira, R. and M. Poesio. 2000b. An empirically-based system for processing definite descriptions. Computational Linguistics, v. 26, n.4.
Our work on resolving bridging descriptions eventually led us to research on acquiring and using lexical knowledge - first for application in anaphora resolution, but then also for other types of applications. We began by trying to use existing sources of lexical knowledge such as WordNet for this purpose ( Poesio, Vieira and Teufel, 1997 ); but more recently we started investigating how this commonsense knowledge could be acquired in a psychologically plausible way ( Poesio, Schulte im Walde, and Brew, 1998 ; Poesio, Ishikawa, Schulte in Walde, and Vieira, 2002 ).
In recent years, this work has led to research trying to improve current techniques for the acquisition of lexical knowledge. In collaboration with my PhD student Abdulrahman Almuhareb, I have been studying the possibility of using better concept descriptions, generalizing from simple syntactic relations - for example, including attributes in these descriptions. The results have been extremely good (Almuhareb and Poesio, 2004).
In collaboration with my PhD student Olivia Sanchez-Graillet, I also started to look beyond extracting lexical relations from text to the possibility of extracting more general types of relations; in particular, causal relations ( Sanchez-Graillet and Poesio, 2004).
Abdulrahman Almuhareb and Massimo Poesio, 2004. "Attribute-based and value-based clustering: an evaluation", Proc. of EMNLP, Barcelona, July .
Massimo Poesio, Tomonori Ishikawa, Sabine Schulte im Walde and Renata Vieira, 2002. "Acquiring lexical knowledge for anaphora resolution", Proc. of the Third LREC, Las Palmas, May.
Poesio, M., S. Schulte im Walde, and C. Brew. 1998. Lexical clustering and definite description interpretation. In Proc. of the AAAI Spring Symposium on Learning for Discourse, Stanford, CA, March. AAAI.
Poesio, M., R. Vieira, and S. Teufel. 1997. Resolving bridging references in unrestricted text. In R. Mitkov, editor, Proc. of the ACL Workshop on Operational Factors in Robust Anaphora Resolution, pages 1-6, Madrid. Also available as HCRC Research Paper HCRC/RP-87, University of Edinburgh.
Olivia Sanchez-Graillet and Massimo Poesio, 2004. "Extracting Bayesian Networks from Text", Proc. of the Fourth LREC, Lisbon, May.
One of my main projects in the area of anaphoric expressions has been my work with Rosemary Stevenson on the role of salience in pronoun interpretation. Our first joint project involved a careful analysis of the claims of theories of anaphoric processing such as Sidner's theory or Centering theory at the light of the results of psycholinguistic research. This work, to be reported in a forthcoming book for Cambridge University Press (Poesio and Stevenson, in preparation ), convinced us of the need for revisions to the underlying models of the attentional state, and for additional experimental work.
Our empirical studies involve a combination of traditional psychological techniques, of corpus analysis, and of computational modelling. Some of the questions we studied using standard psychological techniques include the effect of rhetorical structure on pronoun interpretation, and the interaction between animacy and thematic roles in determining the salience of a discourse entity ( Pearson, Stevenson and Poesio, 2001 ). We have been using annotated corpora (in particular, the GNOME corpus - see below) to evaluate the claims of Centering theory by keeping into account different ways of setting its parameters ( Poesio et al. 2000 ). Finally, in collaboration with Janet Hitzeman , we have studied the interaction between global focus and local focus (Hitzeman and Poesio, 1998 ).
In the last two years, I have been carrying out further corpus-based investigations using the results of this previous work and the GNOME corpus - including work with Natalia N. Modjeska on demonstratives, and with Malvina Nissim on possessives.
Pearson, J., R. Stevenson, and M. Poesio. 2001. The effects of animacy, thematic role, and surface position on the focusing of entities in discourse. Proc. of the 1st Workshop on Cognitively Plausible Models of Semantic Processing, Edinburgh, July.
Hitzeman, J. and Poesio, M. 1998. Long-distance pronominalisation and global focus. In Proc. ACL-98, Montreal, August.
The work with Rosemary Stevenson on psychologically plausible models of anaphora resolution has obvious applications in the area of natural language generation - the development of systems capable of producing natural language that find easy to process. This observation led to a joint project on language generation in which Rosemary and I collaborated with Donia Scott from the University of Brighton and Barbara di Eugenio from the University of Illinois at Chicago. GNOME , which ran from 1998 to 2000, was an EPSRC-funded project whose goal was to apply results from psychological research and from corpus analysis to develop and evaluate algorithms for generating nominal expressions.
In GNOME we developed both hand-crafted and statistical models of the processes involved in discourse entity realization, including statistical models of the choice of NP type (Poesio and Henschel, 1999 ; Poesio et al, 1999 ; Poesio, 2000 ) and NP modification ( Cheng, Poesio, Henschel, and Mellish, 2001 ). We were particularly interested on pronominalization ( Henschel, Cheng and Poesio, 2000 ).
More recently, I have collaborated with Nikiforos Karamanis, Chris Mellish and Jon Oberlander on developing statistical models of other types of generation, including text structuring (Karamanis et al 2004a, 2004b).
Cheng, H., Massimo Poesio, Renate Henschel, and Chris Mellish, 2001, "Corpus-based NP Modifier Generation", Proc. of the NAACL, Pittsburgh, June.
Henschel, R., H. Cheng, and M. Poesio, 2000. Pronominalization Revisited. In Proceedings of the 18th COLING, Saarbruecken, August.
Karamanis, N., Massimo Poesio, Chris Mellish and Jon Oberlander, 2004a. "Evaluating Centering-based metrics of coherence for text structuring using a reliably annotated corpus", Proc. of ACL, Barcelona, July.(pdf).
Nikiforos Karamanis, Chris Mellish, Jon Oberlander and Massimo Poesio, 2004b. " A Corpus-based Methodology for Evaluating Metrics of Coherence for Text Structuring ", Proc. of INLG, Brighton, July.
Poesio, M., and R. Henschel. 1999. Statistical NP Generation: A First Report. In R. Kibble and K. van Deemter (eds.), Proceedings of the ESSLLI Workshop on NP Generation, Utrecht, August.
Poesio, M., R. Henschel, J. Hitzeman, R. Kibble, S. Montague, and K. van Deemter. 1999. Towards An Annotation Scheme For Noun Phrase Generation. In Proceedings of the EACL Workshop on Linguistically Interpreted Corpora, Bergen, June.
One of the main stumbling blocks for research in semantics has been the lack of semantically annotated resources of a size comparable to that of the Penn Treebank. As a result, I have been involved in a number of projects concerned with developing appropriate annotation schemes for several types of semantic information (particularly anaphoric information), with the development of tools for annotation, and with the creation of semantically annotated resources.
The first of such projects was the EU-funded MATE project (http://mate.nis.sdu.dk/), concerned with the development of customizable tools for dialogue annotation based on XML standoff technology. I was responsible for the design of the coreference annotation guidelines in the project (see also (Poesio et al, 1999)).
An important aspect of GNOME was the creation of a corpus annotated for syntactic, semantic and discourse information, such as anaphoric reference. The annotation scheme used in GNOME was based on the MATE guidelines, augmented with instructions for other types of semantic annotation. Creating the GNOME corpus involved quite a lot of work on agreement on semantic judgments, continuing the research initiated as part of my collaboration with Renata Vieira (see below); in particular, we found good agreement among our subjects on animacy, less good agreement on countability, and little on genericity, the assignment of thematic roles, and topichood (Poesio, 2000 ). The GNOME corpus is in continuous development, and has already been used to develop statistical models of NP realization and to study Centering theory (see above). Papers describing the corpus and its annotation scheme are available from the corpus web page, http://cswww.essex.ac.uk/Research/nle/corpora/GNOME/.
More recently, I have been involved in a collaborative project with the Universita' di Venezia, VENEX, which has resulted in the creation of an anaphorically annotated corpus of Italian including both written text and spoken dialogue.
Poesio, M., Bruneseaux, F. and Romary, L. (1999). "The MATE meta-scheme for coreference in dialogues in multiple language", Proc. of the ACL Workshop on Standards for Discourse Tagging. Maryland, June 1999.
Poesio, M. 2000. Annotating a corpus to develop and evaluate discourse entity realization algorithms: issues and preliminary results. In Proceedings of the Third Conference on Language Resources and Engineering, Athens, May.
Poesio, M. 2004a. "The MATE/GNOME Scheme for Anaphoric Annotation, Revisited", Proc. of SIGDIAL, Boston, April.(pdf)
Poesio, M. 2004b. "Discourse Annotation and Semantic Annotation in the GNOME Corpus", Proc. of the ACL Workshop on Discourse Annotation, Barcelona, July (pdf).
My research on the semantics and pragmatics of dialogues involves formal work on general models of the common ground and how it is established; some of this work has been done in collaboration with Robin Cooper , David Traum , and Reinhard Muskens (Poesio, 1994a ; Poesio and Traum, 1997 ; Poesio and Muskens, 1997 ; Poesio and Traum, 1998 ; Matheson, Poesio and Traum, 2000 ). I am also interested in developing models of semantic interpretation that take into account the fragmentary nature of spoken input and the disfluencies it may contain (Poesio, 1995a; Poesio, 1995b; Poesio,1996b; Poesio, 1996c; Poesio, 1997 ).
One of the goals of my work with David Traum is to come up with a theory of the updates resulting from utterances other than assertions; in particular, we have been studying the process by which the common ground is established (Clark's grounding, see also (Traum, 1994)) and how to interpret utterances that contribute to it, such as acknowledgments and repairs ( Poesio and Traum, 1997 ; Poesio and Traum, 1998 ;Matheson, Poesio and Traum, 2000 ). This work started in Rochester and continued after I joined HCRC and the Centre for Cognitive Science at the University of Edinburgh.
At HCRC I had the opportunity to take advantage of resources such as the MapTask corpus , and initiated a collaboration with Robin Cooper that began in the EU-funded FraCaS project and continued in the European Union-sponsored project TRINDI , which was concerned with dialogue management and, more generally, with the notion of information state in dialogue.
I am actively involved in promoting the semantics of dialogues as an attractive research area, and in organizing events in the area, such as the annual series of workshops on the Semantics of Dialogues .
C. Matheson, M. Poesio, and D. Traum. 2000. Modeling Grounding and Discourse Obligations Using Update Rules. In Proc. NAACL-2000, Seattle, April.
Poesio, M. and D. Traum. 1998. Towards an Axiomatisation of Dialogue Acts. In J. Hulstijn and A. Nijholt (eds.), Proceedings of the Twente Workshop on the Formal Semantics and Pragmatics of Dialogues, Enschede, May, 207--222.
Poesio, M. and R. Muskens. 1997. The dynamics of discourse situations. In P. Dekker and M. Stolhof, editors, Proceedings of the 11th Amsterdam Colloquium. University of Amsterdam, ILLC, December.
Poesio, M. and D. Traum. 1997. Conversational actions and discourse situations. Computational Intelligence, 13(3):309-347.
Poesio, M. 1996c. Formal semantics and spoken dialogues. In J.J. Quantz, editor, Proc. of the ECAI workshop on Corpus-Based Semantic Analysis , Budapest, August. ECAI.
Poesio, M. 1995b. A model of conversation processing based on micro conversational events. In Proceedings of the Annual Meeting of the Cognitive Science Society, Pittsburgh, July.
Semantic and pragmatic interpretation in spoken dialogues were the main focus of the five-year Advanced Fellowship I was awarded from the UK Engineering and Physical Sciences Research Council (EPSRC) . The main research goal of the fellowship was to apply my work on ambiguous expressions and semantic interpretation from spoken input to develop modules for the robust interpretation of spoken dialogue utterances whose performance can be evaluated in a quantitative fashion. Work on the fellowship involved, in addition to the theoretical work just mentioned, extensive investigations of the empirical foundations of one specific subarea of semantic interpretation-the interpretation of definite descriptions-as well as work on robust systems performing this kind of interpretation (see below), and more recent work on dialogue act interpretation in the MapTask.
Together with Andrei Mikheev (Poesio and Mikheev, 1998 ), I looked at whether `hierarchical' models of game structure such as conversation games theory lead to improved predictive power over `flat' models, answering the question in positive. More recently, together with Helen Wright and Stephen Isard we found that these results still hold even when the hierarchical structure has to be reconstructed from the input, using intonation and other such clues (Wright, Poesio and Isard, 1999 ; Wright, Poesio and Isard, 2002 )
Poesio, M. and Mikheev, A. 1998. The Predictive Power of Game Structure in Dialogue Act Recognition: Experimental Results Using Maximum Entropy Estimation. In Proc. of ICSLP-98, Sydney, December.
A lot of my research has been motivated by problems that I ran across while I was involved with the TRAINS project at the Department of Computer Science, University of Rochester . TRAINS is a long-term project devoted to the study of task-oriented spoken dialogues; the activities related to the project include collecting and analyzing corpora of spoken dialogues, developing theories of specific phenomena, and developing prototype spoken dialogue systems ( Allen et al., 1995; Traum et al., 1996 ).
My work on TRAINS originated my interest in semantic and pragmatic interpretation in spoken conversations, as well as my research on anaphoric expressions , robust methods for semantic interpretation , and underspecification . While working on TRAINS I also got interested in the methodological issues involved in collecting corpora and using them as source of data and to evaluate the performance of natural language processing systems.
The main application for my work on semantic interpretation in dialogues are systems that allow a student to learn a second language by getting engaged in a task. In the past years I collaborated with Sarah Davies on several prototypes of this kind of systems ( Davies and Poesio, 1998a ). We also started a collaboration with a psycholinguist working on second language acquisition, Antonella Sorace, to put our work on sounder empirical footing, and with the Institute for Applied Language Studies (IALS) of the University of Edinburgh. We plan to explore the effectiveness of this kind of systems from a second language acquisition perspective by incorporating them in the classes taught at IALS. The current focus of this research is on dialogue strategies for corrective feedback ( Davies and Poesio, 1998b ).
Davies, S. and Poesio, M. 1998b. The Provision of Corrective Feedback in a Spoken Dialogue CALL System . In Proc. of ICSLP - International Conference on Spoken Language Processing, Sydney, Australia, December 1998.
Davies, S. and Poesio, M. 1998a. A CSLUrp-based spoken dialogue system for TEFL. In Proc. of STiLL - ESCA Workshop on Speech Technology in Language Learning , Marholmen, Sweden, May 1998.
Traum, D. R., Lenhart K. Schubert, Massimo Poesio, Nathaniel G. Martin, Marc N. Light, Chung Hee Hwang, Peter A. Heeman, George M. Ferguson and James F. Allen. Knowledge Representation in the TRAINS-93 Conversation System. In International Journal of Expert Systems 9(1):173-223, 1996. Also appears as TRAINS Technical Note 96-4 and TR 633 , Computer Science Dept., University of Rochester, August 1996.
Allen, J.F., L.K. Schubert, G. Ferguson, P. Heeman, C.H. Hwang, T. Kato, M. Light, N. Martin, B. Miller, M. Poesio, and D. R. Traum. 1995. The TRAINS project: a case study in building a conversational planning agent. Journal of Experimental and Theoretical AI, 7:7-48.
Definite descriptions - expressions like the car - are one of the most frequent constructions in natural language, and have originated a great deal of research in semantics. I have long been interested in the properties of this type of noun phrases, and made proposals concerning both their semantics and the way they are interpreted.
My initial interest was in definite descriptions that seem to have `lost' their presuppositional status - that we called weak definites ( Poesio, 1994 ). This initial work was based on the assumption that the defining characteristic of definites is familiarity, as suggested, e.g., by Heim (1982). More recently, I have been led to reconsider this assumption, primarily because of my work with Renata Vieira on analyzing the uses of definite descriptions in corpora. We found that more than half of definite descriptions are not `familiar' in an obvious sense. As a result, I began considering theories based on the competing hypothesis, that what characterizes definites is uniqueness. Specifically, I concentrated on Loebner's (1987) theory. Using the GNOME corpus, we compared the familiarity-based account with Loebner's; preliminary results of this work are discussed in (Poesio, 2001 )
I've felt for a long time that studying definite description interpretation is a good way to begin exploring the role of commonsense reasoning in language understanding. My earlier work on the topic was motivated by my work on the TRAINS project . I developed a theory of definite description interpretation, which made use of research on the formal semantics of discourse and on defeasible reasoning ( Poesio, 1992; Poesio, 1993b; Poesio, 1994b ). Just as in the case of the work on scope disambiguation I did as part of my dissertation, I felt that the main problems to be faced in developing theories of this type were the scarcity of empirical evidence and the difficulty in evaluating them. As a result, in my most recent work in this area I focused on collecting such empirical evidence, and on finding ways of developing computational models whose performance could be measured.
Massimo Poesio, 2001. "Definites: familiarity or functionality? a corpus-based study", Presented at Sinn und Bedeutung VI, Osnabrueck, October 2001.
Poesio, M. 1994c. Weak definites. In M. Harvey and L. Santelmann, editors, Proceedings of the Fourth Conference on Semantics and Linguistic Theory, SALT-4 . Cornell University Press, pages 282-299.
Poesio, M. 1994b. Focusing in the interpretation of definite descriptions. In P. Bosch and R. van der Sandt, editors, Proceedings of the Workshop on Focus in NLP, Schloss Wolfsbrunnen, June. Available as Working Paper 6, IBM Deutschland.
Poesio, M. 1993b. A situation-theoretic formalization of definite description interpretation in plan elaboration dialogues. In P. Aczel, D. Israel, Y. Katagiri, and S. Peters, editors, Situation Theory and its Applications, vol.3 . CSLI, Stanford, chapter 12, pages 339-374.
Poesio, M. 1992. Conversational events and discourse state change: A preliminary report. In Proceedings of the Third Conference on Principles of Knowledge Representation and Reasoning, Boston, MA, October.
Poesio, M. and A. Zucchi. 1992. On telescoping. In Proceedings SALT-2 , Columbus, Ohio, May.