Abstract:
This study is a first attempt at formalizing the concept of Musi- cal Relevance from a cognitive and computational perspective. We elaborate on Sperber and Wilson's Relevance Theory, and extend it to account for musical cognition, involving both lis- tening and understanding. Our claim is that the application of the concept of Cognitive Relevance to music would permit us to partially explain hearers' behavior and composers' choices. A computational model of Musical Relevance could also con- tribute to the formulation of a general computational theory of musical cognition. In turn, formulating an algorithm to com- pute Musical Relevance can shed light on the computational nature of the broader cognitive principle of relevance. We pro- pose to unify Relevance Theory with the Generative Theory of Tonal Music, in order to compute Musical Relevance. We started implementing a system to test the proposed approach over simple examples and report about the results in a prelim- inary experimentation.

Abstract:
In this work we present the TULSI system (so named after Turin University Legal Semantic Interpreter), a system to produce automatic annotations of normative documents through the extraction of modificatory provisions. TULSI relies on a deep syntactic analysis and a shallow semantic interpreter that are illustrated in detail. We report the results of an experimental evaluation of the system and discuss them, also suggesting future directions for further improvement.

Abstract:
In this paper we present a prototype for automatically identifying and classifying types of modifications in Italian legal text. The prototype is part of the Eunomos system, a legal knowledge management service that integrates and makes available legislation from various sources, while finding definitions and explanations of legal concepts in a given context. The design of the prototype is grounded on the error analysis of a previous prototype. The latter made use of dependency relations provided by the TUP parser, a multi-purpose parser for Italian. Since those syntactic relations were responsible of the majority of errors, we decided in the present tool to ignore them, and to rewrite an ad-hoc shallow parsing, based on the morphological analysis of the legal text (still provided by the TUP parser). We obtained performances much greater than those of the initial prototype. In particular, the level of precision of the classification in output is now close to 100%.

Abstract:
In this paper we address the problem of how autonomous agents acting on behalf of their users reason about contracts inside virtual organizations. Contracts are modelled as legal institutions: systems of legal rules which allow to change the regulative and constitutive rules of an organization. The methodology we use is to attribute to organizations mental attitudes, like beliefs, desires and goals. Agents reason about the effect of contracts in the interaction with other agents by means of game theoretic techniques

Abstract:
In this paper we introduce an ontology based on the notion of agent to represent and reason about social reality. We model social constructions as agents, for example, groups, organizations, normative systems, and roles, and we attribute mental attitudes to them. Roughly, we define obligations or regulative norms as goals of the normative system, constitutive norms as beliefs of the normative system, joint, shared, mutual and social beliefs, desires and goal as beliefs, desires and goals of group, responsibilities of an agent as goals of the role he plays, and the required expertise of an agent as beliefs and actions of the role he plays. In this way, we achieve a uniform framework for a large variety of concepts using a small vocabulary, and, in particular, basing it on notions, like mental attitudes, which are commonly used in agent theories. The proposed ontology is modelled using a description logic.

Abstract:
In this paper we address the problem of defining roles in organizations like e-trade ones. The methodology we use is to model roles according to the agent metaphor: we attribute to roles mental attitudes, like beliefs, desires and goals, we relate them to the agents required expertise and responsibilities, and we model role behavior in game theoretic terms. Analogously, the organization is modelled as an agent which acts as a normative system: it imposes obligations to roles and to the agents playing the roles

Abstract:
A theory of rational decision making in normative multiagent systems has to distinguish among the many reasons why agents fulfill or violate obligations. We propose a classification of such reasons for single cognitive agent decision making in a single normative system, based on the increasing complexity of this agent. In the first class we only consider the agents motivations, in the second class we consider also its abilities, in the third class we consider also its beliefs, and finally we consider also sensing actions to observe the environment. We sketch how the reasons can be formalized in a normative multiagent system with increasingly complex cognitive agents.

Abstract:
In this paper we formalize the specification of games in the trias politica using Rao and Georgeff's specification language BDICTL*. In particular, we generalize Rao and Georgeff's specification of single agent decision trees to multiagent games, for which we introduce observations and recursive modelling, in this setting we formalize obligations, and we characterize four kinds of agents, called legislators, judges, policemen and citizens. Legislators are characterized by their power to create and revise obligations, judges are characterized by their power to count behavior of citizens as violations, and policemen are characterized by their ability to sanction behavior.

Abstract:
We discuss a model of cooperation among autonomous agents, based on the attribution of mental attitudes to groups: these attitudes represent the shared beliefs and objectives and the wish to reduce the costs for the members. When agents take a decision they have to recursively model what their partners are expected to do under the assumption that they are cooperative, and they have to adopt the goals and desires attributed to the group: otherwise, the other members consider them uncooperative and thus liable.

Abstract:
We are interested in the design of policies for virtual communities of agents based on the grid infrastructure. In a virtual community agents can play both the role of resource consumers and the role of resource providers, and they remain in control of their resources. We argue that this requirement imposes a distinction between the authorization to access a resource given by the virtual community and the permission to do so issued by the resource providers. Our model is based on a logical multiagent framework that distinguishes the three roles of resource consumption, provision, and of authorization.

Abstract:
We are interested in the design of policies for virtual communities of agents based on the grid infrastructure. In a virtual community agents can play both the role of resource consumers and the role of resource providers, and they remain in control of their resources. We argue that this requirement imposes a distinction between the authorization to access a resource given by the virtual community and the permission to do so issued by the resource providers. Our model is based on a logical multiagent framework that distinguishes the three roles of resource consumption, provision, and of authorization.

Abstract:
Agent theory proposes to model the behavior of complex software systems in terms of mental attitudes like belief, desires, goals, intentions and obligations, ranging from, e.g., the PRS system [7] to the more recent BOID architecture [3]. Decision-theoretic deliberation captures concepts and reasoning mechanisms from agent theory in standard decision-theoretic terms. This is an ambitious enterprise, as it has the objective to bridge the worlds of decision theory and agent theory. Thus far, several partial results on the decision-theoretic characterization have been obtained. The relation between beliefs (as well as defaults) and probabilistic techniques has been studied for some time, there are characterizations of desires and goals in decision-theoretic terms [6], there are various interpretations of obligations and norms, for example as social laws [8], and there are preliminary results on intention [1]. See our comparison paper [4] for an overview. The most problematic issue in decision-theoretic deliberation is the characterization of intention. Roughly, whereas beliefs have been related to probabilities, desires to utilities, and obligations to social laws, intentions do not seem to have an obvious counterpart in classical game and decision theory. However, most discussions on the popular BDI model have focussed on the role of intention in deliberation [2]. Consequently, we believe that intention is the benchmark example of decision-theoretic deliberation.

Abstract:
In this abstract we consider dialogues that have a normative aspect. A dialogue is regulated by norms, but can also establish new norms. Certain utterances count as a particular dialogue move in some dialogue game, which creates obligations and permissions for the participants. But norms do not operate in isolation; we study their relation to mental attitudes of participants, in particular beliefs (information), desires, goals or intentions.

Abstract:
The concept of power plays an important role in the social sciences and Castelfranchi [2, 3] emphasizes the importance of this concept for multiagent systems. In [1] we build upon this work by distinguishing four viewpoints on multiagent systems: a mind structure, a power structure, a dependence structure and a coalition structure. These viewpoints are increasingly abstract conceptualizations of systems as collections of autonomous cognitive agents. In this paper we formally define one kind of power and coalition structures, and an abstraction from this power to its coalition structure.

Abstract:
An important aim in the field of Multiagent Systems is to study emergent social structures, such as groups and collectives. The relevance of social structures in Distributed Artificial Intelligence, Artificial Life, Sociology necessitates a well motivated definition of their characteristics. The research question of this paper is to formalize the results obtained by Castelfranchi that try to bridge the gap between the BDI model and the macro-level of Multiagent Systems. We presented in (AAMAS04)general social viewpoints on Multiagent Systems and in (ECAI04) we have detailed a derivation of coalition structures from power structures. In this paper we continue the research, and we define power in terms of cognitive (or mind) view, i.e., starting from the characteristics of the single agents as their goals and skills. Properties of collectives, as mutual dependence or cooperation, are also hierarchically defined by means of the definition of power. It is shown that upper levels of this hierarchy need mental model of the agents more and more complex.

Abstract:
In this paper we present a formal model of Multiagent Systems to analyze the relations of power and dependence underlying group behaviors such as cooperation. Inspired by the work of Castelfranchi we define these relations by means of a description of goals and skills of single agents. We show how our framework can be used to describe social and organizational structures as emergent properties of a collection of individuals.

Abstract:
We reconsider the BDI agents model by exploiting a it decision theoretic planning framework together with a proposal for bounded rationality. In particular, we show how the interplay between a planner based on an it anytime algorithm and a it meta-deliberation module for dealing with bounded resources sheds light on the notion of it intention as proposed by Michael Bratman.

Abstract:
Using a decision theoretic planner and it bounded rationality reasoning, we reconsider the analysis of intentions proposed by Michael Bratman. Intentions are defined as predominant desires that can be achieved; but identifying which desire is the predominant one is a costly effort: the it information produced during the planning phase is exploited when it meta-deliberating about the reconsideration of the current intentions, giving them stability.

Abstract:
We present a reactive agent architecture which incorporates decision-theoretic notions to drive the deliberation and meta-deliberation process, and illustrate how this architecture can be exploited to model an agent who reacts to contextually instantiated norms by monitoring for norm instantiation and replanning its current intentions.

Abstract:
In some strategic situations, which can be represented as 2x2 matrix games in the it prisoner dilemma style, an agent prefers to enter the interaction only in case he or the other player are subject to some obligations which incentives them to stay it committed to a cooperative behavior. Obligations presuppose the existence of a third agent who is in charge of sanctioning the misbehaving agents and/or rewarding the agents who stick to their commitments. In this paper, we propose a model of game situations where we introduce an explicit it third party agent who has the role of enforcing these obligations.

Abstract:
We present an algorithm for replanning in a reactive agent architecture which incorporates decision-theoretic notions to drive the planning and metadeliberation process. The deliberation component relies on a refinement planner which produces plans with optimal expected utility. The replanning algorithm we propose exploits the planner’s ability to provide an approximate evaluation of partial plans: it starts from a fully refined plan and makes it more partial until it finds a more partial plan which subsumes more promising refinements; at that point, the planning process is restarted from the current partial plan.

Abstract:
In recent years, multi-agent systems have become one of the most promising approaches for organizing software. However, the demand for capabilities of autonomous decision making poses some requirements on agent architectures. It is necessary, in fact, that the behavior of an agent be suitably constrained to make it socially compatible with the community of agents. In this paper, we propose a model for obligations and rules which is inspired to E. Goffman's work in sociology, and which can be integrated in existing BDI (Belief, Desire, Intention) agent architectures. We show that it Decision Theory and it Anticipatory Coordination can be the basic building blocks for an agent model where obligations are associated with sanctions (and with another agent, the it normative agent, who takes care of making the obligation respected). Anticipatory Coordination is required to foresee the normative agent reactions, while Decision Theory enables an agent to choose in a rational way the most promising line of action.

Abstract:
We argue that anaphora cannot be resolved at the level of the formal language representing meaning, but, rather, by making direct reference to the it extension of the sentences. Johnson-Laird's mental models theory provide the tool for coping with extensional representations in a cognitively plausible way.

Abstract:
Agent theories have developed concepts and methodologies that can be a pplied for having a better understanding of reasoning about obligations. This work proposes an agent-based framework for modeling obligations and norms: agents are able to deal with norms and to decide autonomously whether to respect them or not. The key idea is that the addressee of the norm explicitly models the agent who watches on the norms and who can sanction him.

Abstract:
The paper presents theoretical aspects and practical issues related to the development of a grammatical relation's system for corpus annotation. The grammatical relations are arranged on a default inheritance hierarchy based on syntactic and semantic features. Preliminary tests on the annotation of an Italian treebank (the Turin University Treebank) show that the system implements a reasonable trade-off between richness of the representation and tractability of the annotation task.

Abstract:
This paper describes a plan-based agent architecture for modeling NL cooperative dialogue; in particular, the paper focuses on the interpretation dialogue and explanation of its coherence by means of the recognition of the speakers' underlying intentions. The approach we propose makes it possible to analyze and explain in a uniform way several apparently unrelated linguistic phenomena, which have been often studied separately and treated via ad-hoc methods in the models of dialogue presented in the literature. Our model of linguistic interaction is based on the idea that dialogue can be seen as any other interaction among agents: therefore, domain-level and linguistic actions are treated in a similar way. Our agent architecture is based on a two-level representation of the knowledge about acting: at the metalevel, the Agent Modeling plans describe the recipes for plan formation and execution (they are a declarative representation of a reactive planner); at the object level, the domain and communicative actions are defined. The Agent Modeling plans are used to identify the goals underlying the actions performed by an observed agent; the recognized plans constitute the dialogue context, where the intentions of all participants are stored in a structured way, in order to be used in the interpretation of the subsequent dialogue turns.

Abstract:
In this paper we consider the consequences for dialog modeling of a ne w proposal of cooperation based on the concepts of group utility, goal adoption and anticipatory coordination. On the one hand, the model accounts for the contextual occurrence of communicative acts during cooperative activity; on the other hand, it models grounding phenomena in dialog, seen as a cooperative activity, without explicitly prescribing them.

Abstract:
Agent theories have developed concepts and methodologies that can be applied for having a better understanding of reasoning about obligations. This work proposes an agent-based framework for modeling obligations and norms: agents are able to deal with norms and to decide autonomously whether to respect them or not. The key idea is that the addressee of the norm explicitly models the agent who watches on the norms and who can sanction him.

Abstract:
In this paper, we propose an definition of cooperation to shared plans that takes into account the benefit of the whole group, where the group's benefit is computed by considering also the consequences of an agent's choice in terms of the actions that the other members of the group will do. In addition, the members of a group consider whether to adopt the goals of their partners: an agent should adopt these goals only when the adoption results in an increase of the group's benefit.

Abstract:
We propose a model where dialog obligations arise from the interplay of social goals and intentions of the participants: when an agent is addressed with a request, the agent's decision to commit to the requester's linguistic and domain goals is motivated by a trade-off between the preference for preventing a negative reaction of the requester and the cost of the actions needed to satisfy the goals.

Abstract:
The paper describes an annotation schema developed for an Italian treebank, the Turin University Treebank (TUT). The schema is based on the dependency format and is characterized by a rich and flexible grammatical relations system. In fact, dependency relations have been organized in a taxonomy and allows for an annotation with variable degrees of specification.

Abstract:
Many natural language researchers are currently turning their attention to treebank development and trying to achieve accuracy and corpus data coverage in their representation formats. This paper presents a data-driven annotation schema developed for an Italian treebank ensuring data coverage and consistency between annotation of linguistic phenomena. The schema is a dependency-based format centered upon the notion of predicate-argument structure augmented with traces to represent discontinuous constituents. The treebank development involves an annotation process performed by a human annotator helped by an interactive parsing tool that builds incrementally syntactic representation of the sentence. To increase the syntactic knowledge of this parser, a specific data-driven strategy has been applied. We describe the cyclical development of the annotation schema highlighting the richness and flexibility of the format, and we present some representational issues.

Abstract:
In this paper, we propose a logical description of the mechanisms which cause a speech act to be impolite, and of how the indirect expressions may prevent speakers from offending their partners. We specifically focus on conventional indirect speech acts, providing a formal framework to recognize the beliefs underlying them and the way how the possible offenses produced by communicative actions may be blocked by using politeness techniques.

Abstract:
In this paper, a representation formalism based on actions and hierarchical plans is proposed to model the aspectual and temporal composition of sentences. Action verbs are interpreted as instances of action schemata that include a plan body; the other linguistic elements, like tense and adverbs, are evaluated by means of rules that add constraints to event representations.

Abstract:
In this paper, we want to show that it mental models provide the adequate level for explaining complex anaphoric phenomena, like donkey sentences. The key idea is that anaphors can be more easily resolved by referring to an extensional representation of the meaning of sentences, like the one provided by mental models.

Abstract:
We claim mental models are a framework that allow to shed light on the phenomenon of presuppositions. A plan-based lexical representation for verbs, together with the effect of conversational implicatures that discharge possible mental models, are the key f eatures of this proposal.

Abstract:
In this paper, we suggest that AI techniques, especially the ones developed in the field of agent modeling and intelligent interfaces, can be exploited both to build systems for narrative understanding and to allow agents to describe their own behavior in a narrative style. In particular, we will show how it is possible to exploit a model of dialog interpretation for building a system that understands stories and produces a representation of the characters' plans and intentions.

Abstract:
We will consider the role played by the notion of agent intentions in dialog and, particularly, in explaining conversational cooperation. The decision to cooperate towards the success of the communication accounts for grounding, requests of repair, repairs to misunderstandings, and other related phenomena. In the model, intentions arise from goals on the basis of two factors: the relevance of a goal, and the cost of the actions needed to achieve it. The adoption of the intention of cooperating at the conversational level is motivated by the fact that the refusal to do low-cost actions, in case they are supported by low-relevance goals, is usually interpreted as an offense to the requester of the action.

Abstract:
Agent theories take as their paradigm human intentional behavior; however, as far as agent interaction is concerned, they have not yet satisfactorily taken into account the requirements raised by studies on human Natural Language communication, the most developed means of interaction. The fundamental missing point is the role of intention recognition, which is the basis of human dialog interactions. In this paper, we describe a declarative agent architecture for modeling social agent behavior, with particular attention to Natural Language dialog. The architecture can be used both to recognize a speaker's intentions and generate intention-driven behavior in agent interactions; therefore, it is suited to interface agents for HCI, which require a friendly interaction with users.

Abstract:
We describe a plan-based agent architecture that models misunderstandings in cooperative NL agent communication; it exploits a notion of coherence in dialogue based on the idea that the explicit and implicit goals which can be identified by interpreting a conversational turn can be related with the previous explicit / implicit goals of the interactants. Misunderstandings are hypothesized when the coherence of the interaction is lost (i.e. an unrelated utterance comes). The processes of analysis (and treatment) of a misunderstanding are modeled as rational behaviors caused by the acquisition of a supplementary goal, when an incoherent turn comes: the agent detecting the incoherence commits to restore the intersubjectivity in the dialogue; so, he restructures his own contextual interpretation, or he induces the partner to restructure his (according to who seems to have made the mistake). This commitment leads him to produce a repair turn, which initiates a subdialogue aimed at restoring the common interpretation ground. Since we model speech acts uniformly with respect to the other actions (the domain level actions), our model is general and covers misunderstandings occurring at the linguistic level as well as at the underlying domain activities of the interactants.

Abstract:
This paper describes an approach to POS tagging based on the automatic refinement of manually written linguistic tagging rules. The refinement was carried out by means of a learning algorithm based on decision trees. The tagging rules work on ambiguity classes: each input word undergoes a morphological analysis and a set of possible tags is returned. The set of tags determines the ambiguity class: the suitable subset of tagging rules is applied and a single tag is returned. This method can be applied both with the original set of manual rules and with the set of refined rules. The experiments have been performed on the Italian language. The results show that the refined set reduces the tagging error of about 17%. The noticeable feature of the approach is that the original set of rules is not modified, so that they retain their linguistic value, but new rules are added to handle problematic cases.

Abstract:
This paper describes a framework for the representation and interpretation of indirect speech acts, relating them to the politeness phenomenon, with particular attention to the case of requests. The speech acts are represented as actions of a plan library and are activated on the basis of the presence of syntactic and semantic information in the linguistic form of the input utterance. The speech act analyzer receives in input the semantic representation of the input sentence and uses the politeness indicators to climb up the decomposition and generalization hierarchies of acts encoded in the library. During this process, it eliminates the indicators and collects the negated presuppositions (represented as effects of the indirect speech act) that characterize the politeness forms. Some cyclic paths in the hierarchy allow the system to cope with complex sentences including nested politeness indicators. In the proper places of the hierarchy the semantic representation of the input sentence is converted into a domain action in order to start-up, when needed, the domain-level plan recognition process.

Abstract:
This paper discusses some of the problems related to the representation of uncertain knowledge and to the combination of evidence degrees in rule-based expert systems. Some of the methods proposed in the literature are briefly analysed with particular attention to the Subjective Bayesian Probability (used in PROSPECTOR) and the Confirmation Theory adopted in MYCIN. The paper presents an integrated approach based on Possibility Theory for evaluating the degree of match between the set of conditions occurring in the antecedent of a production rule and the input data, for combining the evidence degree of a fact with the strength of implication of a rule and for combining evidence degrees coming from different pieces of knowledge. The semantics of the logical operators AND and OR in possibility theory and in our approach are compared. Finally, the definitions of some quantifiers like AT LEAST n, AT MOST n, EXACTLY n are introduced.

Abstract:
This paper describes a text-to-speech translation system for the Italian language. It is based on a formal translator which converts a graphemic text into the corresponding phonemic representation and on a suprasegmental features descriptor. The translator is implemented according to the formal techniques developed for lexical and syntactical analysis of artificial languages. The suprasegmental features descriptor determines pause positions, phoneme durations and pitch contour on the sole basis of the punctuation marks and the accent marks present in the text.