In 2002 the group carried out research activity on various topics, listed below:
1.Syntactic analysis and treebank devlopment
2.Agent models.
3.Ontologies and Legal Knowledge
1) Syntactic analysis and robust methods for natural language processing.
In 2002, the group enriched the set of linguistic resources available from the Natural Language Group of the Department. In particular, the treebank of Italian sentences annotated according to the Dependency approach has doubled in size, reaching 1200 sentences; 200 of the new sentences have been extracted from the Italian Civil Code, and they constitute a first nucleus of syntactically annotated sentences with legal content, which has been planned to support the activity on the third topic (ontologies), and that will enable us to start up an activity on semantic interpretation. Within this field, the group has also developed a new scheme of labels for dependency arcs, which merges, in a theoretically principled way, information of lexical, syntactic, and semantic nature. Another resource developed by the NLP group is a robust chunk-based parser. A first version was available at the end of 2001. The new version, which has been implemented in 2002 and will be released at the end of february 2003, has considerably higher performances than the previous version (currently, about 19% of arc label errors, including attachment and grammatical functions) and also includes a first-level treatment of traces.
A second line of research concerns the study of formal methods for the representation of syntactic knowledge.
Empirical psycholinguistic analyses show that incrementality is one important feature of human sentence processing. The work in this line of research focuses on the possibilty of introducing this characteristics into the grammars used by a syntactic unit of a natural language processor. Following this idea we defined a new grammatical formlism, called DV-TAG (Dynamic Version of Tag) and we studied its strong and weak generative power. DV-TAG enriches Lexicalized Tree Adjoining Grammars formalism, a well studied mildly context sensitive formalism, by adding some constraints on the derivation process. In this way we respect the strong incrementality hypothesis, which restricts incrementality to the case in which the parser-generator mantains a fully connected tree at each state.
Thirdly, substantial advancements have been made on psycholinguistic models of sentence processing. New experiments have been carried out on the Penn Treebank, supporting the idea that the development of natural parsing strategies (in particular as regards the problem of prepositional attachment) is based on the available linguistic evidence. In particular, it has been shown that psycholinguistic criteria as Minimal Attachment and Late Closure are supported by the fact that the number of structures found in the corpus that respect the criteria is significantly greater (from a statistical point of view) than the number of structures that do not conform to them.
Finally, new results have been obtained on NLP models based on neural net models.
2) Agent models
During year 2002, the research about autonomous agents theory focused on the topics of the reactivity of agents to the changes in their environment and on the formalization of norms in multi-agent systems.
Using a hierarchical planning paradigm we developed a replanning strategy for dealing with the problem of updating the current intentions of an agent to face a new situation which occurred. The replanning process proceeds by making the plan more partial and then refining it again without restarting the planning process by first principles. After the implementation of such algorithm, we started an evaluation phase to compare the performance of the replanning strategy with respect to the performance of a planner which has to build an entire new plan; the evaluation is providing us with promising results.
For what concerns the formalization of the concept of norms for multiagent systems we proposed a definition of obligation based on sanctions: an agent who is subject to an obligation, in order to decide what to do, has to take into account the fact that if he does not fulfil the obligation may be sanctioned by a normative agent. We use this definition both in a framework based on a decision and game theoretic planner and in a logical framework, the latter work is made in cooperation with the Computer Science Department of Vrije University of Amsterdam.
3) Ontologies and Legal Knowledge
The activity on ontologies has been carried out in cooperation with the Department of Scienze Giuridiche of the University of Torino. Our group has developed a new ontological model of norms, based on previous studies on planning. In particular, the model includes a formalization (in a LOOM-like language) of the concepts of plan and behaviour, showing that norms are reasonably integrated in the model as "constraints on behaviour". Within this framework, the notion of "authority" and Hohfeld's Legal Relations find easily their proper place. In order to apply the model to a case study, in 2002 we have faced the problem of the ontological representation of legal Fruits. So, a rich ontology has been developed, including notions such as Good, Asset, Property, Contract, to which Civil Fruits have been suitably connected. A report describing the first version of this ontology has been distributed in December, 2002. Further details will be added in 2003, when also the comparison with other legal systems (where the notion of Legal Fruit does not exist) will be started up.
This research is being carried out also in partial cooperation with the Special Interest Group on Legal Ontologies of the European Network OntoWeb, in particular with groups located in Roma and Trento. The research takes also advantage from the agent models described in the previous section, and will exploit, in the next year, the syntactic treebank described in Section 1.
