Alessandro D'Atri

Structure Preserving Reductions Among Convex Optimization Problems

Abstract In this paper we introduce the concept of convex optimization problem. Convex optimization problems are studied by giving a formalization of the concept of combinatorial structure, in terms of spectra of approximate solutions, and of reduction which preserves such structure. On this basis a classification of convex NP -optimization problems is introduced and is applied to study the combinatorial structure of several optimization problems associated to well-known NP -complete sets. Conditions on the approximability of such optimization problems are also given and it is shown that structurally isomorphic problems have similar approximability properties.

Distance-hereditary graphs, Steiner trees, and connected domination

Distance-hereditary graphs have been introduced by Howorka and studied in the literature with respect to their metric properties. In this paper several equivalent characterizations of these graphs are given: in terms of existence of particular kinds of vertices (isolated, leaves, twins) and in terms of properties of connections, separators, and hangings. Distance-hereditary graphs are then studied from the algorithmic viewpoint: simple recognition algorithms are given and it is shown that the problems of finding cardinality Steiner trees and connected dominating sets are polynomially solvable in a distance-hereditary graph.

Minimal representation of directed hypergraphs

In this paper the problem of minimal representations for particular classes of directed hypergraphs is analyzed. Various concepts of minimal representations of directed hypergraphs (called minimal equivalent hypergraphs) are introduced as extensions to the concepts of transitive reduction and minimum equivalent graph of directed graphs. In particular, we consider equivalent hypergraphs which are minimal with respect to all parameters which may be adopted to characterize a given hypergraph (number of hyperarcs, number of adjacency lists required for the representation, length of the overall description,etc.). The relationships among the various concepts of minimality are discussed and their computational properties are analyzed. In order to derive such results, a graph representation of hypergraphs is introduced.

VirtuE: a formal model of virtual enterprises for information markets

A vital part of a modern economy is an information market. In this market, information products are being traded in countless ways. Information is bought, modified, integrated, incorporated into other products, and then sold again. Often, the manufacturing of an information product requires the collaboration of several participants. A virtual enterprise is a community of business entities that collaborate on the manufacturing of complex products. This collaboration is often ad hoc, for a specific product only, after which the virtual enterprise may dismantle. The virtual enterprise paradigm is particularly appealing for modeling collaborations for manufacturing information products, and in this paper we present a new model, called VirtuE, for modeling such activities. VirtuE has three principal components. First, it defines a distributed infrastructure with concepts such as members, products, inventories, and production plans. Second, it defines transactions among members, to enable collaborative production of complex products. Finally, it provides means for the instrumentation of enterprises, to measure their performance and to govern their behavior.

On hypergraph acyclicity and graph chordality

Abstract Concepts of acyclicity in hypergraphs and chordality in graphs are related by showing that a hierarchy of well-studied classes of chordal graphs corresponds to the hierarchy of classes of acyclic hypergraphs studied in relational database theory (Fagin, 1983).

Dynamic query interpretation in relational databases

Abstract A new dynamic approach to the problem of determining the correct interpretation of a logically independent query to a relational database is formally described. A simple kind of conjunctive query is taken into account and the proposed disambiguating process is based on a simple user-system dialogue that consists of a sequence of decisions about the relevance (or not) of an attribute with respect to the user interpretation. A characterization of unambiguous and disambiguable databases is given and properties of such databases are studied. Approximate answers and weaker concepts of unambiguity and disambiguability are also studied. Finally, criteria to select good disambiguation strategies according to the logical independence degree and the naturalness are discussed.

The KIWIS knowledge base management system

This report describes the functionality and architecture of the KIWI system, an advanced knowledge-base environment for large database systems, which is currently being developed within the framework of the ESPRIT programme (P2424) by a consortium of industrial and research organizations consisting of Alacatel-Bell Tel. (BE), CRAI (I), Enidata (I), FORTH (GR), ORIGIN (NL), SISU (SE), the universities of Antwerp UIA (BE, Coordinating Contractor), Calabria (I), and LAquila (I).

Man-machine interfaces to medical information systems

A joint project to provide an environment for prototyping and validating advanced man-machine interaction techniques for medical information systems is reported. The topics discussed are: data models and knowledge-representation languages for describing medical information, graphical display and interaction environments, user modeling capabilities, and personalization tools. The preliminary results of the project are sketched.<<ETX>>

Browsing in geographic databases: an object-oriented approach

An interaction model for object-oriented geographic databases is presented. A browsing technique based on the model is outlined. The approach is uniform for navigating both the intensional and extensional part of the database. The proposed interaction model is flexible and is suitable for satisfying many purposes. It is able to satisfy users interested in getting a general idea about the contents of the database, as well as those with more specific tasks to accomplish. Depending on their requirements, users may control the degree of complexity of the information presented on the screen. Two basic criteria are adopted for browsing the database: logic adjacency and spatial adjacency between objects. Logic adjacency is determined by four conceptual links, while spatial adjacency is related to the two-dimensional (map-based) view of geographic entities.<<ETX>>

Interpretation of statistical queries to relational databases

A research in progress concerning the problem of extending the logical independence approach to statistical query answering is described. The main ideas and limitations of a system which assists a user in obtaining a statistical table from a relational database by describing only elements that compose the table are discussed. The proposed system assists the user in formulating statistical queries by means of a universal relation interface, i.e., independently from the underlying database structure. To generate the result, an extended relational algebra is needed to process aggregations in queries function; heuristics for the interpretation of queries and their translation in terms of database primitives are the basis of the proposed approach.