In recent years, the widespread of mobile devices has made easier and popular the activities of r... more In recent years, the widespread of mobile devices has made easier and popular the activities of recording locations visited by users and of inferring their trajectories. The availability of such large amount of spatio-temporal data opens new challenges to automatically extract information and get valuable knowledge. The many aspects of this issue have aroused the interest of researchers in several areas, such as information retrieval, and data mining, context-aware computing, security and privacy issues, urban planning and transport management. Recently, there has been a strong interest in understanding how people move during their common daily activities in order to get information about their behaviors and habits. In this paper we describe considerable recent research works related to the analysis of mobile spatio-temporal data, focusing on the study of social habits and behaviors. We provide a general perspective on studies on human mobility by depicting and comparing methods and algorithms, highlighting some critical issues related to information extraction from spatio-temporal data, and future research directions.
ABSTRACT We propose a mapping from a database conceptual design to a schema for XML that produces... more ABSTRACT We propose a mapping from a database conceptual design to a schema for XML that produces highly connected and nested XML structures. We first introduce two alternative definitions of the mapping, one modeling entities as global XML elements and expressing relationships among them in terms of keys and key references (flat design), the other one encoding relationships by properly including the elements for some entities into the elements for other entities (nest design). Then we provide a benchmark evaluation of the two solutions showing that the nest approach, compared to the flat one, leads to improvements in both query and validation performances. This motivates us to systematically investigate the best way to nest XML structures. We identify two different nesting solutions: a maximum depth nesting, that keeps low the number of costly join operations that are necessary to reconstruct information at query time using the mapped schema, and a maximum density nesting, that minimizes the number of schema constraints used in the mapping of the conceptual schema, thus reducing the validation overhead. On the one hand, the problem of finding a maximum depth nesting turns out to be NP-complete and, moreover, it admits no constant ratio approximation algorithm. On the other hand, we devise a graph-theoretic algorithm, NiduX, that solves the maximum density problem in linear time. Interestingly, NiduX finds the optimal solution for the harder maximum depth problem whenever the conceptual design graph is either acyclic or complete. In randomly generated intermediate cases of the graph topology, we experimentally show that NiduX finds a good approximation of the optimal solution.
ABSTRACT The theoretical interest and the practical relevance of a systematic treatment of multip... more ABSTRACT The theoretical interest and the practical relevance of a systematic treatment of multiple temporal dimensions is widely recognized in the database and information system communities. Nevertheless, most relational databases have no temporal support at all. A few of them provide a limited support, in terms of temporal data types and predicates, constructors, and functions for the management of time values (borrowed from the SQL standard). One (resp., two) temporal dimensions are supported by historical and transaction-time (resp., bitemporal) databases only. In this paper, we provide a relational encoding of a conceptual model featuring four temporal dimensions, namely, the classical valid and transaction times, plus the event and availability times. We focus our attention on the distinctive technical features of the proposed temporal extension of the relation model. In the last part of the paper, we briefly show how to implement it in a standard DBMS.
In recent years, the widespread of mobile devices has made easier and popular the activities of r... more In recent years, the widespread of mobile devices has made easier and popular the activities of recording locations visited by users and of inferring their trajectories. The availability of such large amount of spatio-temporal data opens new challenges to automatically extract information and get valuable knowledge. The many aspects of this issue have aroused the interest of researchers in several areas, such as information retrieval, and data mining, context-aware computing, security and privacy issues, urban planning and transport management. Recently, there has been a strong interest in understanding how people move during their common daily activities in order to get information about their behaviors and habits. In this paper we describe considerable recent research works related to the analysis of mobile spatio-temporal data, focusing on the study of social habits and behaviors. We provide a general perspective on studies on human mobility by depicting and comparing methods and algorithms, highlighting some critical issues related to information extraction from spatio-temporal data, and future research directions.
ABSTRACT We propose a mapping from a database conceptual design to a schema for XML that produces... more ABSTRACT We propose a mapping from a database conceptual design to a schema for XML that produces highly connected and nested XML structures. We first introduce two alternative definitions of the mapping, one modeling entities as global XML elements and expressing relationships among them in terms of keys and key references (flat design), the other one encoding relationships by properly including the elements for some entities into the elements for other entities (nest design). Then we provide a benchmark evaluation of the two solutions showing that the nest approach, compared to the flat one, leads to improvements in both query and validation performances. This motivates us to systematically investigate the best way to nest XML structures. We identify two different nesting solutions: a maximum depth nesting, that keeps low the number of costly join operations that are necessary to reconstruct information at query time using the mapped schema, and a maximum density nesting, that minimizes the number of schema constraints used in the mapping of the conceptual schema, thus reducing the validation overhead. On the one hand, the problem of finding a maximum depth nesting turns out to be NP-complete and, moreover, it admits no constant ratio approximation algorithm. On the other hand, we devise a graph-theoretic algorithm, NiduX, that solves the maximum density problem in linear time. Interestingly, NiduX finds the optimal solution for the harder maximum depth problem whenever the conceptual design graph is either acyclic or complete. In randomly generated intermediate cases of the graph topology, we experimentally show that NiduX finds a good approximation of the optimal solution.
ABSTRACT The theoretical interest and the practical relevance of a systematic treatment of multip... more ABSTRACT The theoretical interest and the practical relevance of a systematic treatment of multiple temporal dimensions is widely recognized in the database and information system communities. Nevertheless, most relational databases have no temporal support at all. A few of them provide a limited support, in terms of temporal data types and predicates, constructors, and functions for the management of time values (borrowed from the SQL standard). One (resp., two) temporal dimensions are supported by historical and transaction-time (resp., bitemporal) databases only. In this paper, we provide a relational encoding of a conceptual model featuring four temporal dimensions, namely, the classical valid and transaction times, plus the event and availability times. We focus our attention on the distinctive technical features of the proposed temporal extension of the relation model. In the last part of the paper, we briefly show how to implement it in a standard DBMS.
Uploads
Papers by Donatella Gubiani