Ladjel Bellatreche

Referential Horizontal Partitioning Selection Problem in Data Warehouses: Hardness Study and Selection Algorithms

Horizontal Partitioning has been largely adopted by the database community, where it took a significant part in the physical design process. Actually, it is supported by most commercial database systems (DBMS), where a native Data Definition Language for decomposing tables/materialized views using various modes is proposed. In traditional databases, horizontal partitioning has been largely studied, where several fragmentation algorithms were proposed to partition tables in isolation. In the relational data warehouse environment, horizontal partitioning consists in decomposing the whole warehouse schema into sub schemas, where each schema contains fragments of dimension and fact tables. Dimension tables are fragmented using the primary partitioning mode, whereas the fact table is divided using referential mode. In this article, the authors first focus on the evolution of horizontal partitioning in commercial DBMS motivated by decision support applications. Secondly, they give a formalization of the referential fragmentation schema selection problem in the data warehouse and they study its hardness to select an optimal solution. Due to its high complexity, they develop two algorithms: hill climbing and simulated annealing with several variants to select a near optimal partitioning schema. Finally, extensive experimental studies are conducted using the data set of APB1 benchmark to compare the quality the proposed algorithms using a mathematical cost model. Based on these experiments, some recommendations are given to advise database administrator for well using horizontal partitioning.

Models and data engineering

Data and models are two well established communities that are continuously contributing in new challenges in different research domains including cyberphysical systems (Yoo and Shon, 2016 [1]), cloud computing (Jararweh etal., 2016 [2]), service oriented applications, social networks (Yahyaoui etal., 2013 [3]), big data (with its five Vs characteristics: Volume, Variety, Velocity, Veracity and Value) (Hsu, 2014), etc. The success story of data and models communities is mainly based on the availability of foundations relying on formal methods (Hassan etal., 2016), modelling methods (Ameur etal., 2014), storage systems and platforms (Liu etal., 2015 [7]), advanced optimization structures, benchmarking, scalability, etc. These foundations are usually associated with tools and commercial and academic systems.The selected papers for this special issue address a variety of topics and concerns in models and data fields, including advanced databases, engagement systems, embedded and complex systems, etc.

Dimension table driven approach to referential partition relational data warehouses

Most of business intelligence applications use data warehousing solutions. The star schema or its variants modelling these applications are usually composed of hundreds of dimension tables and multiple huge fact tables. Referential horizontal partitioning is one of physical design techniques adapted to optimize queries posed over these schemes. In referential partitioning, a fact table can inherit the fragmentation characteristics from dimension table(s). Most of the existing works done on referential partitioning start from a bag containing selection predicates defined on dimension tables, partition each one based on its predicates and finally propagate their fragmentation schemes to the fact table. This procedure gives all dimension tables the same probability to partition the fact table which is not always true. In order to ensure a high performance of the most costly queries, the identification of relevant dimension table(s) to referential partition a fact table is a crucial issue that should be addressed. In this paper, we first study the complexity of the problem of selecting dimension table(s) used to partition a fact table. Secondly, we present strategies to perform their selection. Finally, to validate of our proposal, we conduct intensive experimental studies using a mathematical cost model and the obtained results are verified on Oracle11G DBMS.

Effectively and Efficiently Designing and Querying Parallel Relational Data Warehouses on Heterogeneous Database Clusters: The F&A Approach

In this paper, a comprehensive methodology for designing and querying Parallel Rational Data Warehouses PRDW over database clusters, called Fragmentation & Allocation F&A is proposed. F&A assumes that cluster nodes are heterogeneous in processing power and storage capacity, contrary to traditional design approaches that assume that cluster nodes are instead homogeneous, and fragmentation and allocation phases are performed in a simultaneous manner. In classical approaches, two different cost models are used to perform fragmentation and allocation, separately, whereas F&A makes use of one cost model that considers fragmentation and allocation parameters simultaneously. Therefore, according to the F&A methodology proposed, the allocation phase/decision is done at fragmentation. At the fragmentation phase, F&A uses two well-known algorithms, namely Hill Climbing HC and Genetic Algorithm GA, which the authors adapt to the main PRDW design problem over heterogeneous database clusters, as these algorithms are capable of taking into account the heterogeneous characteristics of the reference application scenario. At the allocation phase, F&A introduces an innovative matrix-based formalism capable of capturing the interactions among fragments, input queries, and cluster node characteristics, driving the data allocation task accordingly, and a related affinity-based algorithm, called F&A-ALLOC. Finally, their proposal is experimentally assessed and validated against the widely-known data warehouse benchmark APB-1 release II.

Extending the ANSI/SPARC Architecture Database with Explicit Data Semantics: An Ontology-Based Approach

The database (DB) design process follows the traditional ANSI/SPARC architecture proposed by Bachman [1]. A conceptual model (CM) is translated into a logical model corresponding to a data specification implemented in a DB system. The physical model defines how data are stored and accessed. External models allow a DB designer to adapting data according to users requirements. Regarding the semantic exploitation of data models, this architecture has two major drawbacks [2]: (1) a strong dependency of models with designers and specific application requirements; (2) a gap between conceptual and logical models that increases with the discrepancy of the conceptual modelling languages. n nThe maintenance and/or evolution of the CM, that must be consensual when dealing with semantic integration of data sources (semantics and schema conflicts), are in the kernel of these problems. Recently, some works give more importance to CMs by materializing them in a DB [3]. In these works, the design of a CM is preceded by the design or by pre-existence of ontology. In this case, both ontology and data are represented in the DB. Such a DB is called an ontologybaseddatabase(OBDB). Hence our proposition is to extend the ANSI/SPARC architecture to support OBDBs.

Concept-based clustering of textual documents using SOM

The classification of textual documents has been widely studied. The majority of classification approaches use supervised learning methods, which are acceptable for rather small corpora allowing experts to generate representative sets of data for the training, but are not feasible for significant flows of data. Unsupervised classification methods discover latent (hidden) classes automatically while minimizing human intervention. Many such methods exist, among which Kohonen self- organizing maps (SOM), which gather a certain number of similar objects without prior information. In this paper, we evaluate and compare the use of SOMs for the classification of textual documents in two situations: a conceptual representation of texts and a representation based on n-grams.

Towards Connecting Database Applications to Ontologies

Most database applications are designed according the ANSI/SPARC architecture. When it is used a large amount of semantics of data may be lost during the transformation from the conceptual model to a logical model. As a consequence exchanging/integrating various databases or generating user interfaces for data access become difficult. Ontologies seem an interesting solution to solve these problems, since they allow making explicit the semantics of data. In this paper, we propose an ontology-based approach for designing database applications, and then, for representing explicitly the semantics of data within the database. It consists in extending the ANSI/SPARC architecture with the ontological level. Note that this extension may also be added to existing applications designed according to the ANSI/SPARC architecture, since it preserves an upward compatibility.

SISRO, conception de bases de données à partir d'ontologies de domaine

Database design methodologies require both good modelling capabilities and knowledge of the field to be modelled. The first aspect is a one of the major difficulties for designers, since they have to move from one field to another one. Moreover, resulting databases are heterogeneous and difficult to integrate. In this paper, we propose a database design methodology, called SISRO. It is based on a prior definition of a local ontology for the database to be designed. This ontology is defined by specialization of shared domain ontologies, selective import of properties, and eventually extension with necessary concepts and/ or properties. The conceptual model is defined as a fragment of local ontology. Representing both ontologies and their articulations in the resulting database provides easy access, exchange and integration of data.

Query Interaction Based Approach for Horizontal Data Partitioning

With the explosion of data, several applications are designed around analytical aspects, with data warehousing technology at the heart of the construction chain. The exploitation of this data warehouse is usually performed by the use of complex queries involving selections, joins and aggregations. These queries bring the following characteristics: (1) their routinely aspects, (2) their large number, and (3) the high operation sharing between queries. This interaction has been largely identified in the context of multi-query optimization, where graph data structures were proposed to capture it. Also during the physical design, the structures have been used to select redundant optimization structures such as materialized views and indexes. Horizontal data partitioning (HDP) is another non-redundant optimization structure that can be selected in the physical design phase. It is a pre-condition for designing extremely large databases in several environments: centralized, distributed, parallel and cloud. It aims to reduce the cost of the above operations. In HDP, the optimization space of potential candidates for partitioning grows exponentially with the problem size making the problem NP-hard. This paper proposes a new approach based on query interactions to select a partitioning schema of a data warehouse in a divide and conquer manner to achieve an improved trade-off between the optimization algorithms speed and the quality of the solution. The effectiveness of our approach is proven through a validation using the Star Schema Benchmark (100 GB) on Oracle11g.

Ontologies in engineering: the OntoDB/OntoQL platform

Ontologies have been increasingly used over the past few decades in a wide range of application domains spanning both academic and industrial communities. As ontologies are the cornerstone of the Semantic Web, the technologies developed in this context, including ontology languages, specialized databases and query languages, have become widely used. However, the expressiveness of the proposed ontology languages does not always cover the needs of specific domains. For instance, engineering is a domain for which the LIAS laboratory has proposed dedicated solutions with a worldwide recognition. The underlying assumptions made in the context of the Semantic Web, an open and distributed environment, do not apply to the controlled environments of our projects where the correctness and completeness of modeling can be guaranteed to a certain degree. As a consequence, we have developed over the last decades a specialized standard ontology language named PLIB associated with the OntoDB/OntoQL platform to manage ontological engineering data within a database. The goal of this paper is threefold: (1) to share our experience in manipulating ontologies in the engineering domain by describing their specificities and constraints; (2) to define a comprehensive classification of ontologies with respect to three main research communities: Artificial Intelligence, Databases and Natural Language Processing and (3) to present a persistent solution, called OntoDB, for managing extremely large semantic data sets associated with an ontological query language, called OntoQL. These objectives are illustrated by several examples that show the effectiveness and interest of our propositions in several industrial projects in different domains including vehicle manufacturing and CO