Adriana Marotta

Data warehouse design: a schema-transformation approach

This paper addresses DW design problems, with the goal of improving the DW logical design process. Some of the existing work in transformation oriented methodologies for DW design construct the DW starting from an entity-relationship model of the source database, and arrive to a conceptual or high-level-logical dimensional model of the DW. We propose a mechanism for obtaining the DW logical schema through pre-defined transformations applied to the source logical schema, which can be used as a complement to the existing DW design methodologies. The transformations allow a refined logical design of the DW and provide a trace of the design and a mapping between the source and DW logical structures.

Managing source schema evolution in web warehouses

Web Data Warehouses have been introduced to enable the analysis of integrated Web data. One of the main challenges in these systems is to deal with the volatile and dynamic nature of Web sources. In this work we address the effects of adding/removing/changing Web sources and data items to the Data Warehouse (DW) schema. By managing source evolution we mean the automatic propagation of these changes to the DW. The proposed approach is based on a wrapper/mediator architecture, which reduces the impact of Web source changes on the DW schema. This paper presents this architecture and analyses some selected evolution cases in the context of Web DW.

Managing quality properties in a ROLAP environment

In this work we propose, for an environment where multidimensional queries are made over multiple Data Marts, techniques for providing the user with quality information about the retrieved data. This meta-information behaves as an added value over the obtained information or as an additional element to take into account during the proposition of the queries. The quality properties considered are freshness, availability and accuracy. We provide a set of formulas that allow estimating or calculating the values of these properties, for the result of any multidimensional operation of a predefined basic set.

Towards the construction of quality-aware Web Warehouses with BPMN 2.0 Business Processes

A Web Warehouse (WW) is a Data Warehouse which consolidates data from the Web. The goal of these systems is to act as an intermediary between data publication and the user, pre-processing data and adding value to them. This pre-processing involves data integration, data aggregation, data re-structuring and data quality measurement and improvement. A Business Process (BP) model helps us to specify the users, activities, precedence relations between activities and restrictions, that have to be carried out in order to obtain the desired output. In this paper we present a two level BP specification approach for constructing a WW which has two distinctive characteristics: it manages data quality and it is configurable. The first level BP model is focused on helping the user to configure the web data sources and the desired data quality characteristics, the second level BP uses the defined configuration to generate the WW. Quality characteristics are also defined for the intermediate data sources used to populate the WW.

Automating the process of building flexible Web Warehouses with BPM Systems

The process of building Data Warehouses (DW) is well known with well defined stages but at the same time, mostly carried out manually by IT people in conjunction with business people. Web Warehouses (WW) are DW whose data sources are taken from the web. We define a flexible WW, which can be configured accordingly to different domains, through the selection of the web sources and the definition of data processing characteristics. A Business Process Management (BPM) System allows modeling and executing Business Processes (BPs) providing support for the automation of processes. To support the process of building flexible WW we propose a two BPs level: a configuration process to support the selection of web sources and the definition of schemas and mappings, and a feeding process which takes the defined configuration and loads the data into the WW. In this paper we present a proof of concept of both processes, with focus on the configuration process and the defined data.

Applying a Data Quality Model to Experiments in Software Engineering

Data collection and analysis are key artifacts in any software engineering experiment. However, these data might contain errors. We propose a Data Quality model specific to data obtained from software engineering experiments, which provides a framework for analyzing and improving these data. We apply the model to two controlled experiments, which results in the discovery of data quality problems that need to be addressed. We conclude that data quality issues have to be considered before obtaining the experimental results.

A quality aware service-oriented web warehouse platform

In order to be a useful tool, a Web Warehouse (WW) should take into account the quality of the data it manages and the quality of the services that provide the source data. It should also have enough flexibility to endure the high volatility of web sources. In this work we propose a WW platform that satisfies these two conditions. It manages data and services quality, measuring quality at the different stages of the system life-cycle and at the same time, using these measures as input for information processing. Additionally, the platform achieves flexibility and configurability because it is strongly based on information-processing services.

Data Quality Metrics for Genome Wide Association Studies

Genome Wide Association Studies (GWAS) are developed to find direct or indirect relations from given genomic configurations to physical characteristics or specific diseases. In order to build new GWAS, avoiding the complexities of field based studies, a statistical technique called meta-analysis can be used. Bad or unknown data quality has been largely identified as a major problem in meta-analysis since it generates lack of confidence and inhibits its exploitation. This paper addresses GWAS data quality issues and presents a domain specific model for data quality assessment, which has been developed taking into account meta-analysis requirements.

Reliability Models for Data Integration Systems

Data integration systems (DIS) are devoted to providing information by integrating and transforming data extracted from external sources. Examples of DIS are the mediators, data warehouses, federations of databases, and web portals. Data quality is an essential issue in DIS as it concerns the confidence of users in the supplied information. One of the main challenges in this field is to offer rigorous and practical means to evaluate the quality of DIS. In this sense, DIS reliability intends to represent its capability for providing data with a certain level of quality, taking into account not only current quality values but also the changes that may occur in data quality at the external sources. Simulation techniques constitute a non-traditional approach to data quality evaluation, and more specifically for DIS reliability. This chapter presents techniques for DIS reliability evaluation by applying simulation techniques in addition to exact computation models. Simulation enables some important drawbacks of exact techniques to be addressed: the scalability of the reliability computation when the set of data sources grows, and modeling data sources with inter-related (non independent) quality properties.

Data Warehouse Quality Assessment Using Contexts

Data Warehousing Systems DWS are of great relevance for supporting decision making and data analysis. This has been proven over time, through the generalization of its development and use in all kind of organizations. Many researchers have presented the need to incorporate and maintain Data Quality DQ in DWS. However, there is no consensus in the research community on how or whether it is possible to define a set of quality dimensions for DWS, since such set may depend on the purpose for which the data are used. Moreover, quality requirements may vary among different domains and among different users. The contribution of this paper is twofold: a study of existing proposals that relate DQ with DWS and with contexts, and a proposal of a framework for assessing DQ in DWS. This proposal is the starting point of a broader and deeper investigation that will allow quality management in DWS.