لخّصلي

Emerging Data Warehousing Technologies
By the beginning of this century, the foundational concepts of data warehouse systems were mature and consolidated. Nevertheless, the field has been steadily growing in many different ways. On the one hand, new kinds of data and data models have been introduced. Some of them have been successfully implemented into commercial and open-source systems. This is the case for spatial data. On the other hand, new architectures are being explored for coping with the massive amount of data that must be processed in modern decision-support systems. We comment on these issues in this section.
A simplifying hypothesis used in most data warehouses is that dimensions do not change, and thus facts and their measures are the only data that are associated with a time frame. However, this does not correspond to reality, since dimensions also evolve in time; for instance, a product may change its price or its category. The most popular approach for solving this problem, in the context of relational databases, is the so-called slowly changing dimensions.
An alternative approach to this problem is based on the notion of Temporal databases, which provide structures and mechanisms for representing and
8
1 Introduction
managing time-varying information. The combination of temporal databases and data warehouses leads to temporal data warehouses.
Current database and data warehouse systems give limited support for manipulating time-varying data. Querying time-varying data with SQL involves writing extremely complex and probably inefficient queries. Further, MDX currently does not provide temporal support. What is needed is to extend the traditional OLAP operators for exploring time-varying data, which is referred to as temporal OLAP (TOLAP). Temporal data warehouses are studied in Chap. 11.
In addition to the above, in real-world scenarios, the schema of a data warehouse evolves across time in order to accommodate new application requirements. The common approach to address this situation consists of modifying the data in the warehouse to comply with the new version of the schema: this implies removing data that are no longer needed and adding new data that were not previously collected. When this is not possible or desirable, the versions of the schema and their data should be maintained, leading to multiversion data warehouses. In such data warehouses, new data are added according to the current schema, while data associated with previous schemas are kept for analysis purposes. Thus, users and applications can continue working with the previous schema versions, while new users and applications can target the current version of the schema. Multiversion data warehouses are studied in Chap. 11.
Over the years, spatial data has been increasingly used in various ar-eas, such as public administration, transportation networks, environmental systems, and public health, among others. Spatial data can represent either objects located on the Earth's surface, such as streets and cities, or geographic phenomena, such as temperature and altitude. The amount of spatial data available is growing considerably due to technological advances in areas such as remote sensing and global navigation satellite systems (GNSS), namely the Global Positioning System (GPS) and the Galileo system.
Spatial databases offer sophisticated capabilities for storing and manipulating spatial data. However, such databases are typically targeted toward daily operations and therefore are not well suited to support the decision- making process. As a consequence, spatial data warehouses emerged as a combination of the spatial database and data warehouse technologies. Spatial data warehouses provide improved data analysis, visualization, and manipu-lation. This kind of analysis is called spatial OLAP (SOLAP), which enables the exploration of spatial data in the same way as in OLAP with tables and charts. We study spatial data warehouses in Chap. 12.
Many applications require the analysis of data about moving objects, that is, objects that change their position in space and time. The possibilities and interest of mobility data analysis have expanded dramatically with the availability of positioning devices. Traffic data, for example, can be captured as a collection of sequences of positioning signals transmitted by the cars' GPS along their itineraries. This kind of analysis is called mobility
1.2 Emerging Data Warehousing Technologies
9
data analysis. In addition, since the sequences generated by moving objects' positions can be very long, they are often processed by being divided into segments of movement called trajectories, which are the unit of interest in the analysis of movement data. Extending data warehouses to cope with mobility data leads to mobility data warehouses. These are studied in Chap. 12.
A common characteristic of the web, transportation networks, communication networks, biological data, and economic data, among others, is that they are highly connected. Since connectedness is naturally modeled by graphs, the interest in graph databases and graph analytics lead to the notion of graph data warehousing and graph OLAP. Two main approaches have been proposed in this respect. On the one hand, the property graph data model is used for native graph databases and graph analytics, where graph data structures composed of nodes and vertices are the basis for storing the data. This approach is very effective for computing path traversals. Chapter 13 is devoted to property graph databases and graph analytics, mainly based on Neodj, one of the most popular graph databases in the marketplace.
The web is an important source of multidimensional information, although this is usually too volatile to be permanently stored. The semantic web aims at representing web content in a machine-processable way. The basic layer of the data representation for the semantic web recommended by the World Wide Web Consortium (W3C) is the Resource Description Framework (RDF), on top of which the Web Ontology Language (OWL) is based. In a semantic web scenario, domain ontologies (defined in RDF or some variant of OWL) define a common terminology for the concepts involved in a particular domain. Semantic annotations are especially useful for describing unstruc-tured, semistructured, and textual data. Many applications attach metadata and semantic annotations to the information they produce (e.g., in medical applications, medical imaging, and laboratory tests). Thus, large repositories of semantically annotated data are currently available, opening new opportunities for enhancing current decision-support systems. The data warehousing technology must be prepared to handle semantic web data. In Chap. 14 we study semantic web data warehouses.
In the current big data scenario, which will be predominant in the coming years, massive-scale data sources are becoming common, posing new challenges to the data warehouse community. New database architectures are gaining momentum. As an answer to these challenges, distributed storage and processing, NoSQL database systems, column-store database systems, and in-memory database systems are part of new emerging data warehouse architectures. In addition, traditional ETL processes and data warehouse solutions are unable to cope with the massive amounts and variety of data. Thelutions are unable to cope with the massive amounts and variety of data. The need to combine structured, unstructured, and real-time analytics demands for solutions that can integrate data analysis in a single system. The NewSQL and HTAP paradigms, Data lakes, Delta Lake, Polyglot architectures, and cloud data warehouses are responses to this demand from academia and in-