Data integration is the process of retrieving data from multiple heterogeneous sources and combining it in such a way that a unified view is obtained. This allows us to compare and manage data as if it came from a single source. This can be a difficult and costly process, but it is necessary in industrial groups that manage different databases, or in big data contexts, in which data from different providers are combined to be analyzed in order to obtain improved information about users, processes, and so on.

One of the possible strategies in data integration is to use a common ontology to guide the integration, providing a common semantics into which all the data is converted. This is an approach that comes from the semantic web area.

In this paper, the authors present the application of this ontology-based integration to fast-fashion design industries. In fact, they present how it has been applied to one company, Imperial Fashion. The paper covers the whole process followed to implement it in this company, from the initial interaction with domain experts in order to obtain their approval and collaboration, including the ontology creation and the steps followed for data conversion, to data visualization. A valuable characteristic of this paper is having a process as complete as this one condensed in a single place. It provides guidance to apply a similar approach, that is, ontology-based integration, in this or other domains.

Section 2 presents the arguments used to convince industry experts, with no technical expertise or knowledge about ontologies or the semantic web, about the suitability of ontologies for their data integration problem (they presented an ontology as a kind of “Esperanto language”). This is accompanied by the requirements of industry experts, in essence not affecting existing databases while the process presented in this paper is in progress. Section 3 details the process followed to develop the global ontology, called imperial data ontology (IDO). It includes the methodology used for ontology development (simplified agile methodology for ontology development (SAMOD)), the applications used (graphical framework for OWL ontologies, Protégé, Fuseki), and the reuse of existing ontologies (something really encouraged in the semantic web community). In addition, the different modules in the IDO ontology are presented in detail, so that for each of them a scenario and competency questions are listed, together with the description of the main classes in that module. After this detailed description of the ontology and the way it was built, section 4 gets into the details of data conversion. Again, the methods and applications used are thoroughly covered: each conversion module is a resource description framework (RDF) graph serialized in Turtle, whose structure is explained by means of an excerpt from one of them. Each of these RDF graphs is used by the general conversion algorithm, which implements the data conversion following the Turtle file’s rules. In section 5, the data visualization and interaction phase is presented. Finally, sections 6 and 7 cover related work and conclusions.

In my opinion, the high level of detail used to present each step of this process is one valuable contribution of this paper, making it interesting for data engineers intending to apply a similar strategy in fast-fashion environments or other industrial contexts. But it is also a positive value from a learning perspective, as it can be used as a guide for learning the phases of data integration with ontologies in real life, and what tools can be used in each step.