Integrating Provenance Capture and UML With UML2PROV: Principles and Experience

Abstract

This block will depict the excerpt of the UML diagram with the elements that model the previous key elements. In addition, we provide a table, whose structure is illustrated below, that explains the representation of each key element, by means of UML elements. Additionally, we assign a green label containing a numeric identifier to each UML element, which makes it easier its location in the UML diagram. Key Element UML RationaleName of the element UML element id The fundamental reasons serving to account for the use of the UML element for modelling the key element.2/74 Mapping to PROVThis block contains the PROV template generated from the previous excerpt of UML Diagram, together with a explanation about the transformation, that is, the PROV elements, attributes, and PROV relations generated from the UML elements in the UML Diagram. We assign a numeric identifier to each PROV element that corresponds to the identifier of the UML element from which it comes from. Additionally, each relation among PROV elements appearing in the PROV template is labeled with a letter that helps link such a relation with its description. The structure used to specify this block is the following:The explanation of the mapping between UML element and PROV element. Attributes PROV ElementAttribute / Value Description PROV element id name of attribute / assigned value The description of the meaning of the attribute and its value.Note: Throughout this specification, we have included the attributes tmpl:startTime and tmpl:endTime associated with Activities because we consider such an information very useful from a provenance point of view. Nevertheless, both attributes are optional and the user is free to include them. PROV relations PROV Relation DescriptionPROV relation id Description of the relation.Note: In PROV, two relationships of the form (B, prov:used, A) and (C, prov:wasGeneratedBy, B) may be enriched with (C, prov:wasDerivedFrom, A) to express the dependency of C on A. This structure is a provenance construction called use-generate-derive triangle [3] which explicitly connects a generated prov:Entity to a used prov:Entity. In the realm of this work, it may be applied in those templates in which a prov:Entity is used by a prov:Activity, and such a prov:Activity generated another prov:Entity. However, aiming at avoiding the overburden of the PROV template explanations with information that can be inferred, we have decided to include the relation prov:wasDerivedFrom only when the context of the pattern explicitly refers to such a derivation. DiscussionIssues related to the transformation of UML to PROV. Concretely, we will focus on the explanation and justification of our transformation decisions together with alternative solutions (if any), and some questions that are likely to come up to the reader. 3/743 Index of patterns Modelled by means of UML Sequence DiagramsPattern identifier Context Page SeqP1A participant (the sender) interacts with another participant (the recipient) by calling an operation in the recipient, and then, ...

Automated and non-intrusive provenance capture with UML2PROV

“…The work in [37] provides quantitative data and qualitative arguments to show the benefits and trade-offs of applying UML2PROV, focusing mainly on how the level of detail of several UML designs of an application affect aspects such as provenance design generation, application instrumentation, maintenance of provenance capabilities, storage and run-time overheads, and quality of the generated provenance. In that evaluation only one of the configurations presented in Sect.…”

“…Here, as well as in [37], a Bioinformatics application called GelJ [18] is used as case study. GelJ is a platform-independent, open-source tool that arose for analyzing DNA fingerprint gel-images [18].…”

“…The proposed BGM design and implementation presented here not only fulfils all the requirements stated for the BGM (Table 1), but it also allows developers to follow different approaches (called configurations) to manage provenance. This approach, briefly presented in [37], is a generic event-based proposal developed on top of the AOP paradigm [21] to generate bindings without user intervention.…”

“…UML2PROV [37] is a more recent software-engineering methodology the authors proposed to facilitate the instrumentation of provenance recording in applications by addressing the gap between application design and provenance design. It automates the generation of (1) PROV templates for the provenance to be recorded in a given application and (2) the code to capture values required to instantiate those templates at run time, both from the application's UML diagrams.…”

“…UML2PROV methodology is the subject of several previous publications [36,37,39], which also include its evaluation as such methodology. However, those publications only partially described the definition of the BGM and provided a succinct description of its design.…”