Sunteți pe pagina 1din 2

2017 IEEE International Conference on Big Data (BIGDATA)

Towards a Big Data Requirements Engineering Artefact Model


in the Context of Big Data Software Development Projects
Poster Extended Abstract

Darlan Arruda and Nazim H. Madhavji


Department of Computer Science
University of Western Ontario
London, Canada
darruda3@uwo.ca, madhavji@gmail.com

Abstract — In this paper, we describe our ongoing research artefact-model [1]: (i) identification of elements and
aimed at defining a Requirements Engineering Artefact Model concepts. A Systematic Literature Review (SLR) was
(REAM) in the context of Big Data software applications. This conducted on the topic related to RE artefact models from
model aims to provide a “big picture” of the Requirements which artefacts were identified and a glossary of terms was
Engineering work products created and used in Big Data
created (see Table I); (ii) definition of the artefact
software development projects. REAM are important tools
that can be used as references for the definition of domain- relationships. Using the glossary of terms, we created t a
specific RE models, system life-cycle processes and artefact- table of artefacts and their inter-relationships (see Table II);
centered processes, currently bereft in the Big Data Software (iii) definition of cardinalities. Using the basic table of
Engineering research. artefacts and inter-relationships and interpreting the domain
knowledge from the scientific literature, a relationship
Keywords - Big Data; Requirements Engineering; Artefact cardinality document was created; (iv) synthesizing the
Model; Big Data Requirements Engineering Artfect Model. model using fragments of artefacts, their inter-relationships,
and cardinality information, they were inter-connected
I. INTRODUCTION iteratively as made sense in the RE domain, eventually
Based on literature analysis, it is clear that the field of Big resulting in the artefact model shown in Figure 1.
Data is growing in industry. However, as yet there is no The proposed model is composed of three basic elements:
recognisable body of knowledge concerning the development (a) Artefact: a rectangular shape; (b) Association: a labelled
of Big Data software applications. This void is also reflected line connecting two artefacts; and (c) Cardinality: indicating
in the field of Requirements Engineering (RE). For example, quantity. If the cardinality is not expressed in the association
a search for the state of the art literature in RE involving Big line, it means that it is unity cardinality.
Data applications initially yielded 311 papers which, upon The following relationships are represented in the model:
closer scrutiny, were filtered down to only 13 papers that (i) Is-derived-from represents the relationship between the
actually addressed some aspect of RE involving Big Data artefacts when from one artefact (e.g., Big Data scenarios)
software applications. one or more artefacts can be derived and specified (e.g.,
In an attempt to add to the RE body of knowledge on end- quality requirements are derived from Big Data scenarios);
user applications on Big Data, we mined the filtered (ii) Is-identified-from represents the relationships when from
literature for RE activities, work products, constraints, one artefact (e.g., organisational goals) one or more artefacts
methods, and related elements and relationships involved in (e.g., Big Data Scenarios, Constraints and Concerns, etc.) are
the development of end-user Big Data software applications identified; (iii) Is-part-of relationship represents aggregation
in order to create an artefact-model. An artefact-model, with and it is illustrated when one or more artefacts are part of one
domain elements and inter-relationships, is a means for or more major artefacts (e.g., functional requirement is part
understanding the domain and the context for creating a of software requirements); (iv) Contains relationship is used
software application [1], [2]. The model details are also when one or more artefacts have or hold information from
promoted as an aid in product maintenance and in another artefact within (e.g., software requirements contains
establishing project-specific RE processes [1]. analysed requirements); and (v) Used in relationship means
that one artefact can be used to guide in the definition of
II. RE ARTEFACT MODEL CREATION other artefacts (e.g., project constraints are used in Big Data
We adhered to the following key steps in the definition of the scenarios).

978-1-5386-2715-0/17/$31.00 ©2017 IEEE 4725


Legend: ADK - Application Domain-knowledge | Cons - Constraints | Association line | Aggregation Line | Artefact
Figure 1. Big Data Requirements Engineering Artefact Model (BD-REAM) Depicting its Elements, Relationships, and Cardinalities.

III. SUMMARY AND FUTURE WORK of RE life-cycle processes in the context of Big Data
Software engineering of Big Data end-user applications is software application projects.
relatively new. Currently, to our knowledge, there is no
ACKNOWLEDGMENT
domain or artefact model for Requirements Engineering
(RE). This paper thus aims to fill that gap though the artefact This research is in part supported by grants from CNPq, The
model depicted in Figure 1 is still preliminary and pre- National Council of Technological and Scientific
empirical-validation. The model has been created through Development – Brazil and NSERC, Natural Science and
systematic literature review (SLR). Engineering Research Council of Canada. We thank the
One observation from the SLR results is that there isn’t reviewers for providing helpful feedback that has improved
much research addressing the RE aspects of Big Data. While this paper.
this necessarily limits the richness of the artefact model we
have created, the resultant model does provide a preliminary REFERENCES
baseline to the research community for further research in [1] Berenbach,B.D. J. Paulish, J. Kazmeier, and A. Rudorferx. Software
this field and to the practicing community an understanding & Systems Requirements Engineering: In Practice. New York,
McGraw-Hill, 2009, pp.19-38.
of the key artefacts and relationships in the model which can
be used in the design of RE projects. [2] B. Penzenstadler, D. M. Fernandez, and J. Eckhardt, “Understanding
the impact of artefact-based RE - Design of a replication study,” Int.
Further work is envisaged: (i) validation and improvement Symp. Empir. Softw. Eng. Meas., October, pp. 267–270, 2013.
of the model through empirical studies; and (ii) the definition

4726