Content uploaded by Gustavo Pinto
Author content
All content in this area was uploaded by Gustavo Pinto on Jan 15, 2018
Content may be subject to copyright.
Towards a Model to Transfer Knowledge from Software
Engineering Research to Practice
*†Bruno Cartaxo, ‡Gustavo Pinto, *⌥Sergio Soares
*UFPE, †IFPE, ‡UFPA, ⌥SENAI - Brazil
Abstract
Context: Many researchers argue that Software Engineering (SE) research
lacks connection with practice.
Objective: We propose a model aimed at supporting researchers to transfer
knowledge to SE practice.
Method: This model is built upon the foundation of Rapid Reviews and
Evidence Briefings. These two key elements have been proven e↵ective in
other domains, such as medicine, and initial results suggest that they can
play a prominent role in SE as well.
Results: We discuss how to appl y t h e mode l a s well a s p ossib l e chal l e n ges
that might hinder its adoption.
Conclusion: We believe that both SE practitioners and researchers could
benefit from the proposed model. We expect replications and instantiations
of the model conducted in the future.
Keywords: Knowledge Transfer, Rapid Reviews, Evidence Briefings,
Evidence Based Software Engineering
1. Introduction
Over the last years, a myriad of Software Engineering (SE) empirical
studies have been conducted on a steady pace. Such advances made Evi-
dence Based Software Engineering (EBSE) one of the pillars of the software
engineering research. However, developers still discount empirical evidence
in favor of expert opinion [1]. Although expert opinion is important, its
prevalence is a challenge, since developers might lack evidence to back up
their claims.
Preprint submitted to Information and Software Technology October 26, 2017
In order to favor evidence rather than opinion, EBSE community has
recently recognized the importance of proper ways to transfer knowledge to
practice [2]. In particular, EBSE researchers advocate in favor of System-
atic Reviews (SR), which synthesizes the best research evidence and make it
available to practitioners and researchers [3]. Unfortunately, recent studies
revealed a lack of connection between what is studied in the Systematic Re-
views and what is needed by software engineering practice [3], which hinders
the knowledge transfer process.
In this paper, we argue that evidence from SE researchers should benefit,
and better transferred to, SE practitioners. To bridge this gap, this paper
proposes a knowledge transfer model applied to SE field. The model is an
instantiation of a generic knowledge transfer model proposed by Bozeman [4].
The proposed model is built upon the foundations of two emergent con-
cepts in Evidence Based Medicine, namely: Rapid Reviews [5] and Evidence
Briefings [6]. The former — Rapid Review — is a kind of lightweight sec-
ondary study (details at Section 4). It focus on (1) understanding the prob-
lems that practitioners face in practice and (2) delivering evidence in a timely
manner. The latter concept — Evidence Briefing — is an one-page document
that summarizes the main findings of any empirical study. Its short form is
crucial to provide evidence in a more straightforward and appealing manner
to practitioners (details at Section 5).
2. Related Work
Gorschek et al. [7] developed a model to transfer technology to practice.
Although similar, their model is focused on identifying practitioners’ issues
and propose a direct solution, while our model intends to identify practition-
ers’ issues and synthesize empirical evidence that could support decision-
making. The former approach demands much more e↵ort and commitment
from researchers, which might hinder cooperation between research and prac-
tice. Our proposed model, nevertheless, demands less e↵ort and commitment,
as well as stimulate practitioners to consume empirical evidence, which can
foster a culture of informed decision-making.
As regarding the kinds of mediums, Grigoleit et al. [2] reported that such
mediums can be “artifacts”, like publications and documents, or “human-
intensive”, like conferences and workshops. To make SE research more rele-
vant to practice, Beecham et al. [8] stated the importance of writing shorter
evidence-based reports. However, to the best of our knowledge, this is the
2
first work proposing a model to transfer knowledge strongly based on the
evidence produced by a Rapid Review process and presenting the results
through alternative mediums like Evidence Briefings.
3. The Model
The knowledge transfer model is an instantiation of the influential model
proposed by Bozeman [4]. It has five key elements:
•Transfer Agent: Institution or organization seeking to transfer knowl-
edge. Government agency, university, or a private firm;
•Transfer Medium: Vehicle, formal or informal, which the knowledge
is transferred through. License, copyright, person-to-person, or a for-
mal literature;
•Transfer Object: Content and form of what is transferred; the trans-
fer entity. Scientific knowledge, technological device, process, or know-
how;
•Transfer Recipient: Organization or institution receiving the transfer
object. Firm, agency, organization, consumer, or an informal group;
•Demand Environment: The characteristics of the environment the
Recipient Agent is immersed in.
In our instantiation, software development companies (Demand En-
vironment) are the sources used to identify the issues that practitioners
face. Once an issue is detected, it motivates researchers (Transfer Agents)
to conduct a Rapid Review along with practitioners to identify evidence
(Transfer Object) that could help practitioners (Transfer Recipient)to
address the issue. The evidence is presented to practitioners through Evi-
dence Briefings (Transfer Medium). Figure 1 presents the model instanti-
ated.
Since Demand Environment, Transfer Agent, and Transfer Recipient are
straightforward to understand, at this moment, we focus on two elements
that deserve more attention: the Transfer object and the Transfer medium.
We operationalize the latter in terms of Evidence Briefings [9] and the former
in terms of Rapid Reviews [10].
3
Figure 1: Proposed knowledge transfer model.
4. Rapid Reviews’ Evidence: The transfer object
Rapid Reviews are lightweight secondary studies focused on delivering
evidence in a timely manner to practitioners. Due to its strong focus to
practice, researchers should (1) work in close collaboration with practitioners
and (2) reports the results through alternative mediums more suitable for
practitioners’ needs.
In spite of the limitations inherently natural of this relaxed method, the
interest in rapid reviews is growing in health-care fields [5]. To illustrate,
the prestigious Systematic Reviews1journal published an editorial2recog-
nizing rapid reviews as one of the foundations of Evidence Based Practice.
Additionally, Cochrane — a global renowned group of researchers and practi-
tioners specialized in evidence di↵usion in health-care — announced a group3
to guide the production of rapid reviews in medicine.
Following the promising results in medicine, we conducted a rapid review
in a SE context [10]. The rapid review was aimed to (1) understand the prob-
lem that the company had and to (2) provide evidence that could support
the decision-making of company’s representatives towards fixing the prob-
lem. The problem reported was low costumer collaboration.Wethen
(1) created a search string to search for relevant literature (limited to Scopus
search engine only); (2) conducted a selection procedure to filter out papers,
screened by only one researcher; (3) synthesized the findings, and finally
(4) reported the results to practitioners using Evidence Briefings. Through-
out this process, practitioners worked on close collaboration with researchers.
1Systematic Reviews’ Journal – http://systematicreviewsjournal.biomedcentral.com
2Systematic Reviews Journal’s Editorial http://bit.ly/2boZTgv
3Cochrane Rapid Review Group - http://bit.ly/2u77FUN
4
The entire process took six days, and the first author was full-time dedicated.
Although the company’s representatives were unfamiliar with this approach,
they considered that the rapid review was applicable to software engi-
neering practice, specially due to its short duration. They reported that
rapid review is more reliable than the approach they use to seek in-
formation (e.g., informal sources and expert opinion). Additionally, they
mentioned that rapid review process helped them to better comprehend
and structure the problem they were facing. These initial results suggest
that rapid reviews might play the role of Transfer Object in SE.
5. Evidence Briefings: The transfer medium
Researchers in medicine argue that systematic reviews often neglect prac-
titioners’ needs [5], avoiding them to consume that kind of content. This led
to alternative mediums to transfer knowledge that better fit practitioners’
needs rather than traditional research paper format.
Following initial findings from medicine peers [5], we recently introduced
the concept of Evidence Briefings [9] in SE. The Evidence Briefings’ tem-
plate is based on information design best practices. Both the template and its
guidelines can be found at http://cin.ufpe.br/eseg/evidence-briefings.
Figure 2 shows an Evidence Briefing. It has six main parts: (1) title; (2)
a short paragraph presenting briefings’ goal; (3) the main section, presenting
the findings of the study; (4) an informative box outlining the intended au-
dience and explaining the nature of the briefings’ content; (5) references to
primary studies; and (6) an area for logos of research groups, universities, or
companies involved in the study [9].
Both format and content of Evidence Briefings were positively evaluated
with a group of SE practitioners [9]. Respondents perceived that information
was easy to find and the format was clear, understandable, and reliable.
Due to these characteristics, we believe that Evidence Briefings are a proper
Transfer Medium.
6. Discussion
The proposed knowledge transfer model, or parts of it, can be applied
in many contexts. For instance, a Rapid Review could synthesize challenges
and strategies to establish agile practices in a distributed team. Similarly,
researchers studying acceptance testing could conduct a rapid review with
5
Figure 2: Evidence Briefing Structure.
testing teams facing similar issues. Regulatory agencies and/or companies
departments that need to define its software improvement process can invite
researchers to conduct rapid reviews in situ. Still, Evidence Briefings can
be used to guide technical discussions inside companies, or even to serve as
internal dissemination mediums.
However, some challenges might hinder the adoption of the proposed
model. For instance, some companies may oppose to conduct rapid reviews
arguing difficulties to allocate employees’ e↵ort on that kind of initiative.
To overc o m e t his challenge, re s e a rch ers can argue t h a t the be n e fi ts of rapid
reviews overcome the e↵orts on the long run [5]. Likewise, researchers can
take most of the e↵ort, as they already do with traditional systematic reviews.
Moreover, companies may claim that the model might introduce delays on
the projects’ schedules. One possible mitigation plan is to avoid problems on
the critical path of a project’s schedule (at least until practitioners are not
fully convinced about the benefits of such kind of approach).
Some companies may want to avoid information disclosure or even may
be afraid of admit their problems. To mitigate, researchers might take
6
advantage of Informed Consent Forms (or any similar approach that guaran-
tees data confidentiality, participants anonymity, and the right that partic-
ipants have to withdraw from the research at any moment). For skeptical
practitioners, researchers can also highlight that rapid reviews can provide
evidence to support decision-making based on data gathered from previous
experiences. Therefore, before conducting an internal rapid review, com-
pany’s representatives can evaluate its e↵ectiveness by learning from other
companies’ experiences.
7. Conclusions
In this paper we propose a model to transfer knowledge from scientific ev-
idence to SE practice based on Rapid Reviews and Evidence Briefings. Rapid
reviews are a kind of secondary study that deliberately omit or simplify some
systematic reviews’ steps to deliver evidence in a timely manner and, more
importantly, connected to practitioners’ issues. Evidence Briefings synthe-
size any research in an one-page document, that practitioners could easily get
acquainted with. We report two case studies on the usage of Rapid Reviews
and Evidence Briefings. Due to encouraging initial results, we believe that
our model (or part of it) could be better explored in software engineering
research.
References
[1] A. Rainer, et al., Persuading developers to ’buy into’ software process
improvement: Local opinion and empirical evidence, in: ISESE, 2003.
[2] F. Grigoleit, et al., In quest for proper mediums for technology transfer
in software engineering, in: ESEM, 2015.
[3] R. E. S. Santos, F. Q. B. d. Silva, Motivation to perform systematic
reviews and their impact on software engineering practice, in: ESEM,
2013.
[4] B. Bozeman, Technology transfer and public policy: a review of research
and theory, Research Policy 29 (45) (2000) 627 – 655.
[5] A. C. Tricco, et al., A scoping review of rapid review methods, BMC
Medicine.
7
[6] D. Chambers, P. Wilson, A framework for production of systematic
review based briefings to support evidence-informed decision-making,
Systematic Reviews Journal.
[7] T. Gorschek, P. Garre, S. Larsson, C. Wohlin, A model for technology
transfer in practice, IEEE Software.
[8] S. Beecham, P. O’Leary, S. Baker, I. Richardson, J. Noll, Making soft-
ware engineering research relevant, Computer 47 (4) (2014) 80–83.
[9] B. Cartaxo, et al., Evidence briefings: Towards a medium to transfer
knowledge from systematic reviews to practitioners, in: ESEM, 2016.
[10] B. Cartaxo, G. Pinto, S. Soares, Supporting decision-making with rapid
reviews in software engineering practice, unpublished manuscript (Oct.
2017).
8
