LUND UNIVERSITY

PO Box 117221 00 Lund+46 46-222 00 00

It Takes Two to Tango – An Experience Report on Industry–Academia Collaboration

Runeson, Per

Published in:[Host publication title missing]

DOI:10.1109/ICST.2012.190

Published: 2012-01-01

Link to publication

Citation for published version (APA):Runeson, P. (2012). It Takes Two to Tango – An Experience Report on Industry–Academia Collaboration. In[Host publication title missing]. (pp. 872-877). IEEE--Institute of Electrical and Electronics Engineers Inc..10.1109/ICST.2012.190

General rightsCopyright and moral rights for the publications made accessible in the public portal are retained by the authorsand/or other copyright owners and it is a condition of accessing publications that users recognise and abide by thelegal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of privatestudy or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal ?

Take down policyIf you believe that this document breaches copyright please contact us providing details, and we will removeaccess to the work immediately and investigate your claim.

Download date: 16. Jun. 2016

It Takes Two to Tango — An Experience Report on

Industry–Academia Collaboration

Per RunesonLund University

per.runeson@cs.lth.se

Abstract

Industry–academia collaboration is critical forempirical research to exist. However, thereare many obstacles in the collaboration pro-cess. This paper reports on the experiencesgained by the author, in a 2-year collabora-tion project on software testing which involvedon-site work by the researcher in the industrypremises. Based on notes, minutes of meet-ings, and progress reports, the project historyis outlined. The project is analyzed, using col-laboration models as a frame of reference. Weconclude that there must be a balance betweencompany ‘pull’ and academia ‘push’ in the col-laboration Management support is inevitablya key factor to success, while other factors likecross-cultural skills and interfaces towards keyresources also contribute.

1 Introduction

Industry–academia collaboration is a must forempirical research to take place, as the sourceof the empiricism is industry. Like-wise, forindustry to gain from knowledge created inacademia, collaboration must take place. Thepredominant work products from academicresearch are journal and conference papers.These are not designed to be a communica-tion channel between academia and industry,and hence there are needs for a more directcommunication in order to get mutual benefitsfrom the collaboration.

This paper reports on experiences gained

during a 2-year industry–academia project onsoftware testing, where the researcher was lo-cated part-time in the industry premises, work-ing in a role as consultant on strategic issuesrelated to software testing. The objective is toidentify success and failure factors for such col-laboration, to help future improvements. Theproject was funded by a Swedish research fund-ing body, with the aim of “increased personalmobility between sectors, universities, coun-tries and disciplines”1.

The information presented here on theproject is based on personal notes, minutesof meetings, and progress reports. The anal-ysis is based on two collaboration models forindustry–academia research. The report is in-evitably subjective, as it reports the author’sown experiences and not the industry view, anddoes not hence qualify for being called a casestudy [9].

We analyze the sequence of events in theproject in relation to two published models forindustry–academia collaboration, one process-based by Gorschek et al. [5] and one relationalby Sandberg et al. [10]. Having these modelas the theory framework for the analysis helpsreducing the subjectivity of which items are re-ported.

The experience report concludes thatindustry–academia collaboration would ben-efit from having both industry ‘pulling’ andacademia ‘pushing‘ the project. The need formanagement support and champions in the

1http://www.stratresearch.se/en/Ongoing-Research1/Mobility-grants/

Page 1

Problem formulation

Problem/ issue

Industry

Academia

Study state of the art

Candidate solution

Static validation

Validation in

academia

Dynamic validation

Release solution

Figure 1: An activity model for technology-transfer in industry–academia collaboration,from Gorschek et al. [5].

company are reported elsewhere (e.g. [11]),and supported in this analysis. Other factorsinclude adaptability to different cultures, anda preparedness for continuous change.

Section 2 presents related work on the topic,including models for industry–academia col-laboration. In Section 3, the plans for theproject are presented, and in Section 4 the ac-tual project is described. Section 5 analysesthe project outcome, and Section 6 concludesthe paper.

2 Related work

Models for the collaboration between indus-try and academia involve a clear process fromidentified needs to improved industry prac-tices [5], see Fig. 1. Being a good rolemodel for industry–academia collaboration, ex-perience reports show that the collaborationin practice is far from straightforward, lead-ing Sandberg et al. [10] to define more rela-tional models for the collaboration, as shown inFig. 2. This model has two main foci; researchactivity and research result. Each of them havefive aspect, which are interconnected, enablingeach other in a chain of dependencies.

Success factors for collaboration are reportedto be “management engagement” in the re-search activity, and “need orientation” in the

Need orientation Research result

Industry goal alignment

Research activity

Deployment impact

Industry benefit

Innova-tiveness

Management engagement

Network access

Collaborator match

Communi-cation ability

Continuity

Figure 2: A relational model for industry–academia research, from Sandberg et al. [10].Bowed arrows mean “enables”.

choice of research topic, according to Sand-berg et al. [10]. Wohlin et al. [11] corrobo-rate their results in a survey, identifying “com-pany management support” and “champion atthe company” being the top two factors. Theresearcher’s “attitude and social skills” comethird. They surveyed both industry practition-ers and academics, and found industry rank-ing higher the “researcher’s commitment to fo-cus on industry needs”, “support from com-pany management” and a “visible presence inindustry”, while academics ranked higher the“champion’s network within the company” and“buy in and support from industry collabora-tors”.

One factor in the industry–academia collab-oration the credibility of knowledge for soft-ware practitioners. Rainer et al. [8] identifiedlocal opinion-based knowledge be more trustedin industry than empirically-based knowledge,especially if it comes from a remote context,see Table 1. This is a special challenge facingindustry–academia collaboration.

Industry or practice-academia collaborationis important for several fields of research, suchas Chemical Engineering [7], Civil Engineering[2], Pharmaceutical research [1], and Physics[12]. Each field of study seem to have theirspecific challenges. In Software Engineering aspecific characteristic of the industry–academiacollaboration is that it must involve the oper-

Page 2

Table 1: Matrix of the suggested credibilityof knowledge for software practitioners, fromRainer et al. [8].

Source Type of knowledgeof know-ledge

Opinion Empirical

Local 1 (most credi-ble)

2

Remote 3 4 (least credi-ble)

ational departments on the practice side. Itis not sufficient for software engineering re-searchers to collaborate with research depart-ments in industry; collaboration must involvethe real development projects.

3 Plans for The Industry–Academia Project

The project, which is reported here is a 2-year industry–academia project, funded froma program at Swedish Foundation for Strate-gic Research2 with a specific goal of increas-ing the “strategic mobility” between industryand academia. Researchers could get fundingto spend time in industry, or industry peoplecould get funding to spend time in an academicresearch environment.

The industry and academia partners in thisproject are active in a long-term collabora-tion, which has endured in various forms formore than a decade, including part-time em-ployment of researchers in the company. Inparallel with the current project, other collab-oration projects took place; most prevalently a10-year industry excellence center on softwareengineering3, for which the author is director.

The project in this experience report had thefollowing objectives defined in the application:

The main objectives of this strate-gic mobility project is to 1) boost the

2http://www.stratresearch.se/en/3http://ease.cs.lth.se

effects of the new industrial excel-lence center by widening the industry–academia contact network for the cen-ter director, 2) establish a ‘bridge-head’ within the company organiza-tion for empirical research on soft-ware testing, and 3) support transferof knowledge between two of the com-pany’s development sites.

The scope was software testing in a broadsense, with a focus on processes and manage-ment as the intersection of the researcher’scompetence and the industry needs.

An application, following an easy andlightweight process, was sent in in September2008, and the funding was granted in Decem-ber 2008, with a project plan covering 2009and 2010. The application was supported byletters of intent from the industry partner andthe University, to make personnel and otherresources available for the project. It was de-ferred to a later stage to define which depart-ment of the company the work should takeplace in, as well as formal employment roles– whether the researcher should be employedby the company or by the University and theexternal funds be transferred accordingly. Thenegotiations on the industry side was madeby their university relations coordinator, andnot a specific department within the company,which later should turn out as a problem.

The work plan comprised three phases, oneat the regional office of the company, the sec-ond at another site abroad, and an evaluationphase to the end. These were the plans, butthe outcome became different, as described inthe next section.

4 Actual Industry–AcademiaProject

This section is based on personal notes, min-utes of meetings, e-mail correspondence andprogress reports from the project. The descrip-tion is made by the researcher solely, as the in-dustry contact persons were not available, and

Page 3

is hence naturally biassed, despite all attemptsto be as objective as possible.

During Spring of 2009, the university rela-tions coordinator at the company and the re-searcher met and discussed options at whichtest department to find a ‘liasion’ or ‘cham-pion’ to act as the host for the researcher. Thecompany was in a phase of major reorganiza-tion, and change of technical platform, as wellas development process. Consequently, the or-ganization was not very attracted by activi-ties that might require some resources to ini-tiate. Being busy with various other activi-ties, neither the company coordinator nor theresearcher pressed on very hard to get thingsgoing.

In October 2009, a champion was identified,who saw the value in getting access to a re-searcher “for free”, and the project got started.It was decided that the researcher remainedfull-time employed at the University, and theexternal funding be transferred to the Univer-sity. The researcher signed a nondisclosureagreement, got company ID card, access to thecompany intranet and premises.

The industry champion was responsible forthe development of the new test processes, andhad a strategic leadership role in coordinat-ing improvement projects in the field of soft-ware testing. The champion was chairman forweekly meetings with different test managersin the organization: every second week it wasa broader information forum, the other weeka more focused steering group forum for testimprovement projects. The researcher was in-vited to those meetings, which formed the pulseand contact point towards the organization.

The first annual report to the funding agencysummarizes:

The project is finally started, after 10months’ delay. In the current finan-cial situations, with accompanying re-organizations, it has been hard to findan organizational ‘home’ at the hostcompany. [...] Causes of the delayare, in addition to the current focuson short term goals, lack of experience

in collaboration with academia. Theydon’t really see beforehand what anacademic may contribute to industry.Some obstacles are at the formal level:how to register an co-worker in thepersonell-administrative system, whoworks as a consultant, but is not paidfrom the company? The solution isthat the researcher currently is classi-fied as ‘internship’ in the company’sdatabase.

One additional effect of the reorganizationwas the closing-down of the remote site, atwhich the last six months of the project wereplanned to be spent.

The researcher got specific task forces to in-vestigate, but had also to search for his ownprojects to investigate. An example task forcewas to summarize the research contribution towhether code coverage is a good measure oftest progress, or not. One own project identi-fied was the acceptance testing, run at a dif-ferent site, which seemed not to be effective inmaking acceptance decisions. A master thesisproject was launched during Spring of 2010,that got access to data from the acceptancetesting. The master student developed a statis-tical model, utilizing the data more effectivelyto support decisions [6]. The outcome of theproject was presented to the quality depart-ment, and they took some ideas into practicewhen restructuring the acceptance testing pro-cedures.

Another master thesis project on prioritiza-tion of regression test cases, initiated before theproject start, was concluded in May 2010. Itwas later presented as a conference paper [4].

At this point in time, the champion andpoint of contact left his position. His respon-sibilities were shared between two others. Theone running the meeting fora did not have a socentral position in the organization, and con-sequently, the value and attendance rate of themeetings decreased. The other took over theresponsibility for the test strategy, as well as anoverall operational responsibility. He utilizedthe researcher in some discussions, but not as

Page 4

systematically as the first champion did.In the fall of 2010, a PhD student project

was initiated, based on the researcher’s identi-fication of redundant testing being conductedin the different test organizations, working inparallel. The PhD student was given access tothe environment, and conducted a case studyabout the redundancy of testing across differ-ent test organizations. The results were pre-sented to the champion in December 2010, andfound being additional support for an ongo-ing, more lightweight investigation, with simi-lar goals.

The second annual project report summa-rized:

During 2010, the researcher hasspent one day per week in thecompany. On this day, a steer-ing/reference group for test activitieson the company had regular meetings.In this group, the researcher has afunction as external expert, and hasgiven advice and comments from ascientific perspective.

During spring 2011, the PhD student projectwas finalized and additional presentations weregiven. The researcher spent less time in thecompany during Spring 2011, due to other com-mitments and lack of requests from the com-pany. The requests, in turn, were reduced dueto limited presence in the company.

During the Fall of 2011, the researcher spentabroad as a visiting professor. However, sincethe company had closed down its premisesin the host university’s vicinity, the plannedtransfer of experiences could not take place.

5 Evaluation

In this section, the procedures and the outcomeof the project are evaluated. Firstly, it is ana-lyzed with respect to the models presented inSection 1, by Gorschek et al. [5], Sandberg et al.[10], and Rainer et al. [8]. Second, the set-ting up of the project are discussed, thirdly,general organizational aspects are discussed,

and finally improvement for further projectsare identified.

5.1 Collaboration Models

This project is hard to fit into the Gorscheket al. [5] technology transfer model (see Fig. 1),since there was no specific technical goal to ad-dress. Rather, the project goal was to fill acertain function of constituting “a ‘bridgehead’within the company organization for empiricalresearch on software testing”. The specific taskforces had more of the technical goal, and theworking process resembled then the technologytransfer model in the small scale.

The Sandberg et al. [10] model is more suit-able to analyze the outcome of the project (seeFig. 2). Admittingly biassed, due to self re-porting and lack of industry participation inthe analysis, the model items are commentedas instantiated in the project, in Tables 2 and3.

For the research activity part of the model,as outlined in Table 2, the management en-gagement played a dual role. Initially, it waslacking due to business on both sides, but oncein place, the champions demonstrated manage-ment engagement for the project. The weeklymeeting fora in the company were instrumen-tal in providing network access as well as com-munication ability, in terms of personal net-works. Since there were two champions duringthe project, they can be compared with respectto collaborator match. One of them was moreactively ‘pulling’ results from the researcher,creating more mutual benefit, while the othermore broadly asked for contributions from theacademic perspective. Spending only one dayper week reduced the ability to create continu-ity. Much of the time was spent catching upwhat happens in the organization, and the op-portunities for spontaneous meetings also weretoo small.

Regarding the research results part of themodel, as outlined in Table 3, the need ori-entation aspect was fully developed. Perhapsit was even too far stretched? The collabora-tion was based on short term needs, aligned to

Page 5

Table 2: Research activity – model items from to Sandberg et al. [10].

Model item Project instantiation

Management engagement The project did not start until management engagement in the form ofa champion was found.

Network access Meeting forum at the core of communication.Collaborator match The first champion and the researcher made up a good match, with the

champion ‘pulling’ results from the researcher.Communication ability The researcher had the option to communicate through the regular meet-

ing.Continuity Spending one day per week in the company was not enough to create

continuity.

Table 3: Research results – model items from to Sandberg et al. [10].

Model item Project instantiation

Need orientation The project was fully need oriented, which made it change completelycompared to original plans.

Industry goal alignment There was no overall goal, pulled by industry; rather it was academiapush to provide results.

Deployment impact The researcher was not working close enough to practice, to impact ondeployment.

Industry benefit From the researcher point of view, industry long term benefits wereclear, but decisions are made on short term gains.

Innovativeness Solutions were not specifically innovative, but the company ratherneeded many small improvement, than one major innovation.

industry goals, and tended consequently to lackan overall orientation. The ability to impact ondeployment is a matter of closeness to or dis-tance from practice. The champions belongedto top management of the company, and hencetheir focus was more on high-level issues, whichwere distant from the day-to-day testing prac-tice. The industry benefits should be evaluatedby industry representatives, but the researchertried to make sure that the activities were ben-eficial for the company. The long-term goalsof the collaboration were clear, but sometimesit was hard to connect the to the short-termtrade-offs between costs and gains. Finally,the solutions worked on were not very radi-

cal, and hence not very innovative either. Themajor ground-breaking results, envisioned byresearch, are rarely what benefits industry inthe short term, but rather a sequence of smallimprovement actions.

In terms of the ‘buy-in’ model by Raineret al. [8], the knowledge presented to the com-pany was mostly remote empirical knowledge,i.e. in the least credible category (see Table 1).However, begin a researcher, present in person,and tailoring the presentation of the knowledgeto the local context, it seemed to be appreci-ated as ‘localized’. If the knowledge could beconnect to the practitioners own experiences, italso seemed to gain credibility. However, this

Page 6

is only subjectively experienced appreciation ofthe knowledge, not investigated by any inde-pendent actor.

5.2 Setting Up the Project

The biggest hurdles in setting up the project,was finding a host (liaison or champion) to con-nect the research project to. This factor isidentified among the top key factor in the sur-vey by Wohlin et al. [11] too. In the currentproject, a university professor was offered “forfree”; still managers did not see the benefits be-forehand of such a project. Perhaps, researchprojects should not be “for free”; rather somekind of company commitment or funding maylead to more ‘pull’ from the company.

Collaborating with the organization throughan established series of meetings, was veryhelpful. The meetings as such gave a natu-ral introduction to the company and its on-going test improvement projects, and to thekey staff at various positions. Further, usingshares of the time of an already gathered meet-ing is much easier than to gather them to aseparate meeting. Having said this, it shall benoted most people willingly made themselvesavailable for specific meetings if the project soneeded. Some people also indicated a curiosityabout what “research tells” on the topic underdiscussion.

Getting permissions from the developmentorganization was rather straightforward, oncethe right organizational home was found.However, supporting organization, like hu-man resources and legal departments were notadapted to this type of collaboration. The hu-man resources department had problem to finda suitable job title for the researcher, as men-tioned above, which is more of a curious than aserious issue. However, more serious was thatthe legal department was not up to date withthe Swedish law about university–industry col-laboration. Secrecy issues about informationhanded over in a university–industry collab-oration is strongly protected by the SecrecyAct (2009:400). However, the legal departmentenforced signing a non-disclosure agreement,

which is not at all valid, since law outperformscontract. However, instead of fighting the de-bate, the researcher signed the agreement andabided with the law.

5.3 Business Models and Organiza-tions

In an industry–academia collaboration project,it soon becomes clear that the business modelsand organizations are different [3].

Firstly, for industry, the driving force is todevelop new products on time to put on themarket and earn money. For academia, thedriving force is to have an impact on the aca-demic scene through publications, and for anapplied researcher, to have an impact on in-dustry practice. In the current project, thisdistinction was well known and respected, al-though it sometimes was questioned what theresearcher would gain from spending time inthe industrial context. Once explained, the re-searcher’s presence was fully accepted and ap-preciated.

The incentives in the two types of organi-zations, are also different. Appreciation in in-dustry – and sometimes bonus – comes withfulfillment of short to medium term projectgoals, while the incentives in academia is re-lated to publications and citations – with year-long feed-back cycles – which in the long runmay help career promotion and grants for newresearch. Naturally, this controls priorities inindustry towards near-sighted goals, at the costof long-term perspectives, and vice versa foracademia.

This difference in time scale is also re-flected in the planning horizon in personal cal-endars. Researchers make commitments farahead of time for e.g. conference organizationand teaching, while industry staff re-plan theircommitments on daily, or even hourly basis, forhigher management. Spending only one dayper week in a “high-speed” organization makesit very hard to keep up the pace, and practi-cally challenging to negotiate schedules.

Industry organizations change frequently,while academic organizations are very stable.

Page 7

In this project, re-organizations lead to changein contact person and focus of the project,which lead to re-establishment of network andwork procedures.

Last, but not least, is shall be noted thatthe target organizations for the collaborationproject are fully operational product devel-opment organizations, working at the criticaltime-line of product deliveries. Industrial re-search organizations naturally have more simi-larities to the academic organization; however,the do not offer an environment to study andimpact on the software engineering practices,possibly on the software technology only.

5.4 Synching the Tango – Discussion

How could industry–academia collaboration bein better synch, and thus more effective? Manyof the factors impacting on this project are dueto “bad luck”, e.g. coinciding with the financialcrisis. Other factors are independent of suchevents.

This project was not initiated by the com-pany; it suffered from lack of ‘pull’ from thehost organization and was mostly ‘pushed’ bythe researcher. Ideally, the demand shouldcome from inside the organization, driven bya commitment to and expectations on the col-laboration. Means to achieve this may involvedirect costs for the collaboration – if an orga-nization pays the costs from their own budget,they feel obliged to utilize the results – or in-centives for the managers and the organizationrelated to their business goals.

The stress for management support can notbe made clear enough. As the collaborationinteracts with short-term projects, while theresearch goals are more long-term, this conflictmust be resolved by company management atall levels. This observation corroborates earlierexperience [10] and survey results [11]. The fre-quency of industry reorganizations can proba-bly not be impacted on, but the hurt causedby them may be reduced by anchoring the col-laboration high enough in the management hi-erarchy.

The timing aspect includes both the share oftime spent in industry, and the industry ver-sus academia planning horizons. To improvethe collaboration, a larger share of the workweek should be spent in industry, which on theother hand, conflicts with the long term com-mitments of the researcher. PhD students arefrom that perspective better candidates for col-laboration, since their schedules and commit-ments tend to be more flexible, but a student’slack of experience reduces the strategic valuefor the industry. Industrial PhD students (i.e.industry employees, conducting PhD studies inparallel) are an attractive option, although it isa demanding work situation for the individual.

Being prepared for change is a recommendedstrategy for researchers. The pace of changeis much higher in industry than in academia,and is driven by other forces than drivingthe industry–academia collaboration. Hence,strategies for change and risk mitigation, e.g.for thesis projects, must be part of the aca-demic side of a collaboration project.

For broad software engineering improvementprojects, the access to an existing meeting fo-rum of key people was valuable. Getting toknow formal and informal organizational struc-tures take time, and if that time can be spenton the improvement project itself, it becomesmore efficient.

6 Conclusion

Was the collaboration project a success or afailure? With respect to its original plans,it was definitely a failure. The start was 10months delayed, the transfer part was canceledsince the other premises were closed, and theproject had to restart with a new champion,half-way. However, with respects to contactsfor student projects and performed academicstudies, the project still fulfilled some of itsgoals. Ironically, the academic gains are morevisible than the company gains.

For more mutual benefit of the collabora-tion, a balance between company ‘pull’ andacademia ‘push’ must be achieved. Manage-

Page 8

ment support is inevitably a key factor to suc-cess. However, the ability by the researcherto adjust to different cultures, including timeperspective is another key to success, as isthe access to the organizational key resourcesthrough, for example, meeting fora.

Acknowledgement

This work was supported by the SwedishFoundation for Strategic Research under grantSM08-0040. The PhD student work wasconducted within the SWELL (Swedish Ver-ification & Validation Excellence) researchschool, supported by the Swedish Govern-mental Agency for Innovation Systems (VIN-NOVA). The author also thanks the companyrepresentatives for their support. The paperwas written during the author’s sabbatical atNorth Carolina State University, USA.

References

[1] T. Agres. Unlocking hESC success.Drug Discovery & Development, 13(4):6–7, 2010. ISSN 1524783X.

[2] M. Arif, C. Egbu, M. Alshawi, S. Srini-vas, and M. Tariq. Promoting greenconstruction in India through industry-academia collaboration. Journal ofProfessional Issues in Engineering andEducational Practice, 136(3):4 pages,2010. doi: doi:10.1061/(ASCE)EI.1943-5541.0000019.

[3] T. Bjerregaard. Industry and academia inconvergence: Micro-institutional dimen-sions of R&D collaboration. Technovation,30(2):100–108, 2010. ISSN 0166-4972. doi:10.1016/j.technovation.2009.11.002.

[4] E. Engstrom, P. Runeson, and A. Ljung.Improving regression testing transparencyand efficiency with history based prioriti-zation – an industrial case study. In Pro-ceedings of the 4th International Confer-

ence on Software Testing Verification andValidation. IEEE Computer Society, 2011.

[5] T. Gorschek, P. Garre, S. Larsson, andC. Wohlin. A model for technology trans-fer in practice. IEEE Software, 23(6):88–95, 2006. ISSN 0740-7459. doi:10.1109/MS.2006.147.

[6] L. Gustafsson. Software reliability model-ing for test stopping decisions – binomialapproaches. Master thesis, Dept. of Com-puter Science, Lund University, Septem-ber 2010.

[7] A. S. Mujumdar. Editorial on industry–academia collaboration in R&D. Dry-ing Technology, 28(4):431–432, 2010. doi:10.1080/07373931003609427.

[8] A. Rainer, T. Hall, and N. Baddoo. Per-suading developers to ’buy into’ softwareprocess improvement: Local opinion andempirical evidence. In International Sym-posium on Empirical Software Engineer-ing, pages 326–335, 2003.

[9] P. Runeson, M. Host, A. W. Rainer, andB. Regnell. Case Study Research in Soft-ware Engineering. Guidelines and Exam-ples. Wiley, 2012.

[10] A. Sandberg, L. Pareto, and T. Arts. Ag-ile collaborative research: Action prin-ciples for industry-academia collabora-tion. Software, IEEE, 28(4):74–83, july-aug. 2011. ISSN 0740-7459. doi:10.1109/MS.2011.49.

[11] C. Wohlin, A. Aurum, L. Angelis,L. Phillips, Y. Dittrich, T. Gorschek,H. Grahn, K. Henningsson, S. Kagstrom,G. Low, P. Rovegard, P. Tomaszewski,C. van Toorn, and J. Winter. Success fac-tors powering industry-academia collabo-ration in software research. IEEE Soft-ware, PP(99):1, 2011. ISSN 0740-7459.doi: 10.1109/MS.2011.92.

Page 9

[12] I. Yamada, J. Matsuo, and N. Toyoda.Summary of industry-academia collabo-ration projects on cluster ion beam pro-cess technology. AIP Conference Pro-ceedings, 1066(1):415–418, 2008. doi:10.1063/1.3033651.

Page 10