choosing a knowledge dissemination approach

Knowledge and Process ManagementVolume 19 Number 3 pp 160–170 (2012)Published online in Wiley Online Library(www.wileyonlinelibrary.com) DOI: 10.1002/kpm.1391

■ Research Article

Choosing a Knowledge DisseminationApproach†

John Kingston*

Health and Safety Laboratory, Buxton, UK

*CorBuxtE-ma†Its care tHSE

Cop

Knowledge management has been described as ‘getting the right knowledge to the right people in the right place atthe right time’. Knowledge dissemination is a crucial part of knowledge management because it ensures knowledge isavailable to those who need it.This paper reviews four well-known knowledge dissemination techniques. Each technique is classified according to arecently proposed classification scheme, and advice is given regarding when it is appropriate to use each technique.Copyright © 2012 John Wiley & Sons, Ltd.

INTRODUCTION

Knowledge management (KM) has been defined as ‘aconscious strategy for moving the right knowledge tothe right people at the right time, to . . . improveorganizational performance’ (O’Dell and Grayson,1998). Knowledge dissemination—distributing knowl-edge to those who may need it—is therefore a crucialpart of KM.

Awide range of approaches to knowledge dissem-ination is available. It is important that organisationschoose an approach that is appropriate for the knowl-edge being disseminated and for the organisation’sstructure, culture and business goals.

This paper begins by describing a recently pub-lishedmethod that can be used to classify knowledgedissemination techniques. It then describes fourknowledge dissemination techniques and when theyshould be used, including real life examples fromthe domain of health and safety. It concludes bysummarising the pros and cons of the four categorieswithin the chosen classification scheme.

CLASSIFYING KNOWLEDGEDISSEMINATION TECHNIQUES

Milton (2010) presents a four-way classification ofsystems for capturing and disseminating lessonslearnt from past experience. It is presented here

respondence to: John Kingston, Health and Safety Laboratory,on SK17 9JN, UK.il: [email protected], including any opinions and/or conclusions expressed,hose of the authors alone and do not necessarily reflectpolicy.

yright © 2012 John Wiley & Sons, Ltd.

because it can also be applied to classify approachesfor knowledge dissemination.Milton’s classification uses two dimensions

(Figure 1). The first dimension distinguishes ‘collect’approaches (in which knowledge is recorded orwritten into a repository) and ‘connect’ approaches(in which knowledge is communicated directlybetween individuals, whether verbally or throughwritten messages). The second dimension distin-guishes ‘formal’ and ‘informal’ approaches. By‘formal’, Milton means ‘operating within a definedframework or set of rules’, whereas ‘informal’means‘unmanaged and bottom–up’ (Milton, 2009). In prac-tical terms, this often means that ‘formal’ techniquessupply knowledge to users in a structured format,whereas informal techniques supply knowledge inconversational text.This categorisation is powerful because it reflects

one of the biggest philosophical debates in KMresearch. This is the epistemological debate between‘cognitivists’ and ‘constructivists’. According toHeylighen (2003), cognitivists believe that

. . . knowledge consists of models that attempt torepresent the environment in such a way as tomaximally simplify problem-solving. It is assumedthat no model can ever hope to capture all relevantinformation, and even if such a complete modelwould exist, it would be too complicated to use inany practical way. [. . .]. The model that is to bechosen depends on the problems that are to besolved. The basic criterion is that the model shouldproduce correct (or approximate) predictions(which may be tested) or problem-solutions, andbe as simple as possible.

Connect

Collect

Informal Formal

Social networking

Virtual teams, formal

networks

Wikis,blogs

Lessons databases

Figure 1 Milton’s classification of ‘lessons learnt’ approaches

Choosing a Knowledge Dissemination Approach 161

Whereas the constructivist view is that

(i) Knowledge is not passively received eitherthrough the senses or by way of communicationbut is actively built up by the cognising subject.

(ii) The function of cognition is adaptive and servesthe subject’s organisation of the experientialworld, not the discovery of an objective onto-logical reality.

A phrase that is frequently used to summarise theconstructivist view is that knowledge is a ‘justifiedpersonal belief’ (Alavi and Leidner, 1999).

The practical implications of these views are thatconstructivists see knowledge as inseparable froman individual’s mental model(s); this makes themfavour ‘connect’ approaches to KM, as it seems tothem that anything ‘collected’ into a repositorycannot be knowledge. The justification for this is thata repository is not animate and is therefore notcapable of holding a belief (Goodwin, 2009); soanything in a repository must be information ratherthan knowledge. Constructivists also often preferinformal approaches, as there is no guarantee thatthe framework that accompanies any formalapproach will match well with a learner’s mentalmodels. Cognitivists, however, aim to build modelsthat represent the knowledge in someone’s head butare separate from it (for a detailed methodologybased on this approach, see Schreiber et al., 2000);they must therefore use ‘collect’ approaches. Theyalso often prefer formal approaches, either becausethere are rules that describe how their models canbe built or because they are aiming to build a modelwhose structure reflects the expert’s mental model(s).Therefore, it should be expected that there will bemore proposed techniques (or at least, moreenthusiasm for techniques) that fall into the ‘informalconnect’ and ‘formal collect’ categories than in theother two categories.

Another KM debate that is informed by Milton’sclassification is the debate over whether tacit orexplicit knowledge predominates in experts’ headsand what effect that should have on KM techniques.Key arguments include the following:

Copyright © 2012 John Wiley & Sons, Ltd.

• Dave Snowden, a well-respected KM practitioner,argues that because there are multiple factors inour environment and in our experience thatinfluence us in ways we can never understand,we do not know what we know until we needto know it (Snowden and Boone, 2007). In otherwords, he argues that much of our knowledge istacit until the context when it is needed arises.This is often taken to imply that ‘collect’approaches to KM will be ineffective, as theseapproaches necessarily abstract knowledge awayfrom its context.

• There are some situations where the context itselfforms part of the necessary knowledge, but eventhe experts are unaware of this; perhaps, thebest known involved the original design of aspecialised laser, inwhich the laser only functionedif a certain wire was below a certain length—noone explicitly ‘knew’ this, but scientists who hadviewed the original working set-up of the laserwere able to recreate it and produce a workinglaser, whereas scientists who merely followed thecircuit diagrams often could not do so (Collinsand Harrison, 1975). In such situations, it is clearthat the ‘collected’ knowledge is incomplete.

• The well-known ‘knowledge creating cycle’proposed by Nonaka and Takeuchi (Nonaka andTakeuchi, 1995) suggests that innovation occursthrough a process that involves both tacit andexplicit knowledge, but that the ‘combination’step that creates new knowledge occurs whenknowledge is tacit. This implies that ‘collect’approaches will never fully support innovationbecause knowledge that can be collected isalways explicit; this implication is argued explicitlyin Swan et al., (1999).

In summary, any transfer of tacit knowledge isoften deemed to require ‘connect’ approaches.Cognitivists might reply to the aforementioned

objections to ‘collected’ knowledge by arguing thatcontext-dependent tacit knowledge can be madeexplicit and then ‘collected’ by using realistic scenariosin knowledge capture sessions (e.g. see van Somerenet al., 1994). Dave Snowden himself accepts the needfor some ‘capture’ of knowledge or information aboutthe context in models (Snowden, 2011). Cognitivistswould also argue that abstracting knowledge fromits context is not only useful, it is essential if theimportant knowledge is to be clearly identified andits structure understood (cf. Collins and Harrison,1975). Regarding innovation, they would argue thatfor many business applications, it is not the creationof new knowledge that is important but rathermaking existing knowledge available to those whoneed it in a usable form.From a neutral standpoint, the most obvious

point that emerges from these debates is thatthere appear to be two separate strands of practicethat are both called ‘knowledge management’ but

Know. Process Mgmt. 19, 160–170 (2012)DOI: 10.1002/kpm

162 J. Kingston

which are based on different theories, use differenttechniques and support different business goals.Therefore, one of the keys to successful KM inorganisations is to understand the circumstances inwhich different techniques should be applied.

The remainder of this paper will examine fourknowledge dissemination techniques, chosen becausethey were perceived to be commonly used incommerce, industry and government; classify theminto Milton’s scheme; present examples of how theyhave been applied in a particular domain; and identifycircumstances in which that technique is likely tobe successful. It concludes with a review of theusefulness of Milton’s classification scheme inselecting knowledge dissemination techniques.

COMMUNITIES OF PRACTICE—‘INFORMALCONNECT’

Communities of practice (COPs) have been definedas follows:

‘A group of individuals who share a commoninterest, a set of problems or a passion and whoincrease their knowledge and the understanding ofthese aspects through interpersonal relationships’.(Wenger et al., 2002)

The key to using COPs as a KM disseminationtechnique is to encourage regular contact betweenCOP members about knowledge need, availableknowledge and ideas. Contact can be either throughface-to-face contact (in meetings and workshopsrather than one-to-one chats; the goal is for sharedknowledge to be available to the whole community)or through internet-based discussion fora and/orother collaboration spaces.

Communities of practice are an ‘informal connect’approach1 toKMdissemination, according toMilton’scategories; they connect those who need knowledgewith those who have knowledge, and the knowledgethat is expressed is usually in the form ofconversations.

Example: a community of practice in The Healthand Safety Executive

The UKHealth and Safety Executive (HSE)’s missionis to prevent death, ill health and injury to those atwork and those affected bywork activities. HSE staffsinclude (amongst others) inspectors, who inspectworkplaces for health and safety risks or breaches ofhealth and safety legislation; specialists, whose task

1It would be possible to turn COPs into an ‘informal collect’approach by recording conversations and storing them in a central,searchable repository. This is rarely done, however, perhapsbecause wikis are a more effective method of collecting andorganising knowledge from communities.


is to maintain an in-depth up-to-date knowledge ofa particular type of health and safety risk; and policymakers. HSE staffs are located in several offices,spread throughout Great Britain. HSE inspectors alsospend considerable time visiting workplaces; if anincident has taken place at a workplace, they maybe away from the office for several days at a time.The Process Safety Community of Practice and

Interest was originally conceived as a web-basedhelp desk in connection with setting out acceptableprocess safety standards for legal compliance.However, the ability to communicate widely and tointeract with practitioners not only allowed theconsultation and dissemination of new guidanceand standards but also allowed engagement of thewider community to consider particular technicalproblems and to agree an approach to take whereclear decisions could not easily be made.A clear advantage of the Process Safety community

is that it engages a technical audience, with a broadbase of skills and experience across wide anddiffering organisational boundaries. The communityhas been widened to include not only membersof the process safety discipline but also staff frommechanical engineering, electrical engineering,control and instrumentation, predictive specialistsand nuclear engineering.One community member stated that this knowl-

edge sharing forum provides an accessible conduitto ‘knowledge that is nascent in inspector’s brains’and also provides a feeling that ‘you are part of awider technical community, which can supportyou across organisational boundaries’. It has beenresponsible for storing information, technical refer-ences, web-based links, material for developingnew members of staff and documenting approachesto particular technical issues. It has also beenresponsible for the development of new Regulators’Development Needs Assessment CompetencyStandards, which are implemented to support theperformance review process for inspectors.This community is successful because of the

following:

• It allows a widely dispersed group of differingtechnical specialists to communicate with eachother across organisational boundaries;

• Community members can seek out knowledgeand experience on a technical issue from a topicexpert, who may otherwise remain unidentifiedwithin the organisation;

• The community can seek an independent opinionabout a technical issue without formality andsupport an approach to take on challenging issues;

• Information is uploaded, downloaded, stored andcommunicated for the interest and benefit of all;

• The community’s web site can be searched usingtechnical search terms, which quickly generates aview about what is known within the communityabout a particular issue;



• The community promotes networking andproblem solving;

• The community can generate new leads; and• The community reduces stress for inspectors who

can otherwise feel very alone in making difficulttechnical decisions.

When should communities of practice be used?

Communities of practice work best when

• Members have a shared history together becauseestablishing an effective COP requires a feeling ofbelonging to the community (Akkerman et al.,2008). This factor highlights the importance ofhaving face-to-face meetings of a community aswell as an online discussion forum (Dube et al.,2005). In the aforementioned example, theactivities of the Process Safety community weresupported by occasional face-to-face meetings;

• Geographically dispersed staff who have onlyoccasional face-to-face contact need to worktogether or to access each other’s expertise(Ardichvili et al., 2002);

• There is a need to obtain knowledge more quicklythan it could be obtained from written sources(Ardichvili et al., 2002);

• It is important to integrate new staff into a workcommunity quickly (Ardichvili et al., 2002; Rollaget al., 2005);

• There is an occasional need for new or innovativeknowledge. New knowledge is often generatedwhen a discussion of somebody’s question leadsnot just to a solution but also to an exchange ofideas about improved or new solutions (as seenin the development of new standards by theprocess safety community; see also Ardichviliet al., 2002).

• There is a possibility of unexpected or unpredictableinteractions with users, such as demands for‘knowledge’ that cannot be answered accuratelywithout addressing the user’s underlying assump-tions.2 In such cases, ‘collected’ knowledge mustbe accompanied by explanations of how the user’smental model needs to be modified to make use ofthe knowledge correctly. Humans who are havinga conversation—or the written equivalent—arebetter than almost any collection of knowledgecan be at diversifying responses, understandingcontext-based misapprehensions and providingrelevant explanations.

The size of a community should ideally be (as arule of thumb) between 10 and 100 people. If the

2An example is the infamous (and all too frequent) request tocomputer helpdesk technicians or computer repair shops for areplacement cup-holder because the one supplied with thecomputer has broken. Before answering the query, it is highlyadvisable to explain to the user the difference between a cup-holderand a CD-ROM drive.


group is less than 10, it is likely that each of themwill know all the others quite well, removing someof the need for a COP, and that any online forumwill be accessed only rarely, thus reducing itsusefulness. As for the upper limit, most literatureon KM implicitly assumes that large knowledge-sharing groups are more effective and efficient thansmall groups (Hansen et al., 1999; Mahnke, 1999)because ‘there is a big enough group to create auseful database of information’ (Cabrera and Cabrera,2002). However, Butler and Murphy (2007) arguedthat large numbers could negatively affect acommunity, as having too many members canmake it difficult to converge on a joint enterprise,interactions between any subset of members maybe occasional and unsustainable, and sharedrepertoires may be difficult to build, maintainand understand. There is also the considerationthat a large community that asks many questionsmight create unacceptable demands on the timeof the community’s experts. Voelpel et al., (2008)also discovered that smaller groups are morelikely to respond or engage in the act of know-ledge sharing while still providing a comparablyhigh level of knowledge. Voelpel et al. proposethat COP group sizes should be limited to around100 members.

KNOWLEDGE PORTALS—‘FORMALCONNECT’

Knowledge portals (KP) are information technology(IT) systems, often with a web page as a userinterface, that provide a single point of access tokey information and knowledge resources (Renekerand Buntzen, 2000). The information supplied canbe customised according to user requirements. Theycan bring together information from internal andexternal sources that are relevant to an individual ora group. Commonly used sources include corporatemessage boards, document search results, internaldatabases, specialised websites or news channels(Cloete and Snyman, 2003).In technical terms, KP are ‘middleware’: they

provide a mediating layer between the informationapplications and the individuals using them(Figure 2).Knowledge portals were first developed as

organisations sought to unify information accessand improve the management of informationresources by trying to replicate the popularity andsuccess of public domain, web-based portals suchas Yahoo and Google (Harris et al., 1999; Macket al., 2001). The KP approach can be seen as anatural evolution of intranets and groupwaresolutions into a common information infrastructure(Watson and Fenner, 2000).In Milton’s classification, KP are ‘formal connect’

systems. They connect users with other sources


Figure 2 The portal as ‘middleware’ (van Brakel, 2003)

164 J. Kingston

of knowledge rather than storing knowledge in arepository. The formal design of the KP acts as astructure for the knowledge contained within it.

Example: a knowledge portal in The Health andSafety Executive’s Chemical RegulationDirectorate

The Health and Safety Executive’s Chemical Regula-tion Directorate (CRD) has the task of approvingchemical substances (including pesticides) for useand recording that approval with a Europe-widedatabase. CRD needs to access relevant regulations,associated guidance and studies that have been car-ried out on the substance. There is a great deal ofguidance available for pesticides; the regulations,the guidance and other necessary information arestored on a variety of databases and websites. Someinformation is buried several levels down in EuropeanCommission websites, which are not always easy tonavigate. Furthermore, requests for approvals aresubmitted electronically in a structured format, andrecords of requests (and responses) must be storedon a database run by the European Commission ina similar structured format.

The Chemical Regulation Directorate introduced aKP to address the challenge of providing its staff withaccess to relevant technical information on pesticides.When the user selects a substance from the master listof substances (there are about 200 substances in total),a page of further links is displayed, including thebackground to decisionmaking and issues discussed.Staffs from each section who have received brieftraining from the IT department are able to createcustomised portals that provide extra informationrelevant to that section.

This KP approach is successful because ofthe following:


• it can provide users with links to all and only theinformation sources relevant to their work;

• it gives users direct access to hard-to-find webpages;

• it is not technically difficult to set up, so users canbe trained to set up their own systems; and

• by eliminating intermediate pages and by cachingfrequently accessed pages, it reduces ‘click-through’time and produces considerable improvement ininternet efficiency.

Another indirect advantage is that the file-sharingfacilities of the underlying software (MicrosoftSharePoint) mitigate the potential risks to reputa-tion and information security associated withsharing knowledge between staff at CRD’smultiple sites.

When should knowledge portals be used?

Knowledge portals should be used

• Where there are several key IT systems throughwhich staff repeatedly need to access specificinformation (Daniel and Ward, 2005);

• Where frequently used IT systems are difficult tonavigate, so staff would benefit from a customisedinterface to these systems (Mack et al., 2001);

• When there is a need for knowledge to bedrawn together from, or distributed to, looselycoupled or disconnected knowledge sources,for example, two different databases (van Baalenet al., 2005); and

• Where staff would benefit not only from accessto key IT systems but also improved access tocollaborative tools to share information andknowledge with each other (Detlor, 2000; Dias,2001; Firestone, 2003; van Baalen et al., 2005).



KNOWLEDGE CODIFICATION—‘FORMALCOLLECT’

Codified knowledge is defined here as knowledgethat has been reformatted and/or had additionalindexing features added to it. The purpose of this isto make the knowledge more understandable andaccessible, whether by a person or by a computerprogram.

There are several methods of knowledge codifica-tion requiring different degrees of modification ofthe knowledge in question.

Reformatting: small modificationsCodification methods that require only small modi-fications include drawing a diagram to represent aprocess that was originally described in words(creating a ‘knowledge map’); rearranging the orderof sections in a document; or abbreviating longparagraphs of text as checklists—for example,distilling the records of ‘lessons learnt’ exercises intoa list of key lessons. The codified knowledge thatresults from these processes is usually suitable foruser manuals or databases; an example can befound in Johnson, (2005).

Reformatting: large modificationsLarger modifications to knowledge are usuallyrequired if the knowledge is to be encoded into anIT system. Knowledge that has been captured asstories or in structured interviews may be searchedto extract knowledge that fits certain formats(e.g. rules of the form If X is true then Y is true3).Alternatively, one or more knowledge maps basedon the text may be produced, with the final map(s)being used as a specification for an IT system(Kingston, 1998). Databases may similarly beanalysed for associations between data entries thatcan be transformed into rules or a relationshipmap (Langley and Simon, 1995).

Indexing: small modificationsIndexing schemes can also be small scale or largescale. Small-scale indexing schemes include creatingan index to a book or document or creatinghyperlinks within a document to other parts of thedocument.

Indexing: large modificationsLarge-scale indexing includes the following:

• identifying and linking a list of key features to‘knowledge items’ in the document (Watson andMarir, 1994);

3For example, when capturing knowledge about eligibility forwelfare benefits under English law, one rule might be: ‘IF youhave a spouse or partner who is working full-time, then you arenot eligible for benefits X and Y’.


• extracting the ‘knowledge items’ and positioningthem within a newly created or existing taxonomyor other classification scheme (Chaudhry, 2010); and

• running analyses on text in documents to groupthemwith other documents with similar keywords(Marinai, 2008).

Reformatted knowledge is most often used tomakeguidance or procedures available to its intended users.If it is in the form of a manual or guidance document,it can be distributed as a readable document, whetheron paper or electronically. If the knowledge is encodedinto an IT system, the system will typically ask theuser some questions about the features of a currentsituation or problem and will use its encodedknowledge to suggest guidance for the situation or asolution to the problem.Indexed knowledge is used to support searching of

a large collection of knowledge. It is most often usedto support ‘lessons learnt’ processes, that is to searcha large collection of past cases/incidents/projects forthose that are relevant to the current situation. Theenquirer can extract and apply the appropriate lessons.Supporting software is available for all the

codification activities listed previously:

• Small-scale reformatting of knowledge can besupported by computer programs such ascomputer-aided software engineering tools ormind mapping software;

• Expert systems are computer programs thatencode expert knowledge that has been throughlarge-scale reformatting;

• Many technologies that support small-scaleindexing of knowledge are incorporated withinpopular desktop publishing packages;

• Once large-scale indexing has been performed,case-based reasoning is a computer technology thatscans a database of past problems and solutionsand uses indexed key features to match a currentcase against the most relevant past case(s).

Knowledge codification requires that all knowledgeis collected into a repository before being representedin a codified form, and the codification processalways requires organising the knowledge into somekind of structure, framework or model. In Milton’sclassification, knowledge codification techniques cantherefore be considered to be the epitome of ‘formalcollect’ systems.

Example: expert system in occupational hygiene

Estimation and Assessment of Substance Exposure(EASE (Kingston, 1996; Creely et al., 2005; Tickneret al., 2005)) is an expert system built for HSEfor assessing workplace exposure to potentiallyhazardous substances.The objective of the system is to enable assessors

(inspectors or specialists with expertise in occupa-tional hygiene) to produce an estimate of the extent


166 J. Kingston

to which workers will be exposed to a new orexisting hazardous substance. It achieves this byguiding the user through a decision tree, withconclusions that are drawn from an analysis ofHSE’s National Exposure Database. The system’scompleteness was validated by running exhaustivetests that followed every possible path through itsdecision tree and then checking the accuracy of itsoutputs.

The system helps assessors to remember all thenecessary questions to ask when considering anew industrial process and then estimates anexposure level from their answers. The system alsoincorporates backtracking facilities to allow errors(e.g. if the user supplies incompatible answers totwo questions) to be corrected and ‘What If’ analysesto be performed.

Estimation and Assessment of Substance Exposurehas now been superseded by other expert systems,notably the Advanced Reach Tool (Tielemans et al.,2011), but its lasting benefit to HSE has been itsinfluence on approaches to exposure assessmentwithin the European regulatory framework.

4This refers to a ‘community of practice’ in the sense of a group ofpeople with similar skills and interests.

When should codified knowledge be used?

Codification of knowledge is useful

• when there are many potential users of thedisseminated knowledge or where the potentialcost savings are very large. Codified knowledgeis expensive to produce compared with otherknowledge dissemination approaches, especiallyif large-scale modifications are made, so thereneeds to be a large user base to make it cost-effective. See Power et al., (1998) for an exampleof an expert system that saved a large procurementdepartment, with a few hundred staff, an esti-mated £30m per year;

• where there is a strong desire for a group of usersto use a standard form of knowledge and/or astandard approach to solve a particular problem;

• where a task can be performed better by an expertsystem than by a human expert because of timeconstraints. For example, the task of scheduling(or dynamic rescheduling) of aeroplanes by amajor airline, to ensure that aeroplanes arrive atmaintenance hubs just before their regularmaintenance check-ups are due, can be performedbetter by an expert system than by a human expert(Smits and Pracht, 1991);

• where the knowledge concerns a specific task thatincludes several decision points;

• where knowledge consists of an overview of theorganisation, including topics such as who has theauthority to make decisions or which departmentscarry out which procedures. Knowledge mappingcan provide a guide to this type of knowledge andalso to when, why or with what resources tasksare performed (Kingston, 2004);


• where the specific task being considered is one towhich all possible solutions can be listed. This is a‘rule of thumb’ because it is possible to usecodified knowledge to support tasks where it isnot practical to list all solutions (such as the airlinescheduling task described previously), but it isusually less cost-effective to support such tasks;

• when it is important to be sure that the knowledgebeing supplied by a dissemination technique iscomplete (verification) and accurate (validation).This is only possiblewhenusing ‘collect’ approachesand is easier with ‘formal collect’ approachesbecause the restructuring of the collected knowl-edge often highlights issues relating to verificationor validation. An example of such an issue is thepossibility, identified in the EASE system, thatusers might give answers to two questions that areincompatible with each other.

APPRENTICESHIP/MENTORING ANDTRAINING—‘INFORMAL CONNECT/FORMAL CONNECT’

The apprenticeship model is one of the oldest formsof learning, in which a novice obtains practicalexperience by working with and being instructed bya skilled craftsman, artisan or tradesman. A genuineapprenticeship involves a long-term commitmentthat is designed not merely to enable an apprenticeto achieve a threshold level of acceptability but toprepare the newcomer to function as a professionalmember of that trade (Douglas, 1982). Mentoringworks on the same principles as apprenticeship butmay be less formal or of shorter duration: forexample, job rotation or job swaps often requiresetting up a mentoring relationship.Features of successful apprenticeships include the

following:

• Recognition of achievements;• Entry into a COP,4 including social initiation into

the rules of the community;• Purposeful learning by carrying out an authentic

task;• Learning through practice, in the context within

which the task is most functional and useful;• Tasks that are graded and progressive;• Initiation of apprentices being seen as important

by the skilled staff, even when this process mighthinder the speed of work.

Training courses are often used to accompanyapprenticeship/mentoring so that staff learn formallyfrom the trainer and then put their training intopractice in the role of apprentices.Training is a ‘formal connect’ approach, inMilton’s

terms—the trainees are given the chance to connect



with a teacher who conveys structured knowledgeto them. Apprenticeship/mentoring is an ‘informalconnect’ approach, although with a few formalaspects in the progression of tasks that is organisedand in the recognition of achievements.

Example:mentoring inHealth andSafety Executive’sinspector training

The Health and Safety Executive’s trainingprogramme for newly recruited inspectors can beconsidered to operate on a training/mentoringmodel:

• The training programme is based upon sequencedstages and includes a mixture of on the job experi-ence and formal training, designed to develop theskills and knowledge required of an independentinspector.

• The mentoring aspects are provided to newinspectors through joint visits to workplaces withexperienced inspectors from their own and othergroups, as well as observing specific proceduressuch as prosecutions, tribunals and inquests. Thenew inspector’s role progresses from involvedobserver to lead inspector and then to independentinspector.

When should apprenticeship/mentoring be used?

Apprenticeship/mentoring can be seen as the defaultapproach to knowledge dissemination; it will work innearly all situations, but it is more time-consumingthan other methods. It takes a long time, typically2–5years, before the apprentice is fully experienced.It also requires quite a lot of the expert’s time foreach learner; it is not practical for an expert topass on his knowledge to more than a handful ofapprentices while still performing his regulartasks. The use of formal training reduces the latterproblem, but apprenticeship/mentoring cannotcompete for efficiency with other knowledgedissemination techniques, which can reach manylearners simultaneously.

Apprenticeship/mentoring is particularly usefulfor disseminating tacit knowledge or for tasks thatrequire knowledge of graphics or diagrams, perceptionor physiology (i.e. coordinated muscle movements).An example of perceptual, geometric and physio-logical knowledge can be found in the rail industry:the (now largely obsolete) task of ‘wheel tapping’involved striking the wheel of a railway engine orcarriage with a long-handled hammer and listeningto the resulting sound to determine if the wheelwas cracked or damaged. The primary skill usedby wheel tappers was perceptual (identifying‘incorrect’ sounds); there were also secondary skillsin knowing where and how hard to tap the wheel(geometric and physiological knowledge).


CONCLUSION

This conclusion highlights the pros and cons ofknowledge dissemination systems within each ofMilton’s four categories. These features should betaken into consideration when choosing which typeof system is appropriate.Informal connect techniques are typically the easiest

to set up and (if the experts are readily available)quick to use. They allow knowledge seekers to holdconversations with experts, which is important ifthe knowledge must be explained in more than oneway before the seeker can integrate the knowledgeinto his mental model. These conversations can alsobe a source of unexpected knowledge for seekers oreven of creativity or innovation, especially if morethan one expert is involved.It was predicted earlier in this paper that there

would be more, or more enthusiasm for, ‘informalconnect’ than other types of technique. This seemsto be the case: there are several techniques that aredesigned to allow people to share knowledgeverbally and informally, albeit within a structuredcontext. These techniques include peer assists(Knoco Ltd, 2010), storytelling (Denning 2004;Perrett et al., 2004), ‘an audience with’ (a highlyinteractive seminar) and social media. The socialmedia model in which someone makes a statementand others choose to comment on it is similar tothe approach of an online COP if the topic ofthe statement is a request for knowledge or an offerof knowledge.There are several arguments in the literature that

suggest that informal connect techniques arefavoured by knowledge seekers. Indeed, Cross andSproull (2004) found that, given a choice ofapproaches, users will often choose to obtainknowledge by ‘connecting’ with nearby experts,even when collected knowledge is easily availableand of high quality. The reason is that conversationswith more experienced individuals often supplystaff with more information than just the answerto the question. They may also have meta-knowledge (where to go to have more informationon the issue, or conversely, where not to go becausea certain report is outdated), problem reformulation(when the expert suggests a different way to look atthe problem or issue or highlights potentiallyunforeseen consequences of actions), validation(assurance that the approach the seeker is takingis on course and consequent appreciation) andlegitimising (an expression of approval by a personin authority or with known expertise, which theseeker can then use to influence others). Milton(2010) also suggests that informal connect systemsare popular because their informality makes themeasy to use and because users have the freedomto introduce new discussion topics in their ownwords rather than trying to understand the system’slanguage or classification.


168 J. Kingston

However, the potential for new discussions toopen up is also one of the disadvantages of informalconnect systems; in Milton’s words, they can end upmore as gossip than an exchange of knowledge.Another disadvantage is that the knowledge isusually not stored for the benefit of later users whomight need it; there is a risk that only those who usethe system at the time the request is answered willbenefit. It is also difficult for the organisation to verifyor standardise the knowledge that is shared.

Informal connect systems are also relativelydemanding of the time of experts, which becomesa problem if the number of people seeking theirknowledge exceeds 100 or so.

Formal connect systems typically require basic ITskills from developers and users. They are mainlyuseful when there are a number of informationsources (human or IT) that need to be consulted,and the resulting information amalgamated to solvea problem or where there are information sourcesthat need to be consulted regularly because theircontent is regularly updated. These informationsources may be in-house sources that consist ofcollected knowledge; for example, Milton arguesthat formal connect systems are ideal for sharing‘lessons learnt’ in areas of complex or context-specific need or for accessing knowledge abouttopics that are rapidly changing. Formal connecttechniques may therefore be most useful when usedin conjunction with collect approaches.

Formal connect systems usually reduce effort forthe knowledge users but do not necessarily reduceeffort for knowledge providers.

Informal collect approaches arguably include allforms of publishing,5 but a more practical viewrestricts ‘informal collect’ knowledge disseminationtechniques to material that is ‘published’ on avoluntary or as needed basis to address a perceivedlack of available knowledge. Techniques in thiscategory include blogs and wikis. Published text isusually easy to understand and often providesbackground context as well. However, text cansometimes be hard to search or to query, particularlyin cases where multiple terms can be used to describesimilar concepts. Semantic tagging (e.g. withinsemantic blogs (Cayzer, 2004)) can reduce thesedifficulties.

Collect approaches are always more laborious toimplement and to maintain than connect approaches,as the knowledge has to be written down (‘captured’),and the user must then search it for knowledgerelevant to their needs. However, the ‘capture’process

5Many organisations have libraries of internal reports or access tocollections of publications on the web that contain valuableknowledge. The primary technique at present for searching suchrepositories for knowledge is text mining (as introduced inMarinai, 2008; see Kiryakov et al., 2010 and Doms and Schroeder,2005 for some examples). However, because text mining is a tech-nique for searching for knowledge rather than for disseminatingknowledge, it is beyond the scope of this paper.


can be shared between many people in systemssuch as wikis. Perhaps the biggest advantages of‘collect’ approaches within organisations are thatthe knowledge now resides in a record, and thatrecord belongs to the organisation rather than to anindividual. The fact that the knowledge is recordedmeans that the knowledge can be verified andvalidated, and the results can be supplied to users as‘standard’ knowledge. However, a correspondingdisadvantage of voluntary approaches to knowledgesharing is that the body of collected knowledge islikely to be very incomplete; Milton argues thatvoluntary wikis draw on only about 2%–3% of allavailable knowledge about a topic.Formal collect systems require the most effort to

establish, as the knowledge has to go through threedistinct stages: capture, structuring and makingthe knowledge comprehensible for dissemination.However, this very process subjects the knowledgeto greater analysis than in any of the other techniques,which brings advantages if the knowledge needs tobe verified, validated, standardised or made easier tounderstand. The way in which users access or makeuse of this knowledge can also be controlled quitetightly (if desired) by building it into IT systems. Theultimate in control is the expert system, which isunique amongst knowledge dissemination techniquesin that its primary purpose is not to supplyknowledge to users to enable them tomake decisions;instead, it directs queries to the users, makes its owndecisions and then produces an answer. For userswho want to learn how decisions were made or todecidewhether the system is trustworthy, appropriatejustifications are usually made available. Thesesystems are therefore most suitable for tasks wherethere are a large number of users with low levels ofexpertise.Alternatively, IT-based ‘formal collect’ systems

can utilise the power of computing to automatetasks that are too complicated for an expert toperform optimally; these are usually tasks such asconfiguration, design, planning or scheduling thatfollow a set of knowledge-based principles with aset of constraints but have a very large number ofpossible solutions.The biggest difficulty that formal collect systems

have is being applicable to a wide range of problems.The need to follow a structured dialogue mayoccasionally make it difficult for users to enter thecontent into the system that they want to enter, andif any knowledge is required that goes beyond theknowledge stored in the system, then a formal collectsystem is simply unable to supply it or support asearch for it.It was predicted earlier that formal collect

approaches would outnumber, or at least engendermore enthusiasm, than informal collect or formalconnect approaches. In practice, this prediction seemsto be mitigated by users’ natural preference forinformal connect approaches (as identified by Cross



and Sproull) and by the effort required to design anddistribute formally collected knowledge. However,in circumstances where formal collect approachesare justfiied, they can provide organisations withvery large savings through improved efficiency andthrough more effective decision making.

ACKNOWLEDGEMENT

This publication and the work it describes were funded bythe Health and Safety Executive (HSE).

REFERENCES

Akkerman S, Petter C, De Latt M. 2008. Organisingcommunities of practice: facilitating emergence. Journalof Workplace Learning 20(6): 383–399.

Alavi M, Leidner D. 1999. Knowledge ManagementSystems: Issues, Challenges, and Benefits.Communicationsof AIS 1: 14.

Ardichvili A, Page V, Wentling T. 2002. Motivation andBarriers to Participation in Virtual Knowledge-SharingCommunities of Practice. In: Proceedings of the OKLO2002 Conference, Athens, Greece.

van Baalen P, Bloemhof-Ruwaard J, van Heck E. 2005.Knowledge sharing in an emerging network of practice:The role of a knowledge portal. European ManagementJournal 23(3): 300–314.

van Brakel P. 2003. Information portals: a strategy forimporting external content. The Electronic Library21(6): 591–600.

Butler T, Murphy C. 2007. Implementing KnowledgeManagement Systems in Public Sector Organisations:A Case Study of Critical Success Factors. In: Proceedingsof the 15th European Conference on Information Systems, St.Gallen, Switzerland, 12: 612–623.

Cabrera A, Cabrera EF. 2002. Knowledge-sharing dilemmas.Organization Studies 23(5): 687–710.

Cayzer S. 2004. Semantic Blogging and DecentralizedKnowledge Management. Communications of the ACM,47(12): 47–52.

Chaudhry AS. 2010. Assessment of taxonomy buildingtools. The Electronic Library 28(6): 769–788.

Cloete M, Snyman R. 2003. The enterprise portal - is itknowledge management? In: Proceedings of Aslib 55(4):234–242.

Collins HM, Harrison R. 1975. Building a TEA Laser:The Caprices of Communication. Social Studies of Science5: 441–50.

Creely KS, Tickner J, Soutar AJ, Hughson GW, Pryde DE,Warren ND, Rae R, Money C, Phillips A, Cherrie JW.2005. Evaluation and further development of EASEmodel 2.0. Annals of Occupational Hygiene 49(2): 135–46;2005.

Cross R, Sproull L. 2004.More Than anAnswer: InformationRelationships for Actionable Knowledge. OrganizationScience 15(4): 446–462.

Daniel E, Ward J. 2005. Enterprise portals: Addressing theorganisational and individual perspectives of informationservices. In: Proceedings of the 13th European Conference onInformation Systems, Regensburg, Germany, May 26–28.

Denning S. 2004. Telling Tales. Harvard Business Review.Detlor B. 2000. The corporate portal as information

infrastructure: towards a framework for portal design.International Journal of InformationManagement 20(2): 91–101.


Dias C. 2001. Corporate portals: a literature review of anew concept in Information Management. InternationalJournal of Information Management 21(4): 269–87.

Doms A, Schroeder M. 2005. GoPubMed: exploringPubMed with the Gene Ontology. Nucleic Acids Research33(suppl. 2): W783–W786. Oxford University Press.http://nar.oxfordjournals.org/content/33/suppl_2/W783.long. Accessed February 2012.

Douglas RL. 1982. Arbitrators, apprentices, and arbitration.The Arbitration Journal 37(3): 46–51.

Dube L, Bourhis A, Jacob R. 2005. The impact of structuringcharacteristics on the launching of virtual communitiesof practice. Journal of Organisational Change Management18(2): 145–166.

Firestone JM. 2003. Enterprise Information Portals andKnowledge Management. KMCI Press/ButterworthHeinemann: Boston, MA.

Goodwin SC. 2009. Formal Knowledge Sharing inMedium-to-Large Organizations: Constraints, Enablersand Alignment. PhD Thesis, University of Bath.

Hansen JR, Nohria N, Tierney T. 1999. What’s YourStrategy for Managing Knowledge? Harvard BusinessReview, March-April.

Harris K, Phifer G, Hayward S. The enterprise portal: Is itknowledge management? 1999. Gartner; SPA-08-8978,www.gartner.com. Accessed January 2011.

Heylighen F. Epistemological Constructivism. 2003. http://pespmc1.vub.ac.be/EPISTEMI.html. Downloaded March2011.

Johnson AM. Using Concept Mapping to Facilitate StudentsUnderstanding Software Testing. 2005. http://www.testingeducation.org/conference/wtst4/AMJohnsonUsing%20Concept%20Mapping%20to%20Teach%20Software%20Testing%20(2).doc. Accessed January 2011.

Kingston JKC. 1996. Building a KBS for Health andSafety Assessment. In: Proceedings of Applications andInnovations in Expert Systems IV, proceedings of BCSExpert Systems’96, Cambridge. SGES Publications;1996, 16–18.

Kingston JKC. 1998. Designing Knowledge Based Systems:The CommonKADS Design Model. Knowledge BasedSystems Journal 11(5–6): 311–319.

Kingston JKC. 2004. Conducting Feasibility Studies forKnowledge Based Systems. Knowledge Based SystemsJournal 17(2–4): 157–164.

Kiryakov A, Bishop B, Ognyanoff D, Peikov I, Tashev Z,Velkov R. 2010. The Features of BigOWLIM that EnabledtheBBC’sWorldCupWebsite. In:Proceedings of theWorkshopon Semantic Data Management at VLDB 2010, Singapore.http://ceur-ws.org/Vol-637/paper6.pdf. Sep 17.

Knoco Ltd. KnowledgeManagement Processes: PeerAssist.2010.http://www.knoco.com/peer-assist-page.htm.AccessedFebruary 2012.

Langley P, Simon HA. 1995. Applications of MachineLearning and Rule Induction. Communications of theACM 38(11): 54–64.

Mack R, Ravin Y, Byrd RJ. 2001. Knowledge portals andthe emerging digital knowledge workplace. IBMSystems Journal 40(4): 925.

Mahnke V. 1999. The Economies of Knowledge-Sharing:Production- and Organization Cost Considerations. In:Proceedings of the 6th conference of the Danish ResearchUnit for Industrial Dynamics, Winter.

Marinai S. 2008. Introduction to Document Analysis andRecognition. Studies in Computational Intelligence 90: 1–20.

MiltonN. The Polarity of KM Ideologies. 2009. http://www.nickmilton.com/2009/10/polarity-of-km-ideologies.html.Accessed February 2011.

Milton N. 2010. The Lessons Learnt Handbook: PracticalApproaches to Learning from Experience. Chandos Pub-lishing: Oxford.


http://nar.oxfordjournals.org/content/33/suppl_2/W783.long

http://nar.oxfordjournals.org/content/33/suppl_2/W783.long

http://www.testingeducation.org/conference/wtst4/AMJohnsonUsing%20Concept%20Mapping%20to%20Teach%20Software%20Testing%20(2).doc




http://www.nickmilton.com/2009/10/polarity-of-km-ideologies.html

http://www.nickmilton.com/2009/10/polarity-of-km-ideologies.html

170 J. Kingston

Nonaka I, Takeuchi H. 1995. The knowledge creatingcompany: how Japanese companies create the dynamics ofinnovation. Oxford University Press: New York.

O’Dell C, Grayson CJ. 1998. If we only knew what weknow: the transfer of internal knowledge. The FreePress: New York.

Perrett R, Borges MRS, Santoro FM. 2004. ApplyingGroup Storytelling in Knowledge Management. In:Groupware: Design, Implementation and Use. LectureNotes in Computer Science.

Power R, Reynolds S, Kingston J, Harrison I, Macintosh A,Tonberg J. 1998. Expert Provisioner: ARangeManagementAid. Knowledge Based Systems Journal 11(5–6): 339–344.

Reneker MH, Buntzen JL. 2000. Enterprise knowledgeportals: two projects in the United States. Departmentof the Navy. The Electronic Library 18(6): 392.

Rollag K, Parise S, Cross R. 2005. Getting New Hires Upto Speed Quickly,MIT/Sloan Management Review, WinterIssue: 35–44.

Schreiber G, Akkermans H, Anjewierden A, de Hoog R,Shadbolt N, van de Velde W, Wielinga B. 2000. Knowl-edge Engineering and Management: The CommonKADSMethodology. The MIT Press: Cambridge, MA.

Smits S, Pracht D. 1991. MOCA—A Knowledge-BasedSystem forAirlineMaintenance Scheduling. In:Proceedingsof the Innovative Applications of Artificial IntelligenceConference. AAAI.

Snowden D. Sensemaking. 2011. www.cognitive-edge.com, downloaded June 2011.

Snowden DJ, Boone ME. 2007. A Leader’s Framework forDecision Making. Harvard Business Review.


van Someren MW, Barnard YF, Sandberg JAC. 1994. TheThink Aloud method: A practical guide to modellingcognitive processes. Department of Social ScienceInformatics, University of Amsterdam. Academic Press:London.

Swan J, Newell S, Scarbrough H, Hislop D. 1999.Knowledge management and innovation: networksand networking. Journal of Knowledge Management 3(4):262–275.

Tickner J, Friar J, Creely KS, Cherrie JW, Pryde DE,Kingston JKC. 2005. The Development of the EASEModel. Annals of Occupational Hygiene 49(2): 103–110.

Tielemans E, Warren NE, Fransman W, Van Tongeren M,Mcnally K, Tischer M, Ritchie P, Kromhout H, SchinkelJ, Schneider T, Cherrie JW. 2011. Advanced REACHTool (ART): Overview of Version 1.0 and ResearchNeeds. Annals of Occupational Hygiene 55(9): 949–956.See also http://www.advancedreachtool.com. AccessedMarch 2012.

Voelpel SC, Eckhoff RA, Forster J. 2008. David againstGoliath? Group size and bystander effects in virtualknowledge sharing. Human Relations 61(2): 271–295.

Watson J, Fenner J. 2000. Understanding Portals. TheInformation Management Journal 34(3): 18–22.

Watson I, Marir F. 1994. Case based Reasoning: A Review.Knowledge EngineeringReview 9(4): 327–354. Also availableat http://www.ai-cbr.org/classroom/cbr-review.html.Accessed January 2011.

Wenger E, McDermott R, Snyder WM. 2002. CultivatingCommunities of Practice: A Guide to Managing Knowledge.Harvard Business School Press: Cambridge, USA.


http://www.advancedreachtool.com

http://www.ai-cbr.org/classroom/cbr-review.html

choosing a knowledge dissemination approach

Documents