Knowledge Capture for the Design of a Technology Assessment Tool

Daniela Oliveira

and Kimiz Dalkir

Independent Researcher

School of Information Studies, McGill, Montreal, Canada

Keywords: Knowledge Management, Artificial Intelligence, Technology Assessment, Artificial Intelligence Assessment,

Applied Knowledge Management, Artificial Intelligence Documentation.

Abstract: The design of technology assessment tools is an important Knowledge Management endeavour. Technology

assessment is a subject where consensus is far from being achieved. Any project intended to create a

technology assessment tool is expected to generate a lot of discussion or criticism. Among the most critical

kinds of technology, Artificial Intelligence (AI) is a highly polemic kind of technology. Its impacts are

important and multidisciplinary. Moreover, the technology evolves quickly and so do the attitudes toward that

technology. Therefore, business owners intending to produce an AI assessment tool should expect extensive

discussion of different points of view, but also support the continuation of the discussion throughout time and

with different stakeholders. Surprisingly, technology assessment tools developed by business owners have

been particularly neglected in the coalescent discussion about AI documentation, not to mention the support

to create those tools. To foster a continuous innovation flow, business owners should pay particular attention

to how discussions are captured. This paper explores the foundations of knowledge management initiatives

to support the design of an artificial intelligence assessment tool at the business owner, in a process that

supports continuous discussion and innovation. This article also suggests project aspects and supporting

document structure.

1 INTRODUCTION

Making people exchange ideas, validate knowledge,

and create new knowledge together are some of the

challenges from the Knowledge Management field.

The field is also interested in helping experts produce

consumable documents and, for that, going through

the process of deciding which information should be

retained and which should be left out. A challenge

that does not spare documentation efforts around

Artificial Intelligence (AI). “Determining what

information to include and how to collect that

information is not a simple task”, argued Richards et

al. (2020,p. 1), while designing a document structure

intended to support reporting about AI services.

Documentation challenges involve identifying

what knowledge is mature enough to be written down

(such as a new methodology that has been tried out

enough times to have an article written about it) and

articulating knowledge that has not been yet

https://orcid.org/0000-0001-9285-0173

https://orcid.org/0000-0003-3120-6127

expressed (such as the acknowledgement of different

roles and expectations in a new process).

Documentation is particularly challenging when it

involves knowledge about the design phase. Design

is a project phase where several ideas are articulated.

In this perspective, Design is also a knowledge

process. In knowledge processes, some ideas are

retained, others are not (McElroy, 2011). Design is a

process very rich in terms of information about the

end product. In this phase, the values surrounding the

project take shape and the motivation behind the

project defines itself. This definition phenomenon is

maybe more visible when the discussions touch

different knowledge domains, as it happens to be the

case for AI evaluation tools.

AI academicians and practitioners are in the

beginning of some sort of consensus regarding what

to evaluate in AI systems, and when and by whom.

The design of an AI evaluation tool is therefore a

good candidate for heated discussions involving the

interaction of concepts from different fields and

Oliveira, D. and Dalkir, K.

Knowledge Capture for the Design of a Technology Assessment Tool.

DOI: 10.5220/0011551400003335

In Proceedings of the 14th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (IC3K 2022) - Volume 2: KEOD, pages 185-192

ISBN: 978-989-758-614-9; ISSN: 2184-3228

185

having different impact levels on different people at

different stages of the AI lifecycle. This study intends

to help the creation of a support structure to foster the

design of a technology assessment tool. It can be

particularly helpful for business owners to develop AI

assessment tools.

Documentation in the design phase may capture

knowledge at its state-of-the-art portrait at that time.

It may capture motivation definition. The capture of

the state of the knowledge at the design phase might

help the end product evolve throughout time, as that

knowledge also evolves, because dependencies on

outdated knowledge can be more quickly identified.

The capture of the motivation behind the project, in

addition to increasing its transparency levels, might

also help the end product evolve, as this product

acceptability increases, for instance. In this sense,

documentation in the design phase might help

awarding the end product a continuous innovation

flow, where incremental developments have their

barriers lowered.

In the AI ecosystem, technology and approaches

evolve quickly and so does the acceptability of the

resulting products. The field is the perfect candidate

for the adoption of documentation facilitating a

continuous innovation flow.

2 CONVEYING KNOWLEDGE

ABOUT AI

Richards et al. (2020) argue that the diversity of

information needs that different stakeholders might

have makes it impossible that one single document

addresses all needs in a consumable format, even

within the same domain or organization.

For example, Mitchell et al. (2019) have

suggested a documentation paradigm to describe a

machine learning model. In the short documents

produced according to this documentation paradigm,

named “model cards”, performing characteristics of

the model should be conveyed so that potential users

can understand the systematic impacts of the model

before its deployment. Information such as type of

model, intended use cases, attributes for which model

performance may vary, measures of model

performance, as well as the motivation behind chosen

performance metrics, group definitions, and other

relevant factors should be included. Mitchell et al.

(2019) state the tool intends to help stakeholders to

compare candidate models, understand the limits of

each model and better decide on which model is more

suitable for a given situation. In practice, the

definition of stakeholders in this case is somehow

limited. While the tool should “aid policy makers and

regulators on questions to ask of a model, and known

benchmarks around the suitability of a model in a

given setting” ([p.2]), the target audience is

developers, particularly those interested in including

the model in a larger technological solution. For

Richards et al. (2020), if the documentation is to be

useful, it has to be tailored to their target audience and

to the use this target audience is to make of the

product. Indeed, while in Mitchell et al. (2019) there

is a concern regarding the length of the document (the

models cards should be “short”), in Richards et al.

(2020), the perspective of reporting is changed to suit

the needs of developers that would include models in

a larger technological solution: instead of reporting

characteristics of the model, characteristics of the AI

services, that could include many models, are

reported.

It is therefore reasonable to expect that the

documentation support needed for reporting

characteristics of an AI model or an AI service should

be different from the documentation support needed

for assessing the suitability of that AI model or

service. This assessment should evaluate the

alignment of that model or service with other criteria,

for example, the policies and practices in an

organization.

No documentation approach seems to cover the

whole machine learning cycle, neither to address all

the needs of all audiences (Garbin & Marques, 2022;

Laato et al., 2021).

2.1 Technology Assessment

Technology assessment involves, but is not limited to,

approaches and tools that allow:

 The evaluation of the suitability of a

technological solution to a particular situation or

business need;

 As mentioned, the alignment of the

technological solution with the policies and

practices of the organization;

 The evaluation of positive and negative,

intended and unintended, current and future

impact on the situation, people, environment

and other technological solutions;

 The comparison of one technological solution to

another;

 In a larger spectrum, the comparison among

approaches or technologies.

KEOD 2022 - 14th International Conference on Knowledge Engineering and Ontology Development

186

2.2 Artificial Intelligence Applications

Assessment

AI applications may have high positive impact, but

they also have a great risk of generating negative

impact. AI applications may “violate privacy,

discriminate, avoid accountability, manipulate and

misinform public opinion, and be used for

surveillance” (Janssen et al., 2020), among other

risks.

Every professional involved in the AI cycle has a

responsibility towards increasing AI transparency

and limiting its potential to cause harm, be they AI

producers, regulators or executive board members

and managers of organizations of organizations

making use of AI (Laato et al., 2021). On one hand, it

is necessary for AI producers to better report the

characteristics, uses and limitations of AI models and

services. On the other, well-designed and well-suited

governance approaches to AI are necessary, to define

and monitor its potential negative implications and

limit those implications with effective and timely

responses to incidents.

AI risk assessments are necessary at many levels.

At the level of application domains and at the

institutional level (Winfield & Jirotka, 2018) and at

the level of individual systems (Janssen et al., 2020;

Winfield & Jirotka, 2018). Even if the AI solution in

question has an explainable AI approach, what to

explain and how to explain it might differ from one

domain to another and from one organization to

another (Laato et al., 2021). In addition, solutions

containing explainable AI modules still must be

monitored, as “blindly trusting findings from any

usability research in the XAI field would be

counterproductive due to the novelty and formative

state of the research area” (Laato et al., 2021, p. 20)

In each domain or organization, regulations,

culture and then policies, principles and procedures

are mechanisms for the establishment of thresholds of

acceptable behavior, mechanisms that both influence

and are influenced by societal expectations, norms

and values (Janssen et al., 2020). How can these

mechanisms be used for the creation of AI technology

solutions assessment tools? An approach and a tool

example from Knowledge Management research and

practice follows.

3 KNOWLEDGE MANAGEMENT

Knowledge Management concerns all questions

regarding the acquisition, the development, the

sharing, the exploitation, and protection of

knowledge (Dalkir, 2011).

Applied Knowledge Management is about the

development and tailoring of initiatives and tools

from Knowledge Management regarding a particular

field or activity.

3.1 Applied Knowledge Management

and Technology Assessment

The creation of approaches and tools is some ways no

different from other creativity endeavour in a

business environment. The ideas to be generated need

motivation, expertise and creative skills at their origin

and are required to be “appropriate, useful and

actionable” (Amabile, 1998, p. 79). Creative work is

often expected to be developed in groups (Hennessey

& Amabile, 2010) and work involving knowledge is

often linked to connectedness (Nahapiet, 2009). It is

possible that collaborative work increases the

flexibility and robustness of the solution. In any case,

it includes different perspectives and has the potential

to increase buy-in (Oliveira, 2022).

Applied Knowledge Management can support this

process by removing possible roadblocks and

otherwise creating conducive conditions so that better

goals can be attained more quickly, in addition to

providing individuals, groups and organizations with

a positive experience.

3.2 Knowledge Management and

Artificial Intelligence

Knowledge management around the evaluation of

technological solutions using artificial intelligence

tends to raise challenges that may not be raised in the

evaluation of other technologies. Some of these

challenges are:

 The multidisciplinarity of fields required for the

evaluation. Portraits of Artificial intelligence

solutions have raised social, economic,

technological, linguistic, ethical, legal,

management and philosophical issues, to name

only some;

 The global nature of collaboration: research

from academia and from companies around the

world are mutually influenced by new

developments in the field;

 The field is still in its early stages.

Knowledge management initiatives supporting the

development of technology evaluation approaches

and tools for technological solutions involving

artificial intelligence must then take into account

Knowledge Capture for the Design of a Technology Assessment Tool

187

collaborative work among professionals with a

plurality of backgrounds and a high level of

knowledge acquisition and development.

3.3 Knowledge Development and

Innovation

Development, sharing and exploitation of knowledge

are processes strongly related. Knowledge

development is associated with innovation, as the

creation of new knowledge has the potential to propel

the organization into new venues. While the

development of knowledge or of new ideas can be

done individually, more and more frequently this

process is undertaken in groups (Carrier & Gélinas,

2011; Fisher & Amabile, 2008). Knowledge sharing

is then a process that influences knowledge

development. Among other reasons, knowers might

share developed knowledge in order to validate this

knowledge (Mokyr, 2000), a process that also occurs

with knowledge acquired by an individual outside the

organization. Knowledge validation is necessary for

the subsequent application of this knowledge. Once

the knowledge has been embedded in processes,

services or products, it can be said to be exploited. In

the case of the evaluation of technological solutions

involving Artificial intelligence, knowledge

surrounding artificial intelligence, technology

evaluation and relating themes must be sought outside

the organization or developed internally and then

validated. These processes might occur before or

during the process of design of an actual technology

evaluation approach or tool.

3.3.1 Supporting Knowledge Acquisition

Knowledge from outside the organization can be

acquired through a structured organizational

initiative, but it can also enter the organization

through an employee that acquired that knowledge on

their own (Shoham & Hasgall, 2005). This employee

may act as a sponsor of this knowledge and advocate

its integration into the organizational knowledge.

There are many initiatives that can support

knowledge acquisition. Direct support can include

providing access to academic resources or training

and allowing employees time to explore those

resources. The knowledge acquisition process can

also be supported indirectly through the

organizational endorsement of the whole knowledge

management cycle, particularly development, sharing

and exploitation stages. If employees are allowed and

encouraged to validate knowledge externally

acquired, they will feel also encouraged to seek future

acquisition of knowledge.

One important element when validating

knowledge that was acquired outside the organization

is to acknowledge its provenance. Provenance holds

a symbolic weight that might be useful when

advocating for the acquired knowledge. This

symbolic weight might indicate, among other aspects,

maturity of scholarship, interdisciplinary points of

view, importance of the subject or the practical

potential of the knowledge in question. It is therefore

interesting to include provenance in knowledge

management tools designed to support the validation

of acquired knowledge.

3.3.2 Supporting Knowledge Development

Knowledge developed inside the organization might

combine acquired knowledge with previously

internally developed knowledge. Therefore,

knowledge management tools supporting knowledge

development should include the elaboration of the

new idea or statement and the possibility of

mentioning the provenance of both externally

acquired and internally developed knowledge.

3.3.3 Supporting Knowledge Application

Once a particular knowledge claim has been validated

(McElroy, 2011), it is then time to validate the

application of this same knowledge claim. Statements

explaining the application of knowledge tend to be

prescriptive and respect practical constraints. They

are therefore different in nature from the statements

describing the knowledge at their origins, which can

be more abstract or general.

In the design of an evaluation tool, the

presentation of the knowledge acquisition or

development statement beside the knowledge

application statement allows the reader to understand

the reasoning behind the application statement and

imply the organizational constraints that were

considered along with the knowledge acquired or

developed. It is the knowledge about knowledge, or

metaknowledge, helping the understanding of the

knowledge itself.

The promotion of the understanding of the design

process, beyond the end result, is one of the most

important elements in supporting the creation of

evaluation approaches and tools of technological

solutions involving artificial intelligence.

As the technology evolves and more of its impacts

and possible mitigation solutions are known, it is

important to facilitate the identification of which

areas of the evaluation tool are less current or

KEOD 2022 - 14th International Conference on Knowledge Engineering and Ontology Development

188

applicable and should be revised. Therefore, while the

understanding of the design process as a whole is to

be supported, the understanding of the design process

of each application statement is equally important. It

is also important to consider that the team charged

with the revision of the evaluation tool might be

considerably different from the team that developed

the tool. Facilitating the granular understanding of the

design process supports the updating process of the

evaluation tool in general and the increasing diversity

in the revision team in particular.

The support to the granular understanding of the

design process reduces the need for the revision team

to understand all the aspects of the project. This

facilitates the collaboration with experts in different

domains and reduces the pressure on the tacit

knowledge of the team members.

3.3.4 Supporting Explicit and Tacit Kinds of

Knowledge

The term “tacit knowledge” refers to knowledge that

has not been expressed in any kind of language

(Nonaka & Takeuchi, 1995). “Explicit knowledge”

addresses the knowledge that has been articulated to

the extent that it can be understood without needing

direct access to the holder of that knowledge. If a

continuum of the media holding knowledge is

considered, human minds would be on one extreme

whereas documents would be on the other (Oliveira

et al., 2021).

Explicit knowledge can be supported through

fields where the provenance of the knowledge claim

can be indicated. Supporting tacit knowledge requires

different strategies. One of those is a field where the

name of an employee sponsoring the knowledge can

be codified, as well as the name of an employee

endorsing the proposition. This strategy credits

employees with their efforts in the design of the

evaluation tool, for instance, and carries the symbolic

weight of their expertise.

4 CODIFICATION EFFORTS OF

KNOWLEDGE AND

METAKNOWLEDGE, AI AND

INNOVATION FLOW

Documentation, in most technological solutions,

focuses on the resulting tool. It is intended to

accompany the tool and help client developers make

good use of the tool. This documentation will usually

cover only the application of the knowledge claims

that have been validated in the design of the tool. The

documentation would not articulate the knowledge

claims nor describe the knowledge and processes

involved in validating those claims. In other words, it

would only present the knowledge itself, and not the

metaknowledge surrounding the technological

solution. After all, the aim of the documentation is to

support use of the tool, not necessarily the

development of the knowledge involved in designing

the tool.

Transparency in AI applications require that

codification efforts for the client developer go a little

further, both in terms of the knowledge and of the

metaknowledge surrounding the solution. In terms of

the knowledge, AI documentation developed by its

producer should cover how the evaluation the AI

model or service went through before it was made

available (Mitchell et al., 2019; Richards et al., 2020),

as a part of its quality control or application limits

definition. In terms of metaknowledge, the

documentation should cover the motivation behind

the choice of the metrics used to evaluate the AI

application (Mitchell et al., 2019). There is then a

need to promote codification efforts of this

metaknowledge, even for the documentation intended

for client developers’ use. The process supported by

the documentation suggested in this communication,

however, is the assessment of AI applications

performed by the business owner, or the first client of

AI applications.

4.1 The Business Owner’s Assessment

Tool

The business owner’s assessment team should be

composed of developers and members from other

functions in the business working together. Some

portions of the assessment would require more

technical expertise, while other portions would need

a broader view of the business to assess impact and

alignment, but most of the assessment will need

collaboration among professionals of different

backgrounds. The AI application would be evaluated

and compared with other applications with respect to

its technical approaches, but the alignment of the AI

application with the domain and the business culture

and with the policies and regulations surrounding the

business and the intended application environment

must also be ensured.

Reaching coherent decisions in this diverse

knowledge environment is a challenge, particularly if

the members of the assessment team change during

the assessment process. Documentation supporting

the knowledge acquisition, development and

Knowledge Capture for the Design of a Technology Assessment Tool

189

application processes would help the assessment team

remain coherent throughout the assessment. Along

the same lines, this documentation can be especially

helpful when the organization is ready to transform

the ad-hoc AI assessment process into a structured

and more established one and starts the design of an

AI assessment tool to guide the process.

4.2 Documentation Granularity at the

Level of Knowledge Claims

When a technology assessment tool is designed, the

knowledge claims surrounding the tool are validated

or rejected. The validated claims will join the design

process, and they will most likely be embedded in the

resulting tool.

In the case of the design of an AI assessment tool,

one approach would be to have the documentation

cover the assessment tool as a whole. This approach

gives a broad view of the assessment tool as is

probably the best option for the purposes of training

of new members of the assessment team or to present

the work of the assessment team to other departments

of the organization.

A more granular approach is however necessary

when knowledge surrounding AI evolves and

knowledge claims have to be revaluated. In this case,

it is interesting to quickly and unambiguously identify

the knowledge claims that might be affected by new

knowledge. In this way, barriers for a continuous

innovation flow are lowered and the updating process

of the assessment tool is facilitated.

In these codification efforts aiming at a

continuous innovation flow, the connection between

knowledge and metaknowledge must be made clear.

The documentation efforts should present which

portions of the end product are connected to which

knowledge claims, and the knowledge surrounding

the validation of those claims. Some documentation

of the social capital, “the sum of the actual and

potential resources embedded within, available

through, and derived from the network of

relationships possessed by an individual or social

unit” (Nahapiet & Ghoshal, 1998, p. 243), involved

would also help understand the importance of

knowledge claims. The documentation suggested in

this communication is part of the codification efforts

that support the evolution of the knowledge

surrounding AI applications. The documentation

suggested enforces codification efforts of the

knowledge claims validation process, covering the

knowledge management phases of acquisition,

development and application and both tacit and

explicit knowledge.

5 CAPTURING ASPECTS OF THE

KNOWLEDGE CLAIM

An AI assessment tool encompasses a number of

criteria used to identify the degree in which a

particular AI application presents interesting features.

Each criterion comes from a knowledge claim that

was validated. The documentation suggested captures

aspects of the knowledge claim validation process.

The principle is the expression of the knowledge

claim.

The motivation shows the context in which the

knowledge claim was acquired or developed in the

organization.

The academic / legal references present artifacts

of external knowledge, such as journal articles and

proceedings, books and book chapters, law text and

somehow captured (in documents, emails, audio or

video files, for example) legal advice.

The mentions field provides a space for a

description of the knowledge claim internal

validation process: the mention of the knowledge

claim in conferences, formal or informal discussions

in which members of the assessment team or

executive board members took part.

The previous use field offers the possibility to

codify the identification of the knowledge claim in

benchmarking efforts, either internal or external to

the organization.

The criterion is the short sentence that is an

actual part of the assessment tool. It aims to assess a

particular aspect of the AI application. It is the actual

expression of the knowledge claim in the assessment

tool.

The application field offers a space for

alternative ways to express practical aspects of the

knowledge claim.

The fields proposed by and seconded by capture

a little of the social capital surrounding the

knowledge claim validation process, as they should

present the names of assessment team members that

sponsored the knowledge claim.

The decision field adds to the organizational

memory as it captures the result of the knowledge

claim validation process, whether it was retained,

rejected or if the group has not reached a decision

about it yet.

KEOD 2022 - 14th International Conference on Knowledge Engineering and Ontology Development

190

Table 1: Fields of the suggested documentation.

Proposition

sheet field

Values

KM cycle process or

kind of knowledge

orted

Principle The knowledge

claim

Knowledge

acquisition or

develo

ment

Motivation The inspiration or

reasoning behind

the knowledge

clai

Knowledge

acquisition,

development or

application

Academic /

legal

references

External

knowledge

artifacts supporting

the knowledge

clai

Knowledge

acquisition or

development

Mentions Mentions of the

knowledge claim

in the media,

conferences or

internal

discussions

Knowledge

acquisition or

development

Previous use Presence of the

knowledge claim

in a benchmarked

assessment tool

Knowledge

application

Criterion The knowledge

claim applied to a

solution

Knowledge

application

Application More context

about the

application of the

knowledge claim

Knowledge

application

Proposed by Employee

advocating for the

knowledge claim

Tacit knowledge

Seconded by Employee

supporting the

knowled

e claim

Tacit knowledge

Decision The result of the

knowledge claim

validation

rocess

Knowledge

development

6 CONCLUSIONS

Documentation helps to foster transparency. To build

trust in Artificial Intelligence solutions in general,

documentation is needed in many levels (Winfield &

Jirotka, 2018) and in many steps of the AI cycle

(Richards et al., 2020).

This communication explored the documentation

to be developed by the business owner regarding the

assessment of AI applications. Technological

assessment, in general, is influenced by

characteristics of the application domain, of the

organization and of the category of the solution in

question. Intellectual work is then necessary to make

sure these characteristics are included in the

assessment tool design. This intellectual work has to

also be collaborative, as expertise from different

backgrounds is necessary to evaluate technological

solutions not only from a technological viewpoint,

but also from the organization’s mission perspective,

in addition to a management perspective.

The negotiation process of what to assess and how

can be seen as a knowledge process. In this

knowledge process, a knowledge claim is advocated,

supported, defended, discussed, sponsored, rejected

or, in some cases, just left aside until a consensus

among team members can be reached.

Supporting the validation process of knowledge

claims during the design of assessment tools has the

benefit of providing a map of the knowledge

dependencies of the end product, in this case, the

technology assessment tool.

The diversity of knowledge involved in the design

of a technology assessment tool already justifies the

documentation support of the validation process of

knowledge claims. However, the design of a

technology assessment tool for an AI application also

involves the consideration of multifaceted impacts,

that have to be considered under the light of different

disciplines, which adds to the complexity of the

knowledge claim validation process. In addition, AI

technology and the understanding of its impacts

evolve quickly. In this scenario, being able to quickly

identify knowledge dependencies on outdated

knowledge helps to keep documentation updated, by

triggering a new knowledge claim validation process.

For these reasons, a template to codify aspects of

the knowledge claim validation process is suggested.

The document provides fields that capture elements

of the knowledge management steps of knowledge

acquisition, development and application. It also

fosters ways to capture the transformation of tacit

knowledge into explicit knowledge and of the social

capital involved in the knowledge claim validation

process.

REFERENCES

Amabile, T. (1998). How to kill creativity. Harvard

Business Review, 76(5), 76‑87.

Carrier, C., & Gélinas, S. (2011). Créativité et gestion : Les

idées au service de l’innovation. Presses de l’Université

du Québec. http://www.deslibris.ca/ID/438592

Dalkir, K. (2011). Knowledge management in theory and

practice (2nd éd.). MIT Press.

http://public.eblib.com/choice/publicfullrecord.aspx?p

=3339244

Knowledge Capture for the Design of a Technology Assessment Tool

191

Fisher, C., & Amabile, T. (2008). Creativity, Improvisation,

and Organizations. In The Routledge Companion to

Creativity (p. 27‑38).

Garbin, C., & Marques, O. (2022). Assessing Methods and

Tools to Improve Reporting, Increase Transparency,

and Reduce Failures in Machine Learning Applications

in Health Care. Radiology: Artificial Intelligence, 4(2),

e210127. https://doi.org/10.1148/ryai.210127

Hennessey, B. A., & Amabile, T. M. (2010). Creativity.

Annual Review of Psychology, 61(1), 569‑598.

https://doi.org/10.1146/annurev.psych.093008.100416

Janssen, M., Brous, P., Estevez, E., Barbosa, L. S., &

Janowski, T. (2020). Data governance: Organizing data

for trustworthy Artificial Intelligence. Government

Information Quarterly, 37(3). Scopus. https://doi.org/

10.1016/j.giq.2020.101493

Laato, S., Tiainen, M., Najmul Islam, A. K. M., &

Mäntymäki, M. (2021). How to explain AI systems to

end users: A systematic literature review and research

agenda. Internet Research, 32(7), 1‑31. Scopus.

https://doi.org/10.1108/INTR-08-2021-0600

McElroy, M. W. (2011). The new knowledge management:

Complexity, learning, and sustainable innovation.

KMCI Press.

Mitchell, M., Wu, S., Zaldivar, A., Barnes, P., Vasserman,

L., Hutchinson, B., Spitzer, E., Raji, I. D., & Gebru, T.

(2019). Model Cards for Model Reporting. Proceedings

of the Conference on Fairness, Accountability, and

Transparency, 220‑229. https://doi.org/10.1145/

3287560.3287596

Mokyr, J. (2000). Knowledge, Technology, and Economic

Growth during the Industrial Revolution. In B. van Ark,

S. K. Kuipers, & G. H. Kuper (Éds.), Productivity,

Technology and Economic Growth (p. 253‑292).

Springer US. https://doi.org/10.1007/978-1-4757-3161-

3_9

Nahapiet, J. (2009). Capitalizing on connections : Social

capital and strategic management. Social capital:

Reaching out, reaching in, 205‑236.

Nahapiet, J., & Ghoshal, S. (1998). Social Capital,

Intellectual Capital, and the Organizational Advantage.

Academy of Management Review, 23(2), 242 ‑ 266.

https://doi.org/10.5465/AMR.1998.533225

Nonaka, I., & Takeuchi, H. (1995). The knowledge-creating

company

. Oxford University Press.

Oliveira, D. (2022). Integrating tacit knowledge in product

lifecycle management: A holistic view of the innovation

process [PhD]. École de technologie supérieure.

Oliveira, D., Gardoni, M., & Dalkir, K. (2021). A Closer

Look at Concept Maps Collaborative Creation in

Product Lifecycle Management. In D. Tessier (Éd.),

Handbook of Research on Organizational Culture

Strategies for Effective Knowledge Management and

Performance. IGI Global. 10.4018/978-1-7998-7422-

5.ch014

Richards, J., Piorkowski, D., Hind, M., Houde, S., &

Mojsilović, A. (2020). A Methodology for Creating AI

FactSheets. http://arxiv.org/abs/2006.13796

Shoham, S., & Hasgall, A. (2005). Knowledge workers as

fractals in a complex adaptive organization. Knowledge

and Process Management, 12(3), 225‑236.

Winfield, A. F. T., & Jirotka, M. (2018). Ethical

governance is essential to building trust in robotics and

artificial intelligence systems. Philosophical

Transactions of the Royal Society A: Mathematical,

Physical and Engineering Sciences, 376(2133).

Scopus. https://doi.org/10.1098/rsta.2018.0085

KEOD 2022 - 14th International Conference on Knowledge Engineering and Ontology Development

192