Digital Social Matching Ecosystem for Knowledge Work
Jukka Huhtam
and Thomas Olsson
Tampere University of Technology, Korkeakoulunkatu 8, Tampere, Finland
University of Tampere, Kanslerinrinne 1, Tampere, Finland
Social Matching, Digital Ecosystem, Knowledge Management, Human Resources, Collaboration, Decision-
Support Systems.
Knowledge work involves various so-called social matching decisions: who to recruit, who to pair up or team
up, who to ask for consultancy, etc. Despite the scale of effects such decisions can have on organizations,
social matching activities are little supported by technology. In this position paper, we describe an ongoing
venture to develop the enablers and a shared vision for forming digital ecosystems around social matching of
knowledge workers. Rather than developing monolithic, organization-specific systems, we argue for an API-
based ecosystemic approach that helps co-create value and develop more networked, innovative, and viable
business ventures. We elaborate our vision and work-in-progress by presenting requirements for and scenarios
of digital ecosystems for social matching in knowledge work.
This position paper stems from a research venture be-
tween two university research groups and a consor-
tium of companies to develop the enablers and shared
vision for forming digital ecosystems around social
matching of knowledge workers.
What is social matching of knowledge workers?
The general notion of social matching refers to com-
putational ways of identifying and facilitating new so-
cial connections between people (Terveen and McDo-
nald, 2005). Our focus is on social matching in pro-
fessional life, particularly in creative and knowledge
work. Relevant activities include recruitment of new
personnel to knowledge-intensive organizations, team
formation within or across organizations for various
types and lengths of projects, seeking for mentors or
advisers as an individual or an organization basi-
cally, establishing any kind of collaboration relations-
hips related to knowledge work. Today, social mat-
ching activities are labour-intensive and based on hu-
man judgment. Yet, the decisions have significant
impact on the performance of organizations and the
wellbeing of individuals (Rogers and Blenko, 2006).
This makes professional matching decisions prone to
human error, and the risk of making an unsuccess-
ful choice is of high probability and high impact. We
consider this as a fruitful opportunity to envision new
forms of digital decision support systems.
What is an ecosystem, then? The key premise in
business ecosystems is that in order to be competitive,
companies must allow other companies to create ad-
ditional value to their own offering. James F. Moore
(1996) kicked off ecosystem discussion with his se-
minal article Predators and Prey: A New Ecology of
Competition in which he states that companies should
view themselves as part of a
“... business ecosystem that crosses a variety
of industries. In a business ecosystem, com-
panies coevolve capabilities around a new in-
novation: they work cooperatively and com-
petitively to support new products, satisfy cu-
stomer needs, and eventually incorporate the
next round of innovations.”
In some 20 years, academic research has mo-
ved further to define several complementing types
of ecosystems, including knowledge, innovation, and
business ecosystems (Valkokari, 2015). In know-
ledge ecosystems, companies and research organi-
zations come together to create new knowledge in
pre-competitive phase of research and development
arvi et al., 2018). Innovation ecosystems are inter-
connected, interdependent compositions of startups,
founders, investors, enterprises, universities, public
organizations that together drive the emergence of
new companies, products, and services (Russell et al.,
2011). Business ecosystems are composed of inter-
dependent companies co-creating value to their cus-
Huhtamäki, J. and Olsson, T.
Digital Social Matching Ecosystem for Knowledge Work.
DOI: 10.5220/0006950301940199
In Proceedings of the 10th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (IC3K 2018) - Volume 3: KMIS, pages 194-199
ISBN: 978-989-758-330-8
Copyright © 2018 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
tomers (Iansiti and Levien, 2004). Although the use
of ecosystem as concept in business and strategy lite-
rature remains a controversial topic (Oh et al., 2016),
scholars are actively investigating this form of colla-
borative value creation (Ritala and Almpanopoulou,
2017; Russell and Smorodinskaya, 2018).
What makes an ecosystem digital? Digital as-
sets are increasingly important in enabling and fa-
cilitating the emergence and success of ecosystems.
API ecosystems are a recent example of enabling co-
creation between companies through digital interfaces
that help exchange data and simple services (Evans
and Basole, 2016). Although Application Program-
ming Interface (API) is a core concept in modular
software development, in this context it refers speci-
fically to Web APIs, that is, application programming
that are available online for developers to use. Web
APIs have existed since the early days of the World
Wide Web and they had a core role for example in
the Web 2.0 vision (O’Reilly, 2007). However, only
recently technical and business developers have joi-
ned to explore this place of digital value creation they
both find themselves in (cf., Evans and Basole, 2016).
Digital ecosystem researcher Rahul Basole recaps:
“Businesses must both own and participate in
ecosystems. APIs make that happen. #digital
Our research venture has two key objectives. First,
we seek to develop data-driven, interactive service
concepts for professional social matching and rela-
ted methodology. Second, we aim to explore ways to
implement some of these service concepts at ecosy-
stem level, that is, in co-creation between companies
rather than within the corporate firewall. This pa-
per describes our work-in-progress analysis of what
type of ecosystems could be feasible from technology
perspective and desirable from the perspectives of or-
ganization studies and social psychology, particularly
in the relatively understudied area of social matching
in knowledge work. We argue for the opportunities
ecosystem thinking can bring in creating new digi-
tal services that facilitate professional collaboration.
Ecosystem building starts from shared vision (J
et al., 2018; Russell et al., 2011) and our shared vision
for ecosystem thinking starts with the API ecosystem
approach. That is, instead of designing a monolithic
platform-based ecosystem, our objective is to iden-
tify potential API-based collaborations on social ma-
tching, either among project consortium members or
between consortium members and third parties.
Why the social matching activities in knowledge work
require digital support? Choosing potential matches
are traditionally manual tasks performed by busy ma-
nagers or Human Resource professionals (e.g., mat-
ching an employer with a suitable employee). Ho-
wever, it is well known that decision-making is in-
herently limited by the human capacity of informa-
tion processing—based on intuition, heuristics, and
cognitive shortcuts (Kahneman and Tversky, 1973),
and striving for minimizing cognitive effort (Fiske
and Taylor, 1991). Human decision-making can re-
sult in tendencies like homophily, the preference of
interacting with like-minded others (McPherson et al.,
2001), and leaning on existing social networks and a
geographically limited pool of candidates. For exam-
ple, forming working groups in organizations often
display arbitrary and ill-justified choices even though
the combination of people can significantly influence
the productivity of the group. This calls for digi-
tal support that can help considering options beyond
the obvious (cf., Gal et al., 2017) and identify unex-
pected, yet meaningful social ties between actors.
Why should we care about ecosystems in the con-
text of professional social matching? People are in-
herently interconnected with various other individu-
als and organizations, also outside their professional
role. This means that the consideration of certain so-
cial ties should not limit to a particular matching acti-
vity; for example, a “good match” for a new headhun-
ted recruit could turn into a relevant mentor or mentee
in another context. Similarly, the needs for enhanced
collaboration do not limit to one’s primary professio-
nal role but relate also to secondary and tertiary roles,
not to mention collaboration in the third sector, hob-
bies, and others.
Importantly, if we take a textbook approach to im-
plement computational social matching, we end up
developing a people recommender system that seeks
to maximize there relevance of the recommendation.
Actor similarity, the number of shared existing con-
nections, and triadic formation of new connections are
the key predictors of a connection. Therefore, the in-
troduction of social recommender will only boost the
formation of the traditional connections. On the other
hand, knowledge work is fueled by complementary
information, viewpoints and skills; this calls for sys-
tems that help increase epistemic diversity and unex-
pected social ties in and between organizations.
Digital Social Matching Ecosystem for Knowledge Work
3.1 Social Matching Analytics
Our approach to social matching is computational and
data-driven. Specifically, we explore the use of Big
Social Data (BSD) in identifying potential connecti-
ons between actors. Olshannikova et al. (2017) define
BSD as
“any high-volume, high-velocity, high-variety
and/or highly semantic data that is generated
from technology-mediated social interactions
and actions in digital realm, and which can be
collected and analyzed to model social inte-
ractions and behavior.
Three types of BSD is available on individual ac-
tors. First, data on digital relationships that repre-
sent actors’ social network. Second, transactional
data on the interactions between actors. Third, actor-
produced content, including their self-representations
and discussions.
Social matching starts from the creation of net-
work representation of the existing social connections
between actors. Various data processing methods are
used to extract features that represent the knowledge,
competences, and interests of individual actors in al-
gorithms. Once the network is composed and actors
have their representation in algorithms, several alter-
native analytical option are available. Many of these
approaches are based on pairwise analytics of the ac-
tors, including measuring their social distance and the
similarity of their interests or produced content (Tsai
and Brusilovsky, 2018).
Importantly, relevance-first approach is not advi-
sed in social matching analytics. Instead, social mat-
ching should seek ways to nudge actors to diversity-
seeking behavior. Thaler and Sunstein (2008) define
nudging as means to “alter people’s behavior in a pre-
dictable way without forbidding any options or sig-
nificantly changing their economic incentives.” Me-
ans to implement nudging effects in social matching
systems include the transparency, controllability, and
explainability (Tintarev and Masthoff, 2015; Tsai and
Brusilovsky, 2018). At the same time, we want to note
that in general, social matching is an act of optimi-
zing for the diversity-bandwidth trade-off (Aral and
Van Alstyne, 2011). That is, both strong ties with high
bandwidth and weak ties as sources of potential novel
information are valuable to users.
3.2 API-based vs. Platform-based
Two basic architectures exist for digital ecosys-
tems, that is, platform-based and API-based. In the
platform-based approach, a keystone company opera-
tes a platform and provides the other companies me-
ans to develop complementary products. Apps built
on mobile platforms is a prime example of a comple-
mentary product. On the other hand, API ecosystems
are composed of dyadic service combinations. In API
ecosystems, companies co-create value by providing
Web APIs to each others consumption.
APIs are an important example of boundary re-
sources. Formally, boundary resources are “the soft-
ware tools and regulations that serve as the interface
for the arm’s length relationship between the plat-
form owner and the application developer” (Ghazaw-
neh and Henfridsson, 2013). Compared to platform-
based ecosystems, API ecosystems are nimble and
flexible. New combinations are easy to form and re-
move. API ecosystems are loosely connected com-
pared to platform-based ecosystems. Twitter, for ex-
ample, provides an API for developers to access data,
including tweets and user profiles. Moreover, IBM
Watson can be used through APIs.
Two basic types of APIs exist, data APIs and
functional APIs. Data APIs have dominated in the
early years of API development. However, functional
APIs present an approach that does not insist a com-
pany to release their data to other actors but instead
create value by implementing their own data products.
In our venture, the quest to develop ecosyste-
mic social matching functionalities and services starts
from an API-based approach. This is because we do
not have a clear candidate to serve in the role of keys-
tone that operates a platform. From technical view-
point, social matching comes with a versatile set of
use cases on collecting, cleaning, refining, modeling,
and analyzing data. Due to this versatility, it is in
practice impossible to implement a one-size fits all
technical solution.
In order to manage the technical diversity, we take
a component-based approach (Nyk
anen et al., 2007)
to implement data-processing pipelines for social ma-
tching. This approach is an intuitive extension of
our previous work on data-driven visual analytics that
manifests as the Ostinato Process Model (Huhtam
et al., 2015).
From a technical viewpoint, it is relatively straig-
htforward to assign individual components to be run
as services that different companies provision to each
other. That is, reaching technical modularity is achie-
vable simply by following API design principles. Ho-
KMIS 2018 - 10th International Conference on Knowledge Management and Information Sharing
wever, potential issues related to organizational mo-
dularity truly hinder the design of ecosystems. Once
companies enter an ecosystemic collaboration, by de-
finition they become interdependent to each other
both at technical and organizational level (Ghazaw-
neh and Henfridsson, 2011; Yoo et al., 2010). Data is
a core asset to companies and therefore they are un-
derstandably reluctant to share the data to others.
3.3 Legal Constraints
The newly launched General Data Protection Regu-
lation (GDPR) is truly a gamebreaker in Professio-
nal Social Matching. From BSD mining viewpoint,
GDPR adds major restrictions. Data on individual ac-
tors can only be collected for a dedicated purpose and
with informed consent. Moreover, acts of data-driven
profiling must be reported to the actors that the data
and resulting profiles describe.
It seems that platform-based services such as Lin-
kedIn and Duunitori have an advantage in the GDPR
age. These platforms are able to collect informed con-
sent from users to a) profile them and b) make auto-
mated decisions (e.g., position recommendations for
candidates and candidate recommendations for com-
panies). Using harvested data does not seem to be an
option at all.
To make the intersection of social matching, know-
ledge work, and ecosystems more tangible, the fol-
lowing presents two scenarios of desirable futures.
These shed light on our empirical work in progress
and highlight the potential ecosystemic benefits and
different types of API provider and customer roles in
different realms of knowledge work.
4.1 Global Innovation Platform
A global innovation platform operates by taking pro-
blem statements (case projects) from companies and
combining university student teams to work on the
tasks for an intensive period of problem based lear-
ning. The innovation platform has staff to facilitate
the projects, and each project team is able to exchange
views with university-based topic expert. The pro-
jects produce different benefits for different stakehol-
ders: important experience and connections for stu-
dents, new learning platforms for universities, fresh
ideas for the companies, and atmosphere of open in-
novation to the local community. This is an oppor-
tune ground for ecosystemic social matching scena-
rios. But how to identify which students form an
effective team together? How to identify appropri-
ate expert advisers for each project? How to identify
which possible project topics are most suitable for this
kind of innovation projects?
Data-driven team formation refers to a data-driven
approach to compose and structure teams for new
projects (cf., Zhou et al., 2018). Services for mana-
ging the workforce of an organization (here, a pool
of available students) would provide a workable star-
ting point for this scenario. With the growing trend to
create digital portfolios and professional profiles on-
line, these data about students could be analyzed for
team formation, along with the data the university has
about them. While these data can be privacy sensi-
tive and indeed have multiple origins, the innovation
platform would need an access only to an API that
provides a list of key qualities that each student has
or would like to learn. Ideally, the qualities would
include not only skill and knowledge areas but also
information about their general cognitive styles and
suitable roles in team-based innovation work.
Second, universities are increasingly using Cur-
rent Research Information Systems (CRIS) to help
maintain researcher portfolios and gather data about
academic output, such as publications, talks, awards
and intellectual property. This portfolio data inten-
ded to be public anyway could be sourced to identify
key interests and competences for each researcher and
teacher in order to create features representing them
in a matchmaking algorithm. These features would be
matched with the project descriptions and the student
groups. This insists that the university provides an
API to the CRIS data. Interestingly, such API provi-
sion is well in line with open science and open access
ideals but in stark contrast with GDPR.
Third, the innovation platform can accumulate
data about past successful innovation projects and
their characteristics. Using for example machine lear-
ning, we might identify what types of organizations,
which project topics or what kind of customer invol-
vement have yielded best results for the customer or-
ganization, for the student group, or for the society
as a whole. This insight can be used to headhunt for
suitable project from the local businesses and to se-
lect the most suitable teams for each case. As such
data does not yet exist, it is crucial to define suitable
measurements and practices of gathering relevant data
systematically about each project.
Digital Social Matching Ecosystem for Knowledge Work
4.2 Matching as a Service for
Matching as a Service (MaaS) is envisioned as a new
initiative to provide an organization and its employees
with digital support for various internal social mat-
ching cases. Particularly leadership and managerial
activities, such as individuals’ career development as-
sistance, activity partnering for task forces and leisu-
rely activities, finding a mentor, and ad hoc team for-
mation for various short-term projects would be fruit-
ful cases that are currently missing proper digital sup-
port. At the same time, in knowledge-intensive orga-
nizations and professions, people possess latent, un-
tapped potential and tacit knowledge: many epistemic
problems could be solved with the help of peers rather
than by utilizing the conventional chain of command
through the hierarchy. But how to efficiently iden-
tify what latent skills and knowledge different people
have and who might need them?
The vision comes with an online platform that the
employees use to state needs and requirements for lea-
dership services, of which many fall under social ma-
tching. In order for this scenario to become ecosys-
temic, platform developer must provide boundary re-
sources to allow third-party development of comple-
mentary services to the platform. The ecosystemic
perspective could mean, for example, offering the
identified skills or knowledge outside the organization
to customers and partners. Partnering organizations
could offer API-enabled leadership services to each
other, especially in smaller companies without an es-
tablished HR department. Alternatively, ecosystemic
thinking could be advocated in smaller scale within
the organization: giving more room for grass-root ini-
tiatives and for example internal startups. Particularly
in large enterprises the organizational rigidity and si-
los call for enhanced interplay between actors in dif-
ferent parts of the organization.
In this position paper, we described our vision of the
building blocks of digital social matching ecosystems
for knowledge work. We argued for the need for
social matching within and in-between organizations
and point to Web APIs, a key category of boundary
resources, as digital means to implement ecosystemic
co-creation relationships between organizations. We
hope the vision encourages further transdisciplinary
investigation of the practicality and real-life desira-
bility of digital social matching in general and the
ecosystemic approach in particular.
The current shift in legislation toward increased
privacy and right to be forgotten and therefore limited
data access (GDPR), as well as increased user cont-
rol and machine readability (MyData) must be consi-
dered when planning ecosystemic digital social mat-
ching concepts. Need for informed consent from each
individual separately seems to effectively prevent mi-
ning big social data en masse. That is, the existence
of a social supercollider (Watts, 2013) for social ma-
tching seems unlike at this stage. This implies that
global platforms for digital work posses a major ad-
vantage in developing analytical capability for social
Future research should look into what are suitable
and ethically sustainable design goals for social ma-
tching particularly in knowledge work; what kind of
business models best serve API-based ecosystems in
this domain; and how to enable gathering and analy-
zing relevant data in the current legislative landscape.
Ecosystemic service concepts for social matching that
cross the boundaries of individual organizations are
yet to be developed. We will continue our venture
to develop some of these concepts. We call for the
exchange of viewpoints on privacy-opportunity trade-
off for ecosystemic digital social matching.
This work was conducted under Business Finland
project Big Match (3166/31/2017 and 3074/31/2017).
Aral, S. and Van Alstyne, M. (2011). The Diversity-
Bandwidth Trade-off. American Journal of Sociology,
Evans, P. C. and Basole, R. C. (2016). Revealing the API
Ecosystem and Enterprise Strategy via Visual Analy-
tics. Communications of the ACM, 59(2):26–28.
Fiske, S. T. and Taylor, S. E. (1991). Social cognition.
McGraw-Hill, New York, 2nd edition.
Gal, U., Jensen, T. B., and Stein, M.-K. (2017). People
Analytics in the Age of Big Data: An Agenda for IS
Research. In Thirty Eighth International Conference
on Information Systems, South Korea 2017 1, page 11.
Ghazawneh, A. and Henfridsson, O. (2011). Micro-
Strategizing in Platform Ecosystems: A Multiple Case
Study. In ICIS 2011 Proceedings.
Ghazawneh, A. and Henfridsson, O. (2013). Balancing plat-
form control and external contribution in third-party
development: the boundary resources model. Infor-
mation Systems Journal, 23(2):173–192.
KMIS 2018 - 10th International Conference on Knowledge Management and Information Sharing
aki, J., Russell, M. G., Rubens, N., and Still, K.
(2015). Ostinato: The Exploration-Automation Cy-
cle of User-Centric, Process-Automated Data-Driven
Visual Network Analytics. In Matei, S. A., Russell,
M. G., and Bertino, E., editors, Transparency in Social
Media: Tools, Methods and Algorithms for Media-
ting Online Interactions, Computational Social Scien-
ces, pages 197–222. Springer International Publishing
Iansiti, M. and Levien, R. (2004). Strategy as Ecology. Har-
vard Business Review, 82(3):68–81.
arvi, K., Almpanopoulou, A., and Ritala, P. (2018). Orga-
nization of knowledge ecosystems: Prefigurative and
partial forms. Research Policy, 47(8):1523–1537.
Kahneman, D. and Tversky, A. (1973). On the psychology
of prediction. Psychological Review, 80(4):237–251.
McPherson, M., Smith-Lovin, L., and Cook, J. M. (2001).
Birds of a Feather: Homophily in Social Networks.
Annual Review of Sociology, 27(1):415–444.
Moore, J. F. (1996). The death of competition: leadership
and strategy in the age of business ecosystems. Har-
perBusiness, New York, NY.
anen, O., Mannio, M., Huhtam
aki, J., and Salonen, J.
(2007). A socio-technical framework for visualising
an open knowledge space. In Proceedings of the Inter-
national IADIS WWW/Internet 2007 Conference, pa-
ges 137–144, Vila Real, Portugal. Proceedings of the
International IADIS WWW/Internet Conference.
Oh, D.-S., Phillips, F., Park, S., and Lee, E. (2016). Inno-
vation ecosystems: A critical examination. Technova-
tion, 54:1–6.
Olshannikova, E., Olsson, T., Huhtam
aki, J., and
ainen, H. (2017). Conceptualizing Big Social
Data. Journal of Big Data, 4(1):3.
O’Reilly, T. (2007). What is Web 2.0: Design patterns and
business models for the next generation of software.
Communications & strategies, (65):17–37.
Ritala, P. and Almpanopoulou, A. (2017). In defense of
‘eco’ in innovation ecosystem. Technovation, 60-
Rogers, P. and Blenko, M. W. (2006). Who Has the D?:
How Clear Decision Roles Enhance Organizational
Performance. Harvard Business Review.
Russell, M. G. and Smorodinskaya, N. V. (2018). Levera-
ging complexity for ecosystemic innovation. Techno-
logical Forecasting and Social Change.
Russell, M. G., Still, K., Huhtam
aki, J., Yu, C., and Ru-
bens, N. (2011). Transforming Innovation Ecosys-
tems through Shared Vision and Network Orchestra-
tion. In Proceedings of Triple Helix IX Internatio-
nal Conference: “Silicon Valley: Global Model or
Unique Anomaly?”, July 2011, Stanford, California,
USA, page 17, Stanford, California, USA.
Terveen, L. and McDonald, D. W. (2005). Social Matching:
A Framework and Research Agenda. ACM Transacti-
ons on Computer-Human Interaction, 12(3):401–434.
Thaler, R. H. and Sunstein, C. R. (2008). Nudge: improving
decisions about health, wealth, and happiness. Yale
University Press.
Tintarev, N. and Masthoff, J. (2015). Explaining Recom-
mendations: Design and Evaluation. In Recommen-
der Systems Handbook, pages 353–382. Springer US,
Boston, MA.
Tsai, C.-H. and Brusilovsky, P. (2018). Beyond the Ran-
ked List: User-Driven Exploration and Diversifica-
tion of Social Recommendation. In Proceedings of
the 2018 Conference on Human Information Inte-
raction&Retrieval - IUI ’18, pages 239–250.
Valkokari, K. (2015). Business, Innovation, and Knowledge
Ecosystems: How They Differ and How to Survive
and Thrive within Them. Technology Innovation Ma-
nagement Review, 5(8):17–24.
Watts, D. J. (2013). Computational Social Science: Ex-
citing Progress and Future Directions. The Bridge,
Yoo, Y., Henfridsson, O., and Lyytinen, K. (2010). The New
Organizing Logic of Digital Innovation: An Agenda
for Information Systems Research. Information Sys-
tems Research, 21(4):724–735.
Zhou, S., Valentine, M., and Bernstein, M. S. (2018). In
Search of the Dream Team: Temporally Constrained
Multi-Armed Bandits for Identifying Effective Team
Structures. In Proceedings of the 2018 CHI Confe-
rence on Human Factors in Computing Systems - CHI
’18, pages 1–13.
Digital Social Matching Ecosystem for Knowledge Work