You are on page 1of 2
Collective intelligence based algorithms for the coordination and harnessing of the corporate crowd’s i nnovation

Collective intelligence based algorithms for the coordination and harnessing of the corporate crowd’s innovation potential

Ioanna Lykourentzou 1 , Dimitrios J. Vergados 2 and Amedeo Napoli 1

1 INRIA, 615 rue du Jardin Botanique, 54600 Villers-lès-Nancy, France

2 School of Electrical and Computer Engineering, National Technical University of Athens, Zographou Campus, 15773 Athens, Greece

Corporate innovation, i.e. the firm’s ability to generate new knowledge, ideas and solutions faster than its competitors, is acknowledged as a major source of competitive advantage and consequently its development is

a central target of strategic management. A major key asset for the development of corporate innovation is the

effective harnessing of the organization’s intangible human resources, i.e. the expertise, knowledge and ideas

of its staff members.

To perform this harnessing organizations traditionally rely on hierarchical approaches, in which small groups of selected experts collaborate towards the development of new knowledge and solutions under the supervision of a human coordinator. This approach can guarantee the quality and timeliness of the final result, but it often limits the innovation ability of the organization due to a series of disadvantages. These include a narrow solution range, due to the limited number of collaborators, peer pressure and herding, as well as bottlenecks in the knowledge production, linked to the cognitive saturation of the human coordinator.

Viewing the impressive results often achieved by more participative forms of collaboration in the public internet, and in an effort to tap their in-house innovation capabilities more efficiently, corporations have recently started to experiment with crowd-involving approaches. These include prediction markets, crowdsourcing and collaborative content creation applications such as corporate wikis. Such approaches indeed demonstrate more innovation potential, with prediction markets offering impressive accuracy, crowdsourcing producing remarkably novel solutions and wikis covering broad knowledge fields. However, even in their most collaborative forms, these approaches do not seem to tap the full potential of the corporate crowd. For example prediction markets can typically forecast only among predefined sets of events, crowdsourcing does not allow for an organizational-wide skill combination, since it is mainly based on competition and wikis have often been criticized for questionable content quality, not guaranteed content production time and increased overhead costs as a result of the community self-coordination character.

A solution is therefore needed to combine the high quality result that the experts performing in small focused

groups can produce with the diversity of ideas and solutions that a collaborating crowd can come up with. To achieve the above, and in harmony with the latest suggestions of relevant research literature, in this paper we propose a novel, algorithm-based mechanism that actively coordinates the collaboration of the corporate crowd, reducing the self-coordination costs of the latter and systematically facilitating the emergence of its collective potential.

To develop this mechanism we applied a collective intelligence (CI), swarm-inspired approach. In general, swarm-based methods suggest that the coordinated collaboration of multiple entities, performed through a distribution of tasks amongst them and a subsequent aggregation of their localized actions, can produce significant collective results, while keeping the individual workload low. So far these approaches have

produced remarkable results when applied on natural communities, like bee swarms, or non-human artificial entities, like software agents.

In this work we shift this idea to human communities and propose an algorithm that can coordinate the collaboration of a corporate community, in a swarm-based manner in order to tap their collective potential regarding knowledge and innovation development more systematically and more efficiently compared to the self-coordination pattern.

The proposed algorithm uses machine learning, constraint-based techniques and resource allocation schemas to identify the resources (i.e. the tacit knowledge and skills) of the members of the collaborating corporate community and to allocate these resources to the exact knowledge development sub-tasks that they can improve the most, in this way seeking to ensure and enhance the reliability of the final knowledge result produced.

In this way, the corporate community is coordinated in a CI-based, swarm-resembling manner: each staff member is asked to perform only a few simple actions, like contributing their knowledge on a specific subject, or rating the knowledge contribution of another member without the need of being aware of all the activities taking place simultaneously. In parallel, the algorithms retain the broader picture, e.g. which knowledge piece needs to be enhanced to meet the organizational demands, as well as a centralized overview of the capabilities of each participant and allocate the available resources appropriately.

The development and performance evaluation of the algorithms is made combining findings from the social network analysis and the organizational simulation literature. That is, based on recent findings on the functionality of current knowledge creation communities (such as Wikipedia) we build the benchmark system upon which we apply and test the algorithms. Results indicate that the developed CI-based, swarm-resembling algorithms can provide more qualitative collective knowledge results and more efficient allocation of the capabilities of the corporate community.

Concluding, in this work we highlight the emerging and highly promising field of CI-based algorithms to coordinate, affect and enhance the results achieved by collaborating corporate communities and we provide specific results, obtained through organizational modeling and simulation, which indicate the improvement that the use of these algorithms can achieve.