Posts Tagged ‘knowledge conversion’

Mind the Gap: Transcending the Tunnel Vision on Ontology Engineering.

November 17, 2007

De Leenheer, P. and Christiaens, S. (2007) Mind the Gap: Transcending the Tunnel Vision on Ontology Engineering. In Proc. of the 2nd Int’l Conf. on the Pragmatic Web 2007 (Tilburg, NL). ACM Press

The key objective of communal knowledge sharing at the scale of the World Wide Web is the ability to collaborate and integrate within and between communities. Ontologies, being formal, computer-based specifications of shared conceptualisations of the worlds under discussion, are instrumental in this process by providing shared semantic resources. To this end, the pragmatic aspects of the exchange of knowledge and information are crucial. Pragmatics represent the intentions, motivations and methodologies of the persons involved and need to become formalised and unambiguous for effective exchange to occur. On the one hand, this is something that humans manage fluently in their daily face-to-face social discourses. On the other hand, as contemporary knowledge engineering methods consider only the non-human system parts, they usually focus on mere syntactic aspects of concept modelling. The elicitation (semantics) and application (pragmatics) context are often weak or even ignored. This paper aims to bridge this gap between “reality” and its modelling concepts by (i) transcending knowledge engineering methods to a semiotics view on contextualised communal knowledge engineering and sharing; and (ii) by presenting the DOGMA ontology framework and how it provides extension points to this semiotics engineering.

[…] In order to illustrate the gap between knowledge sharing among people and information exchange at the technical system level we mentioned between the social/human knowledge-sharing system and its technical “mirror”, we adopt Nonaka’s well-known four modes of knowledge conversion [18] (see figure).

Through socialisation, people naturally utter and share experience and expertise in face-to-face discourses, and thereby create tacit knowledge such as mental models of ontological concepts and new technical skills. The current application context is about the concept Deliver.

Through externalisation, this tacit knowledge is partly articulated (publicly or privately) into explicit formal knowledge artefacts, taking the shape of e.g., a concept type, a contributed taxonomy, an interface, a workflow definition, etc. This is illustrated by the curved arrows that take a selection from the mental models that is relevant to explicate the concept Deliver in this application context. Note that although externalisation is an incremental process using language of variable expressivity (as we will explain below), there will inevitably remain an important part of tacit knowledge in the utterer’s mental model on which the correct interpretation of the articulated part is dependent [20,21]. Externalisation is done by domain experts, as they have the tacit knowledge about the domain and can sufficiently assess the real impact of the conceptualisations and derived collaborative services on their organization.

Once they are published, combination involves semantic analysis and integration (for an excellent survey, see [15]) of published contextualised knowledge artefacts in order to adapt to new collaborative requirements. This process might be further constrained by community-shared models such as running application tools that commit to certain published consensus, by pre-existing organisational sub-ontologies, and by inflexible data schemas interfacing to legacy data. Furthermore, participating stakeholders usually have strong individual interests, inherent business rules, and entrenched work practices that influence decisions in meaning negotiation rounds. These may be tacit, or externalised in workflows that are strongly interdependent, hence further complicate the conceptual combination. Sometimes it is not necessary (or even possible) to reach for context-independent ontological knowledge, as most ontologies used in practice assume a certain context and perspective of some community [23]. Wenger [26] supported this by stating “Peace, happiness, and harmony are therefore not necessary properties of a community of practice”. Hence pragmatically, combination processes need to support human experts to focus on these “community-grounded” processes of realising the appropriate amount of consensus on relevant conceptual definitions through effective meaning negotiation in an efficient manner [7].
Internalisation concerns the appropriate operationalisation and embodiment of explicit knowledge consensus in the current communication actions. For example, for ontological knowledge, the most widely used recommendations on the Semantic Web are XML, RDF(S) and OWL.

The four modes of knowledge conversion engender an upward knowledge spiral, where individual knowledge opinions become commonly accepted, through an iterative interplay between externalisation and internalisation. This interplay illustrates how knowledge artefacts co-evolve with their communities of use. In socialisation, knowledge is communicated as abstract entities, and humans can easily grasp the context of interpretation. However, when applying formal knowledge engineering methods for the other knowledge conversion modes, we have to emphasize the semiotic dimensions, in order to bridge the gap between socialisation processes in the social system on the one hand, and the other conversion processes (externalisation, internalisation, and combination) in the technical system on the other hand.