Page 222 - DCAP208_Management Support Systems

Unit 13: Knowledge Management

13.2.1 Knowledge Elicitation Notes

According to whether one is interested in tacit knowledge in networks in the organisation or
knowledge buried in information systems, elicitation of the Knowledge Capital uses two distinct
approaches.

Tacit Knowledge Elicitation

Knowledge Transcription: A first type of approach is what one will call the knowledge
transcription: some tacit knowledge can be elicited simply, by transcribing them, in a
more or less structured manner. It is the case in the setting up of systems quality (of which
the first rule is “write what you have to do”), or in return on experience files, or in the
writing of publications. It is also the case of the “secondary documents” that synthesize
knowledge contained in given documents.

Knowledge Engineering: Knowledge Engineering is a manner more sophisticated than the
transcription to capture parcels of tacit knowledge. It appeared with expert systems (or
knowledge-based systems). These systems were supposed to replicate reasoning of experts
on domains of specific knowledge. One perceived quickly that if powerful technologies
were available to design such systems, the essential difficulty resided in the capacity to
transfer knowledge of one or several human experts into a computer program. Knowledge
Engineering put methods therefore in place to collect the knowledge, the most often from
interviews, and to structure it, in general from models.
These methods can be used therefore with profit to clarify, from interviews with knowledge
holders, a part of the capital of tacit knowledge of the organization.
An example typical of this evolution is the MASK method that was method to specify experts
systems and became a method of knowledge capitalization, integrated then in a general KM
problematic.

Knowledge Extraction

Knowledge Extraction from data: All enterprises detain big quantities of data, resulting
of their production activity. Those data are very diverse (technical, management,
marketing…), their mass doesn’t quit to grow (it doubles every 20 months on average).
Moreover there are other considerable masses of data called non-structured or
semi-structured, that are all textual data (and others media) that correspond to the
production of texts, cards, reports and other documents of any kind. This informational
capital is probably a wealth of the enterprise, for its production needs, but that could be
reused efficiently – a posteriori – for other needs. However, it proves to be that hardly 10%
of this capital is exploited. It can be explained by the difficulty to reuse information that
has been structured for objectives different from capitalization and reuse. However, big
efforts are made currently to valorise these layers of information accumulated by data
processing for production.

Did u know? The objective is about producing, from these layers, new information that are
useful to the action in the enterprise, in other words producing “operational knowledge”,
in the KM meaning.
It is a manipulation of information in an objective of knowledge discovery, called
knowledge extraction from data (or Knowledge Discovery from Data or KDD), linked
also to rather equivalent concepts as “Text Mining”, or “Data Mining” or “Data Warehouse”.

LOVELY PROFESSIONAL UNIVERSITY 215