Page 144 -

Knowledge Capture and Codifi cation 127

customization at the level of the different groups. Another benefi t of a good thesaurus
is that a keyword search engine can use each term to retrieve all relevant content (see
chapter 8).
A number of concept sorting techniques may be used in coding organizational
knowledge, ranging from manual to completely automated processes. An example
of a manual process would be to have participants sort cards into groupings. An
automated example would be something like the RepGrid technique developed by
Shaw (1981) based on Kelly ’ s (1955) personal construct theory. Most automated
systems use a form of cluster analysis to identify groupings in a set of data (e.g.,
hierarchical cluster analysis, Johnson 1967 ), multidimensional scaling (e.g., Kruskal
1977 ) or network scaling (e.g., Schvaneveldt, Durso, and Dearholt 1985 ). Cluster
analysis is a method of producing classifi cations from data that is initially unclassi-
fi ed. In hierarchical cluster analysis, the groupings are arranged in the form of a
hierarchical tree. Repertory grid analysis is a technique based on a theory that states
each person functions as a scientist who classifi es or organizes his or her world. Based
on these classifi cations, the individual is able to construct theories and act based on
these theories. A repertory grid depicts this theoretical framework for a given indi-
vidual. The different taxonomic approaches to the codifi cation of explicit knowledge
are summarized in table 4.2 .
In addition to the hierarchy, taxonomies can organize knowledge as lists, trees,
poly-hierarchies, matrices, facets, or system maps (Lambe 2007). Organizational
knowledge is often best represented using a multifaceted taxonomy or poly-hierarchy
that makes use of more than one classifi cation rule (or “ facet ” ). The general guideline
is that each facet must be clearly distinguishable from the others (e.g., shape, color,
and cost are three facets that do not overlap in any way). Another guideline is that
each facet should be clearly understood by all users (and if not, then a thesaurus
should keep track of equivalent terms). Good examples of a faceted taxonomy may
be found at http://wine.com, where wine is classifi ed according to region, taste, price,
and so on, and http://www.epicurious.com, where recipes can be classifi ed according
to type of event, type of cuisine, and time to prepare. A multifaceted taxonomy is
often used for business content, as it is the most fl exible and can deal with the often
messy, overlapping, ill-defi ned nature of knowledge used in a company. Facets are
relatively easy to add, remove, or modify in order to accommodate changes in the
organization, changes in user types, and changes in tasks. Finally, from a user perspec-
tive, each facet can serve as a search term to locate and retrieve content.
Most small and medium-sized organizations will primarily use manuals as a
means of developing taxonomy while larger organizations may be better positioned